Copyright is owned by the Author of the thesis. Permission is given for a copy to be downloaded by an individual for the purpose of research and private study only. The thesis may not be reproduced elsewhere without the permission of the Author. Understanding Aspects of Andesitic Dome-forming Eruptions Through the Last 1000 yrs of Volcanism at Mt. Taranaki, New Zealand A dissertation presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Earth Science Thomas Platz Massey University, Palmerston North New Zealand 2007 The summit of Mt. Taranaki. Photographed by G. Lube. Dedicated to my parents, to Katrin and my son August i Abstract Andesitic volcanoes are notorious for their rapid and unpredictable changes in eruptive style between and during volcanic events, a feature normally attributed to shallow- crustal and intra-edifice magm atic processes. Using the ex ample of eruptions during the last 1000 yrs at Mt. Taranaki (the Maero Eruptive Period), deposit sequences were studied to (1) understand lava dome formation and destruction, (2) interpret the causes of rapid shifts from extrusive to explosive eruption styles, and (3) to build a model of crustal magmatic processes that impact on eruption style. A new detailed reconstruction of this period identifies at least 10 eruptive episodes characterised by extrusive, lava dome- a nd lava flow-producing events and one sub- Plinian eruption. To achieve this, a new ev aluation procedure was developed to purge glass datasets of contaminated mineral-gl ass analyses by using compositional diagrams of mineral incompatible-compatible elem ents. Along with careful examination of particle textures, this procedure can be br oadly applied to build a higher degree of resolution in any tephrostratigraphic record. Geochemical contrasts show that the products of the latest Mt. Taranaki erupti on, the remnant summit dome (Pyramid Dome) was not formed during the Tahurangi eruptive episode but extruded post- AD 1755. Its inferred original maximum volume of 4.9×10 6 m3 (DRE) was formed by simultaneous endogenous and exogenous dome growth within days. Magma ascent and extrusion rates are estimated at ≥ 0.012 ms -1 and ≥ 6 m 3 s-1 , respectively, based on hornblende textures. Some of the Maero-Period dome ef fusions were preceded by a vent-clearing phase producing layers of scattered lithic lapilli around the edifice [Newall Ash (a), Mangahume Lapilli, Pyramid Lapilli]. The type of dome failure controlled successive eruptive phases in most instances. The destruction of a pressurised dome either caused instantaneous but short-lived magmatic fragmentation (Newall and Puniho episodes), or triggered a directed blast-explosion (Newall episode), or initiated sustained magmatic fragmentation (Burrell Episode). The transi tion from dome effusion to a sustained, sub- Plinian eruption during the Burrell Lapilli ( AD 1655) episode was ca used by unroofing a conduit of stalled magma, verti cally segregated into three layers with different degrees of vesiculation and crystallisation. The resultant ejecta range from brown, grey and black coloured vesicular clasts to dense gr ey lithics. Bulk compositional variation of erupted clasts can be modelled by fractionation of hornblende, plagioclase, clinopyroxene, and Fe-Ti oxides. Pre-er uption magma ascent for the Maero Period events is assumed to begin at depths of c.9.5 km. iii Acknowledgements I wish to thank every person who contri buted to the outcome of this study. I went back to New Zealand to study one of the most fascinating phenomena nature has to offer: volcanoes. Of the North Island volcan oes, Mount Taranaki stands out as being the most imposing, and to me the most stri king. Despite its dorma ncy, it proved a real challenge, to climb, find samples and sections, and to draw out some of his secrets. For the opportunity to work ag ain on Mt. Taranaki I am indebted to my chief-supervisor Dr Shane J. Cronin. Through him I learnt much about how to observe volcanic deposits and to understand the various processes involved in generating them. His aid in my receiving a Massey University Doctoral schola rship is highly appreciated. In past years Dr Cronin supported numerous overseas trips, which offered me the opportunity to meet other scientists, either on conferences or at the institutions where my overseas supervisors are based. I also benefited from his extraordinary skills in writing and presenting ideas and thoughts. I also wish to thank all my co-supervisors: Prof. Vince E. Neall and Dr Robert B. Stewart (Massey University, New Zealand), Pr of. Katharine V. Cash man (University of Oregon, USA), and Prof. Stephen F. Foley (U niversity of Mainz, Germany) for their guidance and support over the years, their fruitful discussions and criticisms. I gratefully acknowledge the skill of all my supervisors to catch my thoughts especially when articulated in a Germanic English. During my studies a lot of people were invol ved that helped and supported me from different parts of the globe. New Zealand: First of all I wish to thank Jonathan N. Procter, Michael B. Turner, Anke V. Zernack and Kat A. Holt, my fellow inha bitants in the “Magma Chamber”. Without their friendship, life at Massey would have been colourless. Sharing “ups and downs” made many situations easier to bear. I am also grateful to the staff of the So il and Earth Sciences Group for assisting with everyday needs: Bob Toes (for also helpin g to fix Emma), Ian F. Furkert, Ross W. Wallace, Mike R. Bretherton, Moira Hubbard, Gl enys C. Wallace, Anne R. West, Dr R. Clel Wallace, Dr Jerome A. Lecointre, Julie A. Palmer and Dr Alan S. Palmer. Carolyn B. Hedley (Landcare Research) introduced me to the thermal analysis technique. Dr Ritchie Sims and John Wilmshur st assisted in XRF- and el ectron microprobe analysis at The University of Auckland. For discussions on aspects of Taranaki and access to their iv geochemical data I sincerely thank Assoc. Prof. Ian E.M. Smith (The University of Auckland) and Prof. Richard C. Price (Waikato University). Garry Bastin (Taranaki Research Centre) is thanke d for his assistance in looki ng for historic references. Eugene: From the Department of Geological Scie nces I wish to thank Dr Heather M.N. Wright for introducing and assisting in samp le preparation and measurement using the He-pycnometer and porosimeter. For insp iring discussions on pumice I especially acknowledge Drs Alison C. Rust and Heathe r M.N. Wright. Emily Johnson, Dr Julie Roberge, and Celeste Mercer are also th anked for their help in preparing melt inclusions. For discussions and access to th e FTIR- and SEM facilities, the help of Assoc. Prof. Paul J. Wallace and Dr John J. Donovan are appreciated. Kiel: Encouraging and fruitful discussions w ith Dr Armin Freundt and Sebastian Münn (IfM-GEOMAR) are especially acknowledged. Mainz: For assistance in sample preparation an d measurement I wish to thank the team on the Laser-ICP-MS, Institute of Geosciences, especially Drs Franziska Nehring and Matthias Barth. Copenhagen: For encouragement, discussion, and anal ysis Dr Holger B. Lindgren of the Geological Survey of Denmark a nd Greenland is highly appreciated. Potsdam: My thanks go also to Dr Thom as R. Walter (GeoForschungsZentrum Potsdam) for his unreserved discussions. Studying abroad would have been far more di fficult without the support of friends and family back home. My sincere thanks go to my parents who always supported me in everything I pursued. My friends Dr Gert Lube, Sebastian Münn, Dr Franziska Nehring, Luzie Herklotz, Dr Diana Reckien, and Jonathan N. Procter were always there when I needed them. But the most important part was contributed by Katrin Bauer, thr ough her love, respect, and honesty. This work was funded and supported by a Mass ey University Doctoral Scholarship, the George Mason Trust of Taranaki, the He len E. Akers Scholarship, the FRST-PGST contract MAUX0401, and an Institute of Na tural Resources transitional scholarship. v Table of Contents Abstract i Acknowledgements iii List of Tables xi List of Figures xiii Chapter 1 Introduction 1 1.1 Introduction 1 1.2 Objectives and Strategy 5 1.3 Thesis Outline 6 1.4 Background Geology 7 1.4.1 Regional Geological Setti ng 7 1.4.2 Taranaki Basin 9 1.4.3 Taranaki Volcanic Lineament 12 1.4.4 Mount Taranaki/Egmont 12 1.4.5 Petrology of Mt. Taranaki Rocks 15 1.5 References 19 Chapter 2 Methodology 31 2.1 Brief Outline 31 2.2 Field Studies 32 2.3 Mineralogy 33 2.3.1 Sample Preparation 33 2.3.2 Microscopic Studies 34 vi 2.4 Geochemistry 34 2.4.1 X-ray Fluorescence Spectroscopy 34 2.4.2 Electron Microprobe Anal ysis 35 2.4.3 Laser Inductively Coupled Plasma Mass Spectrometry 36 2.5 Porosity and Permeability 37 2.6 Scanning Electron Microscopy 41 2.7 Fourier Transform Infrared Spectro scopy 41 2.8 Thermal Analysis 43 2.9 References 46 Chapter 3 The Maero Eruptive Period 49 3.1 Introduction 49 3.1.1 Previous Studies 51 3.1.1.1 Previous Work on Lava Flow Stratigraphy 54 3.2 Results 57 3.2.1 Stratigraphic Type and Reference Sec tions 57 3.2.1.1 Type Section of the Maero Formation 58 3.2.1.2 Reference Sections of the Maero Form ation 64 3.2.2 Block-and-Ash Flow Deposits 68 3.2.2.1 Distribution and Flow Pa ths 68 3.2.3 Lava Flows 71 3.2.4 Glass Chemistry 75 3.2.4.1 Special Characteristics of Some Erup ted Units 76 3.2.4.2 Correlation of Tephra and Pyroclastic Flow Deposits 77 3.3 Discussion 81 3.3.1 Eruption Frequency of the Maero Erup tive Period 81 3.3.2 Comparison to the Previously Known Stratigraphy 86 3.3.3 Tephrostratigraphy of the Maero Eruptiv e Period 87 3.4 References 91 vii Chapter 4 Improving the Re liability of Microprobe-based Glass Analyses 95 4.1 Abstract 97 4.2 Introduction 97 4.2.1 Andesitic Volcanism, Tephra Generati on and Dispersal 100 4.2.2 Sample Sites 103 4.3 Results 103 4.3.1 Contrasts in Particle Texture 103 4.3.2 Glass Chemistry and Data Evaluation 105 4.3.3 Estimating Plagioclase Proportions in Contaminated Analyses 111 4.4 Discussion 112 4.5 Conclusions 116 4.6 References 117 Chapter 5 Non-explosive, Dome-forming Eruptions 121 5.1 Introduction 121 5.1.1 Lava Domes 122 5.1.1.1 Lava Dome Distribution in Taranaki 124 5.2 Field Observations 124 5.2.1 Tahurangi Eruptive Deposits 124 5.2.2 The Present Summit Lava Dome 128 5.2.3 Rock-Avalanche Deposit 130 5.3 Sample Sites and Methods 131 5.4 Results 132 5.4.1 Dome Volume Calculati ons 132 5.4.2 Mineralogy and Mineral Chemistry 134 5.4.3 Bulk Rock Composition 139 5.4.4 Microstructure, Density and Permeab ilitiy of Dome Rocks 142 5.5 Discussion 142 viii 5.5.1 Lava Dome Emplacement and Growth 142 5.5.2 Lava Dome Collapse 145 5.5.3 Estimation of Eruption Parameters 148 5.5.3.1 Magma Source Areas 148 5.5.3.2 Magma Storage at the Level of Neut ral Buoyancy 151 5.5.3.3 Inferences for Hornblende Stability 153 5.5.3.4 Magma Ascent Rate 154 5.5.4 Eruption Duration 155 5.5.5 Approximation of the Time of Erupti on 158 5.5.6 Historic References to Volcanic Activity in the 19 th Century 161 5.6 Conclusions 166 5.7 References 167 Chapter 6 Transition to Explosive Eruptions 175 6.1 Abstract 177 6.2 Introduction 178 6.2.1 Geological Setting 179 6.3 Methodology 180 6.4 Results 182 6.4.1 Field Observations 182 6.4.1.1 Fall Deposits 182 6.4.1.2 Pumice Pyroclastic Flow Deposits 182 6.4.1.3 Block-and-Ash Flow Deposits 185 6.4.2 Volume Estimates 185 6.4.3 Clast Types and Textures 186 6.4.4 Mineralogy and Mineral Chemistry 189 6.4.5 Bulk Rock Geochemistry 190 6.4.6 Glass Composition 194 6.4.7 Porosity and Permeability 196 ix 6.5 Discussion 198 6.5.1 Pre-climactic Conditions 199 6.5.1.1 Grey Lithics 199 6.5.1.2 Grey Pumice (Unit 1) 201 6.5.1.3 Brown Pumice (Unit 2) 202 6.5.2 Syn-climactic Conditions 203 6.5.2.1 Black, Banded, and Unit 3 Grey Pumice Clasts 203 6.5.3 Eruption Dynamics 204 6.6 Conclusions 206 6.7 References 208 6.8 Appendix 213 6.8.1 Appendix A: Correlation of Burrell Deposits, NW Sector 213 6.8.2 Appendix B: Location of Pyroclastic Flow Deposits 214 6.8.3 Appendix C: Volatile Contents in Glass – Preliminary Results and Conclusions 215 6.8.3.1 FTIR Spectroscopy 215 6.8.3.2 Thermal Analysis 217 Chapter 7 Reconstruction of Eruption Mechanisms Using Physico-chemical Data 221 7.1 Introduction 221 7.1.1 Viscosity of Magmas 222 7.2 Methods and Approach 225 7.3 Results 225 7.3.1 Petrography 225 7.3.1.1 Pyroclastic Flow Deposits 225 7.3.1.2 Lava Flows 230 7.3.2 Bulk Rock Chemistry 230 7.3.2.1 Pyroclastic Flow Deposits 230 7.3.2.2 Lava Flows 232 x 7.3.3 Physical Properties 232 7.3.3.1 Water Estimates of Volcanic Glasses 233 7.3.3.2 Viscosity 234 7.4 Discussion 238 7.4.1 Comparison of Physico-chemical Prope rties 238 7.4.1.1 Bulk Rock Composition 238 7.4.1.2 Melt and Magma Viscosities 239 7.4.2 Course and Eruption Styles of the Maero Eruptive Period 243 7.4.2.1 Types of Lava Domes 243 7.4.2.2 Causes of Dome Collapse and Associat ed Deposits 244 7.4.2.3 Reconstruction of the M aero Eruptive Period 247 7.5 Conclusion 251 7.6 References 252 Chapter 8 Conclusions 259 8.1 Avenues of Future Research 263 Appendices I xi List of Tables Table 2-1…………………………………………………………………………………………………..36 Detection limit of element oxides measured at the EMP (University of Auckland) including the deviation from a reference glass composition. Table 2-2…………………………………………………………………………………………………..43 Water and carbon dioxide peaks on the FTIR spectra including their bonds within glasses. Table 3-1…………………………………………………………………………………………………..52 Stratigraphy of the youngest deposits of Mt. Taranaki (Druce, 1966). Table 3-2…………………………………………………………………………………………………..53 Stratigraphy of the youngest deposits at Mt. Taranaki until 2003 (Druce, 1966 ; N eall, 1972 ; Neall, 1979; McGlone et al., 1988; Lees and Neall, 1993). Table 3-3…………………………………………………………………………………………………..54 Stratigraphy of the last 1000 yrs of activity at Mt. Taranaki established in 2003 (modified after Cronin et al., 2003). Table 3-4…………………………………………………………………………………………………..73 Sample list, description and stratigraphy of studied lava flows and scoria-and-ash flow deposits. Table 3-5…………………………………………………………………………………………………..83 New tephrostratigraphy of the Maero Eruptive Period. Table 4-1…………………………………………………………………………………………………106 Glass EMPA of Taranaki (Burrell Lapilli eruption) a nd Ruapehu (14. October 1995 eruption) sorted by classed uncontaminated and contaminated data points. Detection limits for Taranaki glass EMPA only. All data in wt. %. Table 4-2…………………………………………………………………………………………………108 Plagioclase microlite rim and centre compositions (B urrell Lapilli eruption, Mt. Taranaki). All data in wt.%. Table 4-3…………………………………………………………………………………………………112 Calculation of various plagioclase mixing proportions within Taranaki hybrid glass EMPA. EMPA data in wt.%. See text for explanation. Table 5-1…………………………………………………………………………………………………132 Sample location and description of studied summit dome rocks. Table 5-2…………………………………………………………………………………………………134 Comparison of the modelled summit dome to other Taranaki flank lava domes. Table 5-3…………………………………………………………………………………………………156 Historic lava dome eruptions compiled from Newhall and Melson (1983) and the Smithsonian Institution Catalogue. Table 5-4…………………………………………………………………………………………………158 Physical parameters and results for equations 5-4 to 5-6. Table 5-5…………………………………………………………………………………………………160 Physical parameters used for conductive cooling and rainfall-quenching of the lava dome. xii Table 5-6…………………………………………………………………………………………………163 Selected eyewitness reports of the 18 th and 19 th century. Table 6-1…………………………………………………………………………………………………186 Minimum volume and mass calculations of erupted tephra. Table 6-2…………………………………………………………………………………………………191 XRF bulk rock geochemistry of Burrell Lapilli clasts. Samples are mostly from pumice flow deposits with one BAF deposit sample (P10). Table 6-3…………………………………………………………………………………………………193 Mass balance calculation for the inferred fractionation process for pumice-lithics. Table 6-4…………………………………………………………………………………………………195 Representative normalised matrix gla ss compositions. All data in wt. %. Table 6-5…………………………………………………………………………………………………214 Location of pyroclastic flow deposits on the NW and S sector of Mt. Taranaki. Table 7-1…………………………………………………………………………………………………226 Sample list and description of studied pyroclastic flow deposits. Table 7-2…………………………………………………………………………………………………228 Summary of petrographic studies of BAF samples. xiii List of Figures Figure 1-1……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….. 8 Tectonic setting and its relation to Holocene volcanism in the North Island, New Zealand. Area between the Hikurangi Trough and the axial ranges is the fore arc. Arrow indicates Pacific Plate movement with a rate of 42 mm yr -1 . AVF – Auckland Volcanic Field; TgVC – Tongariro Volcanic Centre; TVL – Taranaki Volcanic Lineament; TVZ – Taupo Volcanic Zone. AD and RD refers to andesite and rhyolite dominance within TVZ, respectively. Modified after Reyners et al. (2006 ) and Wilson et al. (1995). Figure 1-2……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….10 Regional tectonic setting of the Taranaki peninsula. The Taranaki Volcanic Lineament comprises the Sugar Loaf Islands (SLI), Kaitake (K), Pouakai (P) a nd Mt. Taranaki (T). Major onshore (thick lines) and offshore (thin lines) faults are indi cated: IF-Inglewood Fault, MF-Manai a Fault, NF-Norfolk Fault, OF- Oaonui Fault. Contours are at 300 m intervals for the volcanic edifices only. New Plymouth (NP) as the major settlement is identified. Modifi ed after Sherburn and White (200 5). Figure 1-3……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….14 Orthophotograph of Mt. Taranaki including the lower northwestern flanks. Inset shows upper cone area including Fanthams Peak. Major morphological features are indicated by letters: a-summit crater of Mt. Taranaki, b-Fanthams Peak, c-the Beehives (two lava domes), d-scarp of the Opua amphitheatre, e-Big Pyramid, f-The Dome, g-Skinner Hill (probably a buried dome structure), h-Pyramid Stream, i-Maero Stream, j-Waiweranui Stream, k-Hangatahua River, l-Egmont National Park boundary (forest/pasture border, here 390 m), m-Kapuni Gorge (marks the eastern border of the amphitheatre), n-Sharks Tooth (second highest peak, 2510 m), o-Fanthams Peak (comprising multiple vents), p-remnant summit dome, q-Turtle, r-Bobs Ridge (western border of the Opua amphitheatre), s-NW flank and main path for BAFs. Figure 1-4……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….16 Classification of Mt. Taranaki and Mt. Ruapehu rocks after Gill (1981 ). Modified after Price et al. (1999). Figure 1-5……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….17 Trace element patterns of selected Mt. Taranaki rocks. Warwicks/Staircase refers to lava flow groups of the main cone of Mt. Taranaki; Fanthams describes lava flows of the satellite vent Fanthams Peak. Data compiled from Price et al. (199 2, 1999 ). Normalisation after Sun and McDonough (1989). Figure 2-1……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….38 The specimen illustrates how pumice cores were obtained. As in this case, six cores in three mutually perpendicular orientations were drilled. Figure 2-2……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….39 Standard deviations for diameter and length of the cores (a) and for volumes V c and V He (b). It is noted that one sample in a) is off the chart at a standard deviation of 0.1321 cm. b) Samples are differentiated into those drilled in Oregon and at Massey. Or egon samples show small variations for both V c and V He . Figure 2-3……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….40 Standard deviation for connected porosities. The limit of 0.3 cm 3 for V c and V He in Fig. 2-2b is used as maximum limit. Cores drilled in Oregon are shown for comparison. Figure 2-4……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….40 Permeability measurements of individual cores were performed three to six times, partially using multiple flow rates, in order to assess re producibility. In this case, the red graph suggests higher flow rates compared to the other three runs and was excluded from further calculations. Figure 2-5……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….44 Separated groundmass glass fraction from a pumice clast (SD3 2). xiv Figure 2-6 ………………………………………………………………………………………………….45 Thermal analysis of volcanic glass to 850 °C. The ch ange in weight (green axis) was measured after the isothermal break at 110 °C. Figure 3-1……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….55 Distribution of lava flows on the Mt. Taranaki main cone and Fanthams Peak. Sampled lava flows and scoria-and-ash flow units are highlighted. The thick solid line represents the Opua amphitheatre scarps. The map has been modified after Neall (2003 ) ; map details were kindly provided by J.N. Procter. Figure 3-2 ………………………………………………………………………………………………….57 Location of type and reference sections around the edifice of Mt. Taranaki. TS-type section, P-Pembroke Road, W-Waingongoro Stream, M-Manaia Road. Numbered black squares refer to reference sections. Black diamonds mark distal locations of pyroclastic flow deposits in the area of Saunders Road – Wiremu Road – Waiweranui Stream. Figure 3-3 ………………………………………………………………………………………………….59 Type section of the Maero Formation located in M aero Stream, NW sector of Mt. Taranaki. Reference sections Nos. 1 and 2 are located in Pyramid Stream and Hangatahua River, respectively. The outcrops comprise deposits of BAFs, surges and pumice flows as well as co-ignimbrite ashes. Debris flow-, lahar-, and fluvial deposits are also exposed but not always differentiated. For further details see text and appendices. Labels Py7 etc refer to sample numbers. See also Fig. 3-4 for field photographs. Figure 3-4……………… …… … … …… …… … … …… …… … … …… …… … … …… …… … … …… ….62 Field photographs of deposits from the NW sector of Mt. Taranaki. a) NW sector of Mt. Taranaki as viewed from the summit. Note the depression in vege tation caused by an avalanche (see Chapter 5). H. R. - Hangatahua River . b) exposure in the upper right Pyramid fork. Cliff section is approx. 30 m high and shows mainly lahar and hyperconcentrated flow deposits. Note Shane Cronin for scale (arrow). c) outcrop on the true left side of the Maero Stream near the intersection of Puniho and Holly Hut track. The upper unit shows the Tahurangi Block-and-Ash Flow deposit (unit a) with its upper matrix-supported and lower clast-supported zones. d) reference section No.1, lower Pyramid Stream, true right side. Only major units are labelled. Discolouration of units IV-Bb and II-B is caused by a raised iron-rich water table. Note the black bar has same height as the spate (arrow). e) close-up of c); basal portion of Tahurangi BAF (unit a) which consists of fine to medium ash. Dashed line marks the boundary between main body of the BAF and its basal portion. f) close-up of d); finely laminated and cross-bedded fine to medium ash surge deposit. Some pumice clasts (arrows) are present. g) exposure on the true right side of Maero Stream. Major units are labelled. The reworked section is 2.5 m thick and represents predominantly fluvial deposits. h) close-up of g); unit IV-Bb. Noteworthy are boulder trains and weak reverse grading of the lower to middle portion. i) distal blast deposit [Newall Breccia (a)] showing its typical pocketing appearance. j) stream exposure of a BAF deposit with pervasive red coloured top portion and grey bottom portion (Waiweranui Stream). k) in situ charcolised tree at a distal BAF exposure [Burrell Breccia (A)] near Saunders Road. l) degassing pipe originating at a charred log within the BAF deposit (same unit as in k). Note that the pipe branches at the centre of the photograph. Same unit as in k). Labelled units in c, d and g refer to the lithostratigraphic code. See di scussion of Chapter 3 for further details. Detailed description of reference section No.1 can be found in the appendix. Photographs of f) and i) were taken by Shane Cronin. Figure 3-5 ………………………………………………………………………………………………….64 Reference sections Nos. 3 and 4 located on the east flank of Mt. Taranaki. Reference site: No.3– Pembroke Road cutting; No.4–intersection of Wain- gongoro Stream and Round-The-Mountain-Track. E02-7 2 etc are sample numbers. xv Figure 3-6 ………………………………………………………………………………………………….66 Reference sections Nos. 5 and 6 located on the south flank of Mt. Taranaki. Reference site: No.5–at the site of the Old Mangahume Hut; No.6–near the Old Mangahume Hut, upslope of No.5. Labels E03-3 5 etc are sample numbers. Figure 3-7 ………………………………………………………………………………………………….67 Reference sections Nos. 7, 8 and 9 located on the west and north sector of Mt. Taranaki. Reference site: No.7–Ahukawakawa Swamp; No.8–Mar upakoko Stream near Kahui Hut; No.9–Parihaka Road cutting. Labels E02-4 etc refer to sample numbers. Figure 3-8 ………………………………………………………………………………………………….68 Crater rim stratigraphy as exposed on its SW side at the entrance of Okahu Gorg e. Four lava flows are identifiable with the youngest flow (4) being known as South Flow. Figure 3-9 ………………………………………………………………………………………………….69 Composite photograph of the upper NW flank of Mt. Taranaki showing named and described morphological features. Figure 3-10..……………………………………………………………………………………………….70 Flow paths and distribution of BAF and surge deposits on the N to W sectors. Crater exit areas: 1-north sector, 2-northwest sector, 3-southwest sector. Figure 3-11………………………………………………………………………………………………...75 Histogram of all glass chemical analyses for SiO 2 (a) and K 2 O (b). Figure 3-12………………………………………………………………………………………………...77 Backscatter-electron microscopy imag es of individual glass shards. Di fferent glass shard shapes within individual samples are illustrated (top: vesicular, partially deformed; bottom: dense and angular). Label in images gives analysis number and approximate beam location (black dot). Grey=glass; light grey=minerals; black=epoxy. a-b) Tahurangi Ash sample (T04-98) with chemically homogenous glass shards but of different texture; a) deformed vesicles, b) dense angular. Note that the presence of large minerals (bottom) may alter vesicle distribution. c- d) Newall Ash sample (E02-75) with distinct shard textures; glass chemistry of shard (d) is similar to those of scoria-and-ash flow glass shard (c). Note regular to irregular vesicles in c). e-f) Burrell Lap illi sample (E03-3 5) with Taranaki glass shard (f) and Taupo volcano-derived glass shard (e), which shows a deformed vesicle. Figure 3-13………………………………………………………………………………………………...78 Glass chemistry of Burrell Lapilli-erupted deposits. The SiO 2 vs. K 2 O diagram shows a general positive correlation. Highest mean silica and potassium cont ents are observed for BA F and surge deposits. The crosses represent the 95% confidence interval of the sample mean. The tephra sample E03-35 is bimodal containing glass shards with a signature similar to Taupo volcano (or atypical of Taranaki). If foreign glass shard analyses are excluded the sample mean is located within the field of BAF/surges (=E03- 35’ ). Also included are sample means of a surge deposit pre-dating the Maero Eruptive Period showing a rhyolitic glass chemistry. It is noted that one sa mple (WW19) shows large variations and a bimodal sample population. Figure 3-14………………………………………………………………………………………………...79 Comparison of mean sample values of scoria-and-ash flow units (T04-53 , T04-56) to pyroclastic pumice flow deposits of the Burrell episode (units 1-3) and Puniho Ash, and other Maero deposits (crosses). Individual glass shards (small black squares) within deposits, other than T04-53 and T04-56, that have compositions similar to scoria-and-ash flows. a) SiO 2 vs. K 2 O, b) CaO vs. FeO. xvi Figure 3-15………………………………………………………………………………………………...80 Correlation of individual tephra and pyroclastic flow units based on field studies and glass chemistry (SiO 2 vs. K 2 O). a) Waingongoro and Waiweranui episodes, b) Newall and Puniho episodes, c) Tahurangi and Te Popo episodes. Figure 4-1 ………………………………………………………………………………………………….99 Locations of Holocene vol canic centres in the North Island of New Zealand: Auckland Volcanic Field, Taupo Volcanic Zone (TVZ), Tongariro Volcanic Centre (TgVC), Mt. Taranaki. Numbers 1-4 refer to distal andesite tephra sites: 1-Onepoto Basin/Pukaki Lagoon/Lake Pupuke (Sandiford et al., 2001 ; Shane and Hoverd, 2002 ; Shane, 2005) , 2-Waikato lakes (Lowe, 1988b), 3-Kaipo Bog (Lowe et al., 1999) , 4- Lake Tutira (Eden and Froggatt, 1996 ), 5-Lake Poukawa (Shane et al., 2002) , 6-Kaimanawa Mts. and Ruahine Ranges (Froggatt and Rogers, 1990 ). Figure 4-2 ………………………………………………………………………………………………...104 Particle textures found in Burrell Lapilli sub-Plinian fa ll deposits. a) pumice clast, type 1, with clear to pale brownish glass; b) hypocrystalline groundmass te xture of type 2 clast showing plagioclase, Fe-Ti oxide and minor clinopyroxene microlites; c) semi-ves icular type 3 clast with brown groundmass glass, large crystals are hornblende; d) for comparison, hypocrystalline groundmass of the present summit dome of Mt. Taranaki with abundant microlites of plagioclase, Fe-Ti oxides and minor clinopyroxene. Scale bars are in µm. Figure 4-3 ………………………………………………………………………………………………...105 Standard deviations of all major oxides are shown for the original glass EMPA dataset and the glass dataset classed as uncontaminated by plagioclase for the Taranaki (a) and Ruapehu (b) samples. A clear reduction in SiO 2 , Al 2 O 3 and CaO variations is observed. See text for further details. All Fe expressed as FeO. Figure 4-4 ………………………………………………………………………………………………...109 Bivariate plots of Al 2 O 3 and K 2 O vs. SiO 2 for Burrell Lapilli (Taranaki) and Ruapehu glass data. Data points with lowest SiO 2 show linear relationship towards mean plagioclase compositions (dashed lines). Figure 4-5 ………………………………………………………………………………………………...110 Bivariate oxide plots for Burrell Lapilli eruption, Tara naki (a-d) and 14. October 1995 eruption, Ruapehu (e-f) demonstrate how contaminated glass analys es were identified. Open symbols are classed uncontaminated, closed symbols represent hybrid gla ss-plagioclase analyses. Da shed lines point towards mean plagioclase compositions. Figure 4-6………………………………………………………………………………………………...114 The glass evaluation procedure cannot be directly applied to BAF deposits as demonstrated for the Burrell Lapilli equivalent BAF deposit. Although data points above a threshold value of 17.9 wt.% Al 2 O 3 clearly embrace contaminated glass analyses, the trans itional data between 17.1 and 17.9 wt.% Al 2 O 3 cannot be uniquely classified. Dashed line points toward mean plagioclase composition. Figure 4-7………………………………………………………………………………………………...115 Glass compositions of Taranaki and TgVC tephras (Lowe 1988b; Lowe et al., 1999; Eden and Froggatt, 1996 and Shane and Hoverd, 2002 ) show large variations, here only shown the means and standard deviations for K 2 O and Al 2 O 3 . Contaminated plagioclase-glass analys es and/or the analysis of two or more particle types may have caused the apparent glass compositional heterogeneity. The small variation in the unmodified Burrell Lapilli dataset (Taranaki) is shown for comparison (in grey). Figure 5-1………………………………………………………………………………………………...123 Lava dome types: a) spiny (Rock Mesa ENE, Oregon), b) spiny-lobate (Mt. St. Helens lava dome, Washington, July 2004) , c) lobate-platy (Big Obsidian Flow, Newberry, Oregon). xvii Figure 5-2………………………………………………………………………………………………...125 NW sector of Mt. Taranaki showing the main deposition area of Maero BAF deposits. The extent of the Tahurangi BAF A is outlined (light grey); unit B is onl y observed in the Maero Stream and is omitted for clarity. The rock-avalanche deposit (RAD) is outlined as observed on aerial photographs from 1959 (mid- grey) with an additional area based on field observations (dark grey). Contour interval is 20 m. The upper right inset shows the general slope inclinations. Figure 5-3………………………………………………………………………………………………...127 Correlation of Tahurangi BAF A and B units and the rock-avalanche deposit across the Pyramid-Maero- Hangatahua area. The exposures are sorted by stream and planimetric distance from source (filled squares in Fig. 5-2). See figure legend for further deta ils (RAD – rock-avalanche deposit). Magnetism was measured by a portable fluxgate magnetometer. For clarity, older exposed units were omitted. Outcrop numbers and profiles refer to Platz (200 1) except S04-1 33. Figure 5-4………………………………………………………………………………………………...129 The remnant present summit dome. a) hemispherical shape of the do me as viewed from the SE; arrow points to a person for scale. Photographed by S.J. Cronin. b) dome amphitheatre as viewed from the W; the arrow points to the summit marker (251 8 m); the dashed lines mark the hydrothermally altered central dome portion. c) northern scar of the amphitheatre show ing listric faults. d) the summit marker is a sub- vertical extrusion penetrating the carapace; note weak columnar jointing; summit marker is highest point of the dome. e) orthophotograph of the summit region of Mt. Taranaki; outline shows mapped deposits associated with the lava dome; note the blocky lava flow to the N; crosses mark sample locations. f) the ‘Three Sisters’ (background) mark the NW border of the intra-crater collapse zone with resulting deposit still preserved in the crater (foreground). Figure 5-5………………………………………………………………………………………………...130 Black and white photograph taken between 1898 and 1901 showing the fresh bouldery rock-avalanche deposit (centre to right). The photograph is taken from the Round-The-Mounta in-Track just west of Maero Stream from the top of a buried lava ridge. The view is NNW towards Pouakai. Photographed by the surveyor H.M. Skeet. Figure 5-6………………………………………………………………………………………………...133 Reconstruction of the Pyramid Dome geometry. a) dash ed white line illustrates the former ideal crater wall position; the solid line marks the inferred dome outline; the black dot is the assumed vent location at the break in slope. b) view of the remnant summit dome from the W. c) top view of the combined paraboloid with a composite elliptical base; dimensions of elliptical radii are given. d) side view of the inferred dome geometry; the dome remnants are in dark grey; the inferred underlying slope on the upper flank is estimated to be c.20°. Figure 5-7………………………………………………………………………………………………...135 Hornblende types. a) type 1 with continuous reaction rim. b) type 2 with discontinuous reaction rim; present only in sample SD1; note individual Fe-Ti oxide crystals are visible. c) type 3 partial to fully replaced hornblende crystals. d) type 1 with observed brown glass fringing the reaction rim; only observed in sample SD6. Scale bar is 100 µm in a) otherwise 25 µm. Figure 5-8………………………………………………………………………………………………...136 Histogram of type 1 hornblende reaction rim thicknesses averaged for crystals and entire samples. xviii Figure 5-9………………………………………………………………………………………………...137 Hornblende compositions of the summit lava dome, Mt. Taranaki. a) Na+K (A-site) vs. Al IV . b) Mg# vs. Si; (c.p.f.–cation per formula unit). For comparis on are shown recalculated pargasitic hornblende compositions of Unzen volcano, Japan (Sato et al., 1999; Browne et al., 2006; Nakada and Motomura, 1999 ) [Unzen matrix refers to groundmass crystals], Soufrière Hills Volcano, M ontserrat (Barclay et al., 1998 ; Rutherford and Devine, 2003), Mt. St. Helens , USA (Rutherford and Hill, 1993), Colima, Mexico (Luhr, 2002 ) and Cerro la Pilita, Mexico (Barclay and Carmichael, 2004 ). Figure 5-10……………………………………………………………………………………………….138 Compositions of Fe-Ti oxide phenocrysts and inclusions in clinopyroxene and hornblende. a) Al vs. Ti. b) Ti/Al vs. Fe 3+ # (c.p.f.–cations per formula unit). Cations are calculated on the basis of 32 oxygens. Figure 5-11……………………………………………………………………………………………….139 Glass compositions of inclusions in clinopyroxene and hornblende. Silica is used as differentiation index. Note that inclusions in different host minerals form separate groups. Figure 5-12……………………………………………………………………………………………….140 Bulk rock compositions of the summit lava dome and the Tahurangi BAF A and B deposits. Dome compositions are distinct to Tahurangi rocks as illustrated for Al 2 O 3 (a), Mg# (b, d) Fe 2 O 3 (e) and Zr (f). Figure 5-13……………………………………………………………………………………………….141 Trace element patterns of the Pyramid Dome and Ta hurangi BAF deposits normalised to N-MORB (a) and chondrite (b). Pyramid Dome rocks and Tahurangi BAF deposits show nearly identical trace element patterns. For the light rare earth elements slightly higher abundances in Tahurangi BAF deposits are noted. Normalisation after Sun and McDonough (1989 ). Figure 5-14……………………………………………………………………………………………….142 Texture of rock sample SD6. a) photograph shows su b-vertical, near parallel crack patterns and cavities. b) modified image of a) highlighting cracks and cavities in black. Figure 5-15……………………………………………………………………………………………….144 Lava dome growth patterns are illustrated in a-e. Exogenous and endogenous dome growth occurred simultaneously. f) demonstration of the inferred exogenously (dark grey) and endogenously (light grey) formed surfaces as observed on the dome remnants. Figure 5-16……………………………………………………………………………………………….147 Reconstruction of the summit dome failure. a) erosion scars on upper flank (white arrows) as well as the curvatures of the amphitheatre, and th e scars to the SW and S define the geometry of individual collapse sectors. b) dome geometry with individual sectors I-IV and their flow directions. c) cross-section of the dome showing the dome remnants (grey), and the disintegration of dome rocks along listric faults. Figure 5-17……………………………………………………………………………………………….149 Aluminium-in-hornblende geobarometer shown as histogram for hornblende phenocrysts (core and rim) and microphenocrysts. Calculated after Johnson and Rutherford (1989) and corrected by -1.5 kbar. Figure 5-18……………………………………………………………………………………………….150 Comparison of calculated hornblende crystallisation pressures for various Mt. Taranaki rocks and xenoliths. Granodiorite xenoliths contain all requ ired mineral phases for the Al-in-hornblende geobarometer (Johnson and Rutherford, 1989 ) and therefore were not corrected. Note that some hornblende crystals of hornblende gabbros and hornblende-pyroxene gabbros indicate crystallisation below (<1 kbar) the inferred hornblende stability limit. xix Figure 6-1 ………………………………………………………………………………………………...179 Mount Taranaki (lower right) has produced mainly lava dome eruptions in the past 800 years with Block- and-Ash Flow deposits making up the fan between the Maero and Pyramid Stream and in the Hangatahua River (BAF – Block-and-Ash Flow, ppf – pumice pyrocl astic flow). Star (top left) indicates the most distal outcrop discussed in the text. To the NNW of Mt. Taranaki are the south flanks of Pouakai volcano. The inset shows the Taranaki peninsula with the Tara naki Volcanic Lineament (S LI-Sugar Loaf Islands, K-Kaitake, P-Pouakai, T-Taranaki). Major onshore a nd offshore faults: IF-Inglewood Fault, MF-Manaia Fault, NF-Norfolk Fault, OF-Oanui Fault. Contours are 300 m. Modified after Sherburn and White (200 5). Figure 6-2 ………………………………………………………………………………………………...181 Variations in bulk vesicularity and connected porosity of single clasts for grey pumice (a), banded pumice (b), and black and brown pumice (c). Crosses represent the range in vesicularity per clast using minimum, mean and maximum values (see inset in a). Variations in bulk vesicularity refer to single cores cut in two ( φcore) and overall clast variations with multiple cores ( φclast). Note different scale in b). See text for details. Figure 6-3………………………………………………………………………………………………...183 Field photographs of a) succession of three pumice pyroclastic flow deposits on the upper south flanks; sketch shows a general assembly of pumice types and gr ey dense lithics, b) grey pumice clasts of unit 3, c) eroded surface into unit 2 show ing the scattered grey pumice [1] from airfall, brown pumice [2], banded pale grey to dark brown pumice [4], and the dense fractured andesite clasts [L]; d) lower contact of a distal BAF deposit, c.13.5 km from source (star in Fig. 1). See text for field description. Figure 6-4………………………………………………………………………………………………...184 Distribution of Burrell Lapilli deposits: a) isopachs in cm including the BAF deposit (black) to the NW for reference, black squares represent mapping locati ons for fall deposits only, P-Pouakai, contours 100 m; b) isopleths for pumice clasts an d pumice pyroclastic flow deposits on the upper flanks (black), see inset in a) for location; c) isopleths for lithic clasts, same outline as in b). Numbers in b) and c) are clast diameters in cm. Figure 6-5………………………………………………………………………………………………...187 Thin-section photographs illustrating basic vesicularity di fferences of juvenile clasts. a) dense grey lithic, b) semi-vesicular black pumice, c) vesicular black pumice with isolat ed and coalesced vesicles, crosses mark plagioclase crystals; d) grey pumice with isolated large single vesicles as well as larger coalesced vesicle. Scale bar is 100 µm in a-c an d 500 µm in d. See text for details. Figure 6-6 ………………………………………………………………………………………………...188 SEM images of grey (a) and brown (b) pumice (note a and b are binarised; black=vesicles, white=glass + crystals); c) shows a large coalesced vesicle; d) co nsists of three SEM images showing the transition from grey to brown in banded pumice. Scale bar is 10 µm in c), otherwise 100 µm. Figure 6-7………………………………………………………………………………………………...189 Hornblende reaction textures in different clast types: a) fresh hornblende with no reaction rims in pumice, b) single Fe-Ti oxide crystals are attached to the hornblende rims in black semi-vesicular pumice, c) hornblende in dense grey lithic clast shows resorption textures and is partially replaced by clinopyroxene, plagioclase and Fe-Ti oxide crystals or is fully replaced (lower left); note abundant plagioclase microlites in groundmass. Scale bar is 100 µm. Figure 6-8………………………………………………………………………………………………...190 Bulk rock geochemistry of pumice and grey andesite clasts in a multi element oxide vs. SiO 2 diagram. The calculated fractionation trend pumice – grey lithics is in good agreement for the majority of clasts (solid line) with some variation for the most evolved clast (dashed line). See Table 6-3 for details. xx Fig. 6-9…………………………………………………………………………………………………...194 Groundmass glass compositions of pumice types presented in the Al 2 O 3 vs. SiO 2 diagram. Modelled glass composition changes due to plagioclase and clinopyroxene crystallisation (thick solid line) and is in good agreement with linear regression line (thin solid line). The small inset shows six data points of one lapillus (SD20) demonstrating relative glass homoge neity. The dimensions of the box are 1 wt. % for Al2 O 3 and SiO 2 . Figure 6-10……………………………………………………………………………………………….196 Connected porosity vs. bulk vesicularity of all pum ice types. Brackets represen t 95% confidence limit for the mean of each pumice population. Solid lines represent 0% and 10% and dashed line 5% isolated pore volume. Figure 6-11……………………………………………………………………………………………….197 Connected and bulk vesicularity vs. permeability. a) da ta of this study with upper and lower data limits (black lines) of y=5×10 -19 x -4.5314 and y=6×10 -21 x -4.5314 , respectively. Note there are six specimens with three cores cut in three mutually perpendicular directions. Upper inset shows cores cut in two perpendicular directions. b) comparison of our data with published literature: Montserrat (Melnik and Sparks, 2002 ), Big and Little Glass Mountains (Rust and Cashman, 2004), Pichincha (Wright et al., 2007 ) ; grey lines are limits of Klug and Cashman (1996). Figure 6-12……………………………………………………………………………………………….202 Reconstruction of eruptive events during the Burrell Lapilli eruption. Changes in bulk rock silica contents are illustrated in Stage a. Bubble nucleation levels 1- 3 correspond with erupted units 1-3. See text for further details. Figure 6-13……………………………………………………………………………………………….213 Correlation of pyroclastic flow deposits associated with the Burrell episode. Note that the major unit from medial to distal represents the Burrell Breccia (A). In section S04-133, the thin pyroclastic pumice flow deposits represent Burrell Breccia (B ) units 1-3. Exposures are sorted by stream and planimetric distance from source. See figure legend for further details (RAD – rock-avalanche deposit). Magnetism was measured by a portable fluxgate magnetometer. For clarity, older exposed units were omitted. Outcrop numbers and profiles refer to Platz (200 1) except S04-1 33. For list of samples and coordinates of outcrops see Appendix B. Figure 6-14……………………………………………………………………………………………….215 Estimates of total water and carbon dioxide contents in melt inclusions. a) total H 2 O at 3550 cm -1 vs. molecular H 2 O at 1630 cm -1 , and b) molecular CO 2 at 2350 cm -1 vs. total H 2 O at 3550 cm -1 . Figure 6-15……………………………………………………………………………………………….216 Different shapes of melt inclusions in clinopyroxene. a) overview of crystal 2 (sample SD20) ; note the many inclusions of glass, plagioclase, apatite and Fe-Ti oxide which are mostly oriented along crystallographic planes, b) two close-ups as marked in a); I - represents common but very small melt inclusions found in clinopyroxene, which are unsuitabl e for FTIR analysis. Their shape is near spherical to ovate; II - the bottle-neck shape is typical for leaked melt inclusions, c) overview of crystal 11 (sample SD32) showing a large irregular shaped melt inclusion, d) close-up of c) showing the impossibility of using these inclusions for FTIR analysis; it can be assumed that the inclusion extends further into the crystals as indicated by the diffuse outline of the melt inclusions further to the right, e) section of crystal 4 (sample SD20) showing two types of melt inclusions, the reddish-brown coloured inclusions are probably altered in comparison to the brown inclusions to the right; note again the irregular outline of the inclusions, f) section of crystal 8 (sample SD9D) ; abundant sheet-like inclusions probably oriented along crystallographic planes; the inclusions around the Fe-T i oxide inclusions (black) appear to be connected. Scale bars are 100 µm in a), c), and e), 50 µm in d) and f), and 10 µm in b). xxi Figure 6-16……………………………………………………………………………………………….218 Preliminary results of the thermal analysis studies. a) trial and error series of sample SD20, b) reproducibility results of different pumice samples; it is noted that for the same sample the maximum weight loss is often observed at similar temperatures. Figure 7-1………………………………………………………………………………………………...229 Groundmass texture (a-d) and crystallin ity (e-h) of clasts from pyroclas tic flow deposits. a and c) two groundmass glasses, b-d) differences in degree of groundmass crystallisation in clear translucent and brown glasses, e) semi- to hyaline, clear translucent glass; note microvesicularity, f) same image as in e) under crossed polarised light, g) semi- to holocrystalline brown glass, h) same image as in g) under crossed polarised light. Figure 7-2………………………………………………………………………………………………...231 Bulk rock composition of selected Maero eruptives. Block-and-Ash Flow depos its are not differentiated and the Pyramid Dome and the Turtle are omitted for cl arity. For comparison, selected lava flows of the upper main cone and Fanthams Peak are plotted. Mg#=100 [ Mg 2+ /(Mg 2+ +Fe 2+ )]; all iron as Fe 2+ . Figure 7-3………………………………………………………………………………………………...233 Analytical totals of all EMP glass analyses (a) and sample averages (b) are plotted against silica content. Estimated glass water contents using the water-by-difference method (WBD) are shown on the right axis. The terms andesite, dacite, and rhyolite refer to the TA S-classification scheme of Le Maitre et al. (1989 ). Figure 7-4………………………………………………………………………………………………...236 Calculated melt viscosities, η, are plotted against silica abundances. Va lues of the models of Shaw (1972) and Hui and Zhang (2007) are shown for H 2 O contents of 0.1 wt.%, 1 wt.% and WBD at T=900 °C and P=1 bar. Solid and dashed lines are regression lines of η at WBD for the Hui-and-Zhang- and Shaw- models, respectively. Figure 7-5………………………………………………………………………………………………...237 Calculated magma viscosities, ηa, are plotted against SiO 2 contents. Lower and upper crystal volume fractions of 30% (a) and 55% (b), respectively are us ed for the calculation based on the calculated melt viscosities (see Fig. 7-4). Viscosities are calculated using H 2 O contents of 0.1 wt.%, 1 wt.% and WBD at constant T=900 °C and P=1 bar. Solid and dashed lines are regression lines of ηa at WBD for the Hui- and-Zhang- and Shaw-models, respectively. Figure 7-6………………………………………………………………………………………………...238 Bulk SiO 2 contents (a) and Mg# (b) are plotted against K 2 O in chronological appearance of eruption episodes. Figure 7-7………………………………………………………………………………………………...241 Calculated Mt. Taranaki melt viscosities are compared to calculated melt viscosities of Merapi volcano (Indonesia), Soufrière volcano (St. Vincent), and So ufrière Hills Volcano (Montserrat), using the same parameters. Solid lines are regression lines for Taranaki data. xxii Figure 7-8………………………………………………………………………………………………...242 Calculated Mt. Taranaki magma viscosities plotted against SiO 2 are compared to other andesite to rhyolite volcanoes. Since Taranaki viscosity calculations are based on glass chemical compositions of the Maero Eruptive Period, the range in bulk silica contents of rocks erupted during this period are used to allow comparison to published data. Upper and lower viscosity abundances are taken from Fig. 7-5. Taranaki data are illustrated by two parallelograms with upper and lower limits representing crystal volume fractions of 55% and 30%, respec tively. The grey parallelogram corresponds to 1 wt.% melt water content, whereas the dashed parallelogram relates to water contents determined by WBD. Data source: silicic lava flows (Murase and McBirney, 1973 ; Fink, 1980 ; Navarro-Ochoa et al., 2002 ; Manley, 1996 ; Harris et al., 2004; and McKay et al., 1998); Mt. St. Helens (Murase et al., 1985 ; Scandone and Malone, 1985 ); Unzen volcano (Suto et al., 1993; Goto, 1999 ; Sato et al., 1999 ); Soufrière Hills Volcano, Montserrat (Voight et al., 1999; Sparks et al., 2000 ); Soufrière volcano, St . Vincent (Huppert et al., 1982 ) ; Merapi volcano (Siswowidjoyo et al., 1995). Chapter 1 Introduction 1 Chapter 1 Introduction 1 Introduction Chapter 1 provides information as to why this study was undertaken and what was known about Mt. Taranaki prior to this study. 1.1 Introduction The unpredictable eruption behaviour of volcanoes erupting intermediate to silicic magmas makes them exceptionally hazar dous to surrounding populations. Typically, andesitic volcanoes erupt only small volumes during any one episode (<1 km 3 ). Despite this, they can be extremely e xplosive and during the course of an eruption, the style and types of activity can change rapidly, une xpectedly and repeat edly. A number of cataclysmic eruptions in the last century have underscored this point. Three key eruptions underscored how little we knew a bout such volcanoes, before focussing world scientific attention on andesite systems and their eruption mechanisms: Mt. Pelée, Martinique, Mt. St. Helens, USA, and Soufrière Hills Volcano, Montserrat. Within a few minutes on the morning of May 8, 1902, a single violent phase of an ongoing dome-building eruption at Mt. Pelée (Martinique) t ook the lives of over 23,000 people in the town of St. Pierre (Lacroi x, 1904). This event changed forever our awareness of volcanic hazar ds and raised the spectre of the hitherto unknown Chapter 1 Introduction 2 phenomenon of nuées ardentes. Since then “pyroclastic de nsity currents” have been recognised as a common and worldwide volcanic process involving hot (often >300 °C), rapidly flowing (often >100 ms -1 ) mixtures of gas and particle s, with a range in particle sizes and particle/gas ratios (Moore and Melson, 19 69; Nairn and Self, 1978; McClelland and Druitt, 1989; Scott and Glasspool, 2005; Zane lla et al., 2007). Following this, physical volcanology resear ch has focussed on understanding the driving mechanisms behind the variety of “ pyroclastic density currents”, such as pyroclastic flows and surges, along with specif ic sub-categories of these, including ash flows, Block-and-Ash Flows and scoria-a nd-ash flows (e.g. Smith and Bailey, 1966; Schmincke, 1974; Rodríguez-Elizarrarás et al ., 1991; Bourdier et al., 1997). In addition, detailed sedimentological studies of the deposits of pyroclastic density currents (e.g. Walker et al., 1980; Wright and Walker , 1981; Wilson and Walker, 1982; Boudon and Lajoie, 1989; Boudon et al., 1993; Cas and Wright, 1991; Fisher et al., 1993; Abdurachman et al., 2000; Lube et al., 2007), have been used to interpret (and debate) a variety of eruptional, transportational and depositional processes (Sparks, 1976; Branney and Kokelaar, 1992; 2002; Dade an d Huppert, 1996). Furt hermore, physical and numerical models are under ongoing development to better describe and simulate flow and transport mechanisms based on observational and experimental studies of deposits (Battaglia, 1993; Gi ordano and Dobran, 1994; Fuji i and Nakada, 1999; Choux and Druitt, 2002; Freundt, 1998; 2003; Fre undt and Bursik, 1998; Lube et al., 2004). A stupendous “blast”, triggered in only a few seconds at 8:32 am marked the onset of the 18 May 1980 Mt. St. Helens eruption and caused 57 fatalities, despite the intense focus of scientists on this re-awakening volcano (cf. Lipman and Mullineaux, 1981). This event fundamentally changed our understanding of how eruptions are triggered and how rapidly andesitic volcanoes can generate explosive phases (e.g. Swanson et al., 1983; Blake, 1984; Pallister et al., 1992; Klug and Cashman, 1994; Aldibirov and Dingwell, 1996; Gardner et al., 1996; Blundy and Cashman, 2001; Scandone et al., 2007). Precursors of the 18 May 1980 eruptio n were known for three months but the size of the eruption and its outbreak mechan ism were unforeseen. The intense scientific focus on this eventually 6-year long eruptive episode has resulted in a new generation of models that combine magma ascent processes (Scandone and Malone, 1985; Rutherford and Hill, 1993) with physico/chemical conduit and eruption processes (Dobran, 1992; Melnik and Sparks, 2002; Cashman and H oblitt, 2004; Cashman and McConnell, 2005; Blundy et al., 2006), as well as providing an understanding of the physical dynamics of Chapter 1 Introduction 3 sub-Plinian eruptions (Carey and Sigurd sson, 1985; Papale and Dobran, 1994), and a variety of mass flows. This long and complex episode also allowed detailed description of the emplacement, growth and destruction mechanisms of lava domes (Moore et al., 1981; Anderson and Fink, 1989; Fink et al., 1990; Swanson and Holcomb, 1990; Fink et al., 1992; Anderson et al., 1995). The now 12-year long eruption episode of S oufrière Hills Volcano on Montserrat, has also repeatedly and tragically demonstrated the destructive potential of andesite volcanism, even whilst under th e intense observation of scientists and local authorities. The most devastating activity occurred on June 25, 1997, when a retrogressive partial collapse of the dome generated pulsatory Bl ock-and-Ash Flows and pyroclastic surges that killed 19 people and in jured 7 others (Sparks and Y oung, 2002). This demonstrated that even with our increased knowledge of these systems, andesite volcanoes remain enigmatically diverse, unpredictable and in tensely hazardous. Ongoing studies on this and other recent long-term dome-forming eruptions with similar tragic results from sudden state-changes (from Mts. Unzen, Merapi, and Colima) have helped to derive a new understanding of dome-forming eruption proc esses. A wealth of data have been collected on the dynamics of long-term a nd often sporadic lava-dome growth (e.g., Sparks, 1997; Sparks et al., 1998; Nakada et al., 1995; 1999; Vo ight et al., 1999; Sparks et al., 2000; Harford et al., 2003;) , as well as on the variety of products related to dome collapse and destruction (e.g., Cole et al ., 1998; 2002; Ui et al., 1999; Camus et al., 2000; Grunewald et al., 2000; Vo ight and Elsworth, 2000; Sau cedo et al., 2002; Carn et al., 2004; Simmons et al., 2004; 2005; Voi ght et al., 2006). The major advance generated as a result of study into these eruptions, however , has been the combination of petrological, geochemical, geophysical and physical methods to comprehend the rise, evolution and eruption of dome-forming magm as from within the crust to the dome surface (e.g., Sato et al., 1999; Power et al ., 2002; Couch et al., 2003; Devine et al., 2003; Higgins and Roberge, 2003; Rutherford and Devine, 2003; Cashman and McConnell, 2005). Experimental and numerical modelling inspired by these events have also lead to major advances in unde rstanding the dynamics and rheology of these complex magmas (e.g., Alidibirov et al., 1997; Navon et al., 1998; Alidibirov and Dingwell, 2000; Massol et al., 2001; Lyman et al., 2004; Mueller et al., 2005). Despite this, however, the sudden inte rruption of “quiet” dome erup tions by violent Vulcanian and sub-Plinian events is not always well understood. These sudden transitions between eruptive styles remain difficult to forecast. Chapter 1 Introduction 4 At historically and pre-historically active andesite volcanoes, sequences of pyroclastic deposits attributed to the same types of processes are commonly preserved in the geological record. These sequences also ofte n demonstrate apparent rapid changes in deposition mode and hence inferred eruption style. Such sequences and transitions can be now better interpreted in relation to the insights gained from historic eruption sequences (as described above) to provide ha zard assessments. On the other hand, over longer periods of the geological record, wh ere deposits of many eruption sequences are exposed, there is also a potential to interp ret a wider range and diversity of eruption sequences and transitions that can help to predict sudden changes during future ongoing eruptions. Eruption styles and types similar to thos e exhibited during th e three catastrophic eruptions described above have been used to interpret the depositional record of the past 1000 years at Mt. Taranaki, New Zealand (Topping, 1972; Neall, 1979; Platz, 2001; Cronin et al., 2003). The main features of this record have been the identification of a range of highly variable eruption styles that surprisingly resulted from eruption of very similar magma compositions. Important transiti ons in the style of eruptions have been noted, including (1) an initially extrusive, lava dome-forming event to an explosive, sub-Plinian event (Burrell Lapilli AD 1655), (2) a dome emplacement punctuated by a highly explosive lateral bl ast (Newall episode, ~ AD 1400), as well as (3) a variety of less violent emplacement, growth and destru ction episodes involving lava-domes. Pyroclastic deposits recording this series of eruption episodes are well exposed on the flanks of Mt. Taranaki, along with the most recent dome (previously thought to have been associated with the Tahurangi eruption) which remains in half-section at the volcano’s summit. These conditions present an excellent example and location to examine some aspects of the physico-chemical controls on eruption episodes, including transitions between effusive and explosiv e phases. Through a combined physical and petrological approach, estimates can be gi ven about magma storage depths, ascent and extrusion rates. These data, in combination with reconstruction of detailed physical eruption processes through field studies, can be used to help constrain some of the controls on eruption dynamics at andesitic stratovolcanoes such as Mt. Taranaki. On a local scale, an improvement in the hazard assessments for Mt. Taranaki can be achieved by reconstructing in detail the course of previous eruption episodes. The variety of data sets provide a sound foundation for understanding eruption precursors, and possibly also precursors for anticipating rapid changes in state during eruption episodes. Chapter 1 Introduction 5 1.2 O b jectives and Strategy The aim of this study is to investigate the causes and characteristics of differing eruptive mechanisms experienced during the last 1000 ye ars of activity at Mt. Taranaki. To carry this out, robust stratigraphic, geochemical, petrological, and physical investigations were required to firstly reconstruct the past volcanic episodes at Mt . Taranaki. This also involved developing methods for geochemically and otherwise correlating pyroclastic units emplaced in diverse sectors of the volcano by a range of eruptive processes. From this reconstructed record, specific pe riods and deposits/extrusi ves were chosen to concentrate on the major sub-aims of the thesis work: 1. understanding causes for rapid shifts from extrusive to explosive eruptive styles during eruptive episodes, 2. interpreting causes of lava dome emplacement and failure, 3. developing models for magma assembly and rise, and constraining late-stage (high-level, conduit, dome) magmatic processes and their impacts on eruption style and the texture/petr ology of eruptive products. Sub-objective 1 work focussed on the AD 1655 sub-Plinian Burrell Lapilli eruption, the largest event of the Maero Eruptive Period. The remnants of the present summit dome were the focus for completing sub-objective 2, since this allowed a detailed view through a section of a dome that apparently collapsed without related explosive activity. For deposits of the entire se quence, detailed analyses of different mineral and glass phases along with whole rock analyses were carried out toward answering the third sub- objective. In addition, physical properties such as porosity and permeability of pumice and dense dome rocks were investigated. To further constrain late-stage eruption processes, estimates of water contents in gl ass inclusions (hosted in minerals) and in matrix glass were also made. Chapter 1 Introduction 6 1.3 Thesis Outline The remainder of this chapter provides a background to Taranaki volcanism, and this is followed by another six chapters comprising th is thesis. Chapter 2 outlines the methods applied in this study and the strategy for determining choice of methods (within the constraints of available analytical tools). Chapter 3 outlines the reconstruction of the Maero Eruptive Period stratigraphy and interpre ted sequence of eruptive episodes. This study resulted in a re-interpr etation of the previously defined eruption sequence and enabled collection of many new age determinations of eruption episodes and events. Chapter 4 describes the problems and reasons in fingerprinting andesitic tephra layers and introduces a new procedure to identify contaminated electron microprobe glass analyses. As a result, correlation of andesite tephras using glass chemistry may facilitate a new degree of resolution in tephrostratigraphic records. The remnant summit dome is the subject of Chapter 5. The lava dome is interpreted to represent a single eruptive event following the AD 1755 Tahurangi eruption. Physico-chemical conditions of dome formation are investigated and implications are given for magma storage depth(s), magma extrusion and ascent rate s. Chapter 6 describes the AD 1655 Burrell Lapilli eruption in detail, which bega n with an effusive, dome-forming phase and terminated in an explosive, sub-Plinian er uption. A model is proposed that explains the diversity of erupted ejecta. In Chapter 7 petrographic and geochemical data are used and interpreted to reconstruct the variety of eruption mechanisms that occurred during the Maero Eruptive Period. All collected datasets are documented in the electronic appendices (on DVD). The printed appendix contains stratigraphic colu mns and all geochemical analyses collected by electron microprobe, XRF analysis , and Laser-ICP MS analysis. Chapter 1 Introduction 7 1.4 Background Geology 1.4.1 Regional Geological Setting New Zealand is located on the convergent plate boundary of the Pacific and Australian Plates. The North Island is situated where the oceanic Pacific Plate is subducted beneath the continental crust of the Australian Plate. The Hikurangi Trough marks the southernmost extremity of the Tonga-Kermad ec trench and begins offshore to the east of the most southern part of the North Isla nd (Fig. 1-1). Along the trench the relative plate motion decreases from north to south with relative plate convergence rates of up to 98 mm yr -1 in the Tonga trench and 42-50 mm yr -1 in the Hikurangi Trough (Minister and Jordan, 1978; de Mets et al., 1994). The occurrence of an oceanic plateau, the Hikurangi Plateau, to the east of the lower North Island causes resistance to subduction. The collisional resistance of the Hikurangi Plateau in the south and subduction of normal oceanic crust in the north of the Hi kurangi Trough, forces clockwise rotation of microplates in the forearc, the region betw een the trough and the axial ranges, also creating a spreading centre beyond the axial ranges where vol canism in the North Island is concentrated (Wallace et al., 2004). South of the Hikurangi Trough, the plate movement contains a greater margin-paralle l component and passes into a strike-slip fault system that accommodates most of the relative plate motion in the South Island (van Dissen and Yeats, 1991; de Mets et al., 1994). Under the North Island, the Pacific Plate plunges steeply into the mantle with an approximate depth of 100 km below the c .40 km thick continenta l crust beneath Mt. Ruapehu (central North Island) at a distance of c.280 km from the tren ch (Reyners et al., 2006). The depth of the subducted slab bene ath the Taranaki region (western North Island), located abou t 400 km from the tr ench, is approximately 250 km (Boddington et al., 2004; Reyners et al., 2006) . However, the Taranaki volcanoes are not directly overlying the slab since it is only traceable to the southeastern part of the Taranaki region, c.35 km SE of Mt. Taranaki (per s. comm. S. Sherburn, 2006). Isolated deep earthquakes with foci centred beneath Mt. Taranaki occur at depths of c.600 km and are interpreted to be derived from a horizon tal lying, detached sliver of the plate (Boddington et al., 2004). Quaternary volcanism in the North Island is generated in two regions within the subduction zone: the Taupo Volcanic Zone (T VZ), including the Tongariro Volcanic Centre (TgVC) and the Taranaki Volcanic Li neament (Fig. 1-1). The TVZ is one of the Chapter 1 Introduction 8 most productive rhyolitic magmatic systems on Earth caused by hi gh heat flows up to 700 mWm -2 in a rapidly extending region (e.g. Stern, 1987, Wilson et al., 1995). The TVZ can be subdivided into three zones based on the main eruptive rock types: the NE (e.g. White Island) and SW (TgVC) are char acterised by andesite provenance, whereas the central part is dominated by rhyolites (W ilson et al., 1995). The TgVC is located at the apex of the extending TVZ (i.e. lowest spreading rate) and consists of Mts Ruapehu and Tongariro with the satellite cone of Mt. Ngauruhoe (Donoghue et al., 1995; Cronin and Neall, 1997; Hobden et al ., 2002). The Taranaki Volcan ic Lineament is located 130 km to the west of Mt. Ruapehu and is geochemically and tectonically distinct from the TVZ setting. Other Cenozoic off- and onshore basalt and andesite volcanic centres are located farther north of Taranaki volcanism, for instance the Alexandra, Okete and Ngatutura Volcanics as well as the offs hore Awhitu volcanic centre (Briggs, 1983; Briggs et al., 1989; Stagpoole, 1999). Figure 1-1: Tectonic setting and its relation to Holocene volcanism in the North Island, New Zealand. Area between the Hikurang i Trough and the axial ranges is the forearc. Arrow indicates Pacific Plate movement with a rate of 42 mm yr -1 . AVF – Auckland Volcanic Field; TgVC – Tongariro Volcanic Centre; TVL – Taranaki Volcanic Lineament; TVZ – Taupo Volcanic Zone. AD and RD refers to andesite and rhyolite dominance within TVZ, respectively. Modified after Reyners et al. (2006 ) and Wilson et al. (1995 ). Chapter 1 Introduction 9 1.4.2 Taranaki Basin The Taranaki peninsula, on the west coast of the North Island, lies within the Taranaki Basin (Fig. 1-2). The eastern boundary of the basin is marked by the Taranaki Fault, a crustal thrust fault that has accommodated a dip-slip displacement of 12-15 km in the last 80 Ma and vertically o ffsets the basement by about 6 km (King and Thrasher, 1996; Nicol et al., 2004). The Cape Egmont Fault Zo ne (CEFZ) to the west of the peninsula divides the basin into the Eastern M obile Belt dominated by subduction related extension and compression processes, and the Western Stable Platform (King and Thrasher, 1996). The offshore Taranaki CEFZ represents the limit of deformation related to the convergence of the Pacific and Australian Plates. It is marked by extension and shows a 53 km long and 1-5 m high scarp on the seafloor; average dip- slip movement for the last 225,000 yrs is c.0.8 mm yr -1 (Nodder, 1993). Three active onshore NE-SW-trending faults are identified: the Oaonui Fault to the SW and the Inglewood and Norfolk Faults to the NE of Mt. Taranaki (Fig. 1-2). The Oaonui Fault shows a vertical displacement of 5 m over the last 6500 yr s, whereas the Inglewood has a 3 m scarp with >1 m vertical displacemen t in the last 13,000 yrs (Hull and Dellow, 1993; Hull, 1994). The three active faults ar e oblique normal slip faults and are interpreted to be related to magmatic intrusions at Mt. Taranaki (Sherburn and White, 2005). Sherburn and White (2005) specula te whether the last Inglewood Fault movement, 3-3.5 ka, is related to the formation of the satellite cone (Fanthams Peak) and two lava domes on the south flanks of Mt. Taranaki, since they appear to be of similar age. The Taranaki Basin is on average 6 km thick, reaching 8 km 22 km SSE of Mt. Taranaki (King and Thrasher, 1996). It is composed of a Cretaceous to Cenozoic sedimentary sequence of limestones, coals, silt-, and sandstones, which in onshore sequences are covered by volcan iclastic deposits up to 300 m thick (cf. King and Thrasher, 1996). On- and offshore oil and ga s reserves within the sediment sequence have led to extensive geophysical studie s providing invaluable stratigraphic and lithologic data on the upper to middle crust. In two oil-exploration wells, two andesite sills of 5-8 m thickness are found near Ne w Plymouth at depths of c.2800 m (Allis et al., 1995). However, the baseme nt geology beneath Taranaki is less well constrained and interpretations are mainly based on seismic profiles, basement rocks drilled offshore, and inferred related rocks that are exposed in the South Island. Mortimer et al. (1997) interpreted the Taranaki basement as mainly composed of calc-alkaline Chapter 1 Introduction 10 subduction-related plutonic rocks which are pa rt of a major terrane boundary unit, the Median Tectonic Zone. Figure 1-2: Regional tectonic setting of the Taranaki peninsula. The Taranaki Volcanic Lineament comprises the Sugar Loaf Islands (SLI), Kaitake (K), Pouakai (P) and Mt. Taranaki (T). Major onshore (t hick lines) and offshore (thin lines) faults are indicated: IF-Inglewood Fault, MF-Manaia Fault, NF-Norfolk Fault, OF-Oaonui Fault. Contours are at 300 m intervals for the vol canic edifices only. New Plymouth (NP) as the major settlement is identified. Ka- Kapuni Gas Field. Modified after Sherburn and White (200 5). Heat flow measurements in oil explorati on wells on- and offshor e Taranaki peninsula revealed an anomalous high heat flow zone rising to 73 mWm -2 in New Plymouth with an average heat flow of c.60 mWm -2 for the entire Taranaki Ba sin (Allis et al., 1995; Funnel et al., 1996). Geotherms calculated for 50 mWm -2 (Kapuni, SSE of Mt. Taranaki) and 70 mWm -2 (New Plymouth) correspond to temperatures of 138 °C at 6 km depth and 160 °C at 5 km depth, respectiv ely (Allis et al., 1995). A higher heat flow than 50 mWm -2 for two Kapuni wells (>5 km) is suggested and could be explained by cooling effects of cross-flowing ground water in the upper 3 km of sediments (Allis et al., 1995). Thermal modelling showed that th e heat anomaly beneath New Plymouth is either caused by 5 km of crustal underpla ting over the last 2-4 Ma or by mid-crustal intrusions up to 500 m thick over the last 0.2-0.5 Ma. Recent crustal intrusions could be Chapter 1 Introduction 11 indicated by a conductive heat pulse in a Ka puni well with an observed increase in the temperature gradient below 4 km (Allis et al., 1995). Groundwater flow modelling beneath Mt. Tarana ki predicts water circulation down to 3 km below sea level with lateral meteoric flow between 2-3 km b.s.l. due to presence of sandstone units (Allis et al., 1997). Although Allis et al. (1997) did not discuss the groundwater circulation in the context of heat flow variations within the Taranaki Basin, a decrease in observed heat flow along the Taranaki Volcanic Lineament towards Mt. Taranaki could also be caused by superimpos ed cooling effects of groundwater flows. The recent “disappearance” of warm springs such as Arawhata Spring (Mongillo and Clelland, 1984), c.19 km SW of Mt. Taranaki near/at the Oaonui Fault, the Waiweranui Stream (literally meaning “large hot water”), and the existence of a hot spring near The Dome (between Mt. Taranaki and Pouakai) until at least 1882 (T he Taranaki Herald, 1871; Scanlan, 1961), may be evidence for te mporal lateral groundwater and heat transfers following volcanic activity. High permeabilities of volcaniclastic deposits of the ring plain and Tertiary sandstones, w ith groundwater flow rates of 5 cm day -1 (Allis et al., 1997), may be able to disguise h eat flows and reduce them to the observed 60 mWm -2 at Mt. Taranaki by late ral groundwater movements. The cold Kokowai Springs (c.1200 m a.s.l.) located c.2.5 km NNE of the summit of Mt. Taranaki contain magmatic carbon dioxide and hydrogen sul phide (Childs et al., 1986). Gravity and aeromagnetic studi es along the Taranaki Volcan ic Lineament show large intrusive bodies beneath the volcanoes, probably reflecting complex, multiple dyke intrusions (Locke et al., 1993; 1994; Lock e and Cassidy, 1997). The solidified andesite dyke complex beneath Mt. Taranaki can be m odelled as a cylindrical intrusive body of approx. 5 km in diameter with a density of 2.7 gcm -3 compared to densities of 2.25-2.6 gcm-3 of the surrounding sedimentary rocks. If the intrusive body is modelled to where the density contrast is only 0.1 gcm -3 , the intrusive body extends down to 15 km, i.e. into the basement (Locke and Cassidy, 1997). A c.5 km diamet er intrusive region beneath Mt. Taranaki extendi ng to 10 km depth is also inferred by 3-D models of P- wave velocity, P- to S-wave velocity rati os, and P-wave quality factors. However, no magma storage was imaged within the upper 16 km, probably due to the 5 km spatial resolution of the models (Sherburn et al., 2006). Geophysical studies (2001-200 2 ) using 75 three-component, broad-band seismographs revealed a shallow crustal brittle-ductile tr ansition zone (BDTZ) at 10 km depth beneath Mt. Taranaki with an increase in dept h to 25 and 35 km to the west and east, Chapter 1 Introduction 12 respectively (Sherburn and White, 2005). The shallow BDTZ caused by magmatic intrusions is consistent with observed high heat flows (up to 73 mWm -2 ; Allis et al., 1995). The temperature at the top of the BDTZ is calculated at about 300 °C if the onset of plastic flow occurs in a quartz-do minated crust (Sherburn and White, 2005). 1.4.3 Taranaki Volcanic Lineament Quaternary volcanism in the west of the No rth Island is expressed by the Taranaki Volcanic Lineament (Fig. 1-2) a successi on of four NNW- SSE aligned individual andesite volcanoes consisting of the Sugar Lo af Islands (1.75 Ma), Kaitake (0.75 Ma), Pouakai (0.5 Ma) and Mt. Taranaki/Egmont (c.0.15 Ma) where vol canism has migrated to the SSE (Neall, 1979; Neall et al., 1986). The Sugar Loaf Islands are remnants of a former volcano where the pinnacles, such as Paritutu, repres ent volcanic dykes and plugs (Grant-Taylor, 1964). This former vol cano is offset by c.12 km to the NNW of Kaitake. Kaitake and Pouakai are andesite volcanoes at different stages of erosion, although Pouakai appears considerably larg er – although not approaching the basal diameter of the present Mt. Taranaki. The locus of volcanism for at least the last 150,000 yrs has been Mt. Taranaki. The Taranaki peninsula with its near semi-c ircular coastline is made up of overlapping fans of volcaniclastic sediments, including debris-avalanche, debris-flow, pyroclastic- flow, and fluvial deposits which form the Egmont and Pouakai ri ng plains (lowland below c.300 m; Gibson and Morgan, 1927). Most of the western and southern parts of the peninsula are characterised by deposits representing several collapse episodes of Mt. Taranaki and subsequent re worked material (Neall, 1979; Newnham and Alloway, 2004; Alloway et al., 2005; Zernack et al., 2005). 1.4.4 Mount Taranaki/Egmont Mount Taranaki/Egmont (2518 m), henceforth called Mt. Taranaki, is volumetrically the largest andesite volcano in New Zeala nd, although it is lower than Mt. Ruapehu (2797 m). The Taranaki edifice is a near ly perfectly shaped stratovolcano whose symmetry is only slightly disfigured by Fant hams Peak, a 1967 m high parasitic cone to Chapter 1 Introduction 13 the south (Fig. 1-3). The volcanic history of Mt. Taranaki began more than 150,000 years ago with the earliest know n deposits lying directly above an uplifted marine-cut surface (Alloway et al., 1995, 2005). The volca no’s history has been characterised by several cone-building and collapse episode s with the current edifice having been predominantly constructed over the last c.10 ka (Neall et al., 1986). Grant-Taylor (1964) described Mt. Taranaki as a somma volcano w ith the symmetrical upper cone (i.e. above 1100 m) grown 400 m SW off-centre from the lowe r part of the edifice. His observation is based on erosion features and slope differences with a pronounced “plateau” on the eastern side of the edifice marking the pre- collapse structure. Four lava domes are recognised on the lowe r flanks of Mt. Taranaki: the two Beehive domes on the south flank between 800-900 m ar e in alignment with Fanthams Peak on a NNE trend, whereas the other two domes, Th e Dome and Skinner Hill, are located on the northern flank at elev ations of 900 m and c.1100 m, respectively (Fig. 1-3). Although the latter two domes are not aligned with the summit, they are also considered to be located on radial faults extending from the summit of Mt. Ta ranaki (Neall, 1971). The eruptive activity over the last c.18 ka wa s characterised by moderate to large sized sub-Plinian eruptions on c.330 yr intervals, in terspersed with more frequent effusive, dome-forming and destructive events and lava flows (Alloway et al., 1995). A brief summary of the succession of volcanic events in the last c.10,000 yrs follows based on Neall, 1979 and Neall et al. (1986): - 12,000 - 7 , 0 0 0 yrs B.P.: major period of dome-producing and collapse episodes with BAF deposition on the western flank (e.g. Ok ahu Gorge, Pyramid and Waiweranui Streams) and the eastern flank (e.g. Ma nganui Gorge); immediately following this period was the extrusion of lava flows forming the core of the upper present day cone - c.7000 yrs B.P.: sector collapse of the s outhern flank forming the Opua Formation comprising debris-avalanche, de bris-flow and lahar deposits - subsequent re-building phase of the cone by lava flow extrusion, most of them probably younger than 3500 yrs - interspersed sub-Plinian eruptions with the Korito Tephra (4150 yrs B.P.) and the Inglewood Tephra (3690 yrs B. P.) being the most distin ct eruptions (Alloway et al., 1995) Chapter 1 Introduction 14 Figure 1-3: Orthophotograph of Mt. Taranaki including the lower northwestern flanks. Inset shows upper cone area including Fanthams Peak. Major morphologi cal features are indicated by letters: a-summit crater of Mt. Taranaki, b-Fanthams Peak, c-the B eehives (two lava domes), d-scarp of the Opua amphitheatre, e-Big Pyramid, f-The Dome, g-Skinner Hill (probably a buried dome structure), h-Pyramid Stream, i-Maero Stream, j-Waiweran ui Stream, k-Hangatahua River, l-Egmont National Park boundary (forest/pasture border, here 390 m), m-Kapuni Gorge (marks the eastern border of the amphitheatre), n- Sharks Tooth (second highest peak, 2510 m), o-Fanthams Peak (comprising multiple vents), p-remnant summit dome, q-Turtle, r-Bobs Ridge (western border of the Opua am phitheatre), s-NW flank and main path for BAFs. Chapter 1 Introduction 15 - 3300 yrs B.P.: final construction of the parasitic cone (Fanthams Peak) with a major and distinct eruption c.2900 yrs ago - extrusion of lava domes from subsidia ry vents on the lower flanks (The Dome, Skinner Hill and the Beehives) - 2000 yrs B.P.: dome-producing and collapse episode with BAF deposition on the eastern flanks - climactic construction phase of the u ppermost present day cone by lava flows - 1400 yrs B.P.: sub-Plinian Kaupokonui er uption with widespread pumice deposition throughout the eastern sector of the edifice - 600 yrs B.P.: Maero Eruptive Period ch aracterised by dome-forming and collapse episodes and one sub-Plinian eruption in AD 1655 (Burrell Lap illi eruption). The “last” known event at Mt. Taranaki, the Ta hurangi eruption, was considered to be in AD 1755 (Druce, 1966). 1.4.5 Petrology of Mt. Taranaki Rocks The volcanic rocks of the Taranaki Volcanic Lineament were first described by Hutton (1889) as being hornblende-, augite-, and olivine andesites. Although petrographical, and later geochemical studies were subseque ntly made (Marshall, 1907; Clarke, 1912; Gibson and Morgan, 1927), it wa s not until Gow (1968) that the first classification scheme for Mt. Taranaki rocks, based on fe rromagnesian mineral phases, was made. He recognised five rock types: 1) augite andesite , 2) augite – hornble nde andesite (augite > hornblende), 3) augite – hornblende andesi te (augite ~ hornblende), 4) augite – hornblende andesite (augite < ho rnblende), and 5) augite – olivine andesite (up to 7% modal olivine). This rock classification has been applied since then and was only modified by the addition of other mineral phases, such as plagioclase and biotite, where further distinction was needed (e.g. Neall et al., 1986, Stewart et al., 1996, Price et al., 1999). Common phenocrysts in Mt. Taranaki rocks ar e plagioclase, clinopyroxene, hornblende, Fe-Ti oxides, and minor olivine with a dditional microphenocryst and groundmass phases of orthopyroxene and biotite (Gow, 1968; Neall et al., 1986). Rare accessories are apatite and zircon (Neall et al., 1986). Co mplex mineral textures such as zoning and Chapter 1 Introduction 16 resorption are described in more detail by Gow (1968), Neall et al. (1986) and Stewart et al. (1996). Igneous and sedimentary incl usions within Mt . Taranaki rocks have long been recognised and recently Gründer (2006) classified them into: 1) sedimentary rocks, 2) garnet-gneiss – granodiorite – granite suite , 3) mafic hornfels, 4) banded amphibolitic gneiss, 5) gabbros, and 6) ultramafic rocks. The latter two types are referred to as being of co-magmatic origin whereas the first two are considered to be exotic inclusions. Types 3 and 4 cannot be clearly assigned to either source but their metamorphic character could suggest an affiliation to deformed MTZ basement (Gründer, 2006). The Mt. Taranaki rocks are classified as a high-K suite (Fig. 1-4) based on whole rock geochemistry, and comprise basalt – low-Si andesite – high-Si andesite (Price et al., 1992) according to the classification of Gill ( 1981). The same rock assemblage can also be classified as high-K basalts – basaltic a ndesites – andesites after Le Maitre et al. (1989). Further, a positive relationship with in the rock suite between potassium and silica contents (Fig. 1-4) and time has been recognised; generally, the younger the volcanic rocks the higher potassium and mean silica abundances (cf. Price et al., 1992; Stewart et al., 1996). An increase of alkalis and silica within a rock suite with no iron enrichment is also referred to as a calc-alka line trend. All Mt. Tara naki rocks are highly evolved based on Mg#, which is in general < 53, and low whole rock Cr and Ni contents (Price et al., 1992). Figure 1-4: Classification of Mt. Taranaki and Mt. Ruapehu rocks after Gill (198 1). Modified after Price et al. (1999). Chapter 1 Introduction 17 Trace element abundances are characterised by relatively high proportions of large ion lithophile elements (LILE; e.g. Rb, Cs, Sr, Ba ) and light rare earth elements (LREE) with deficiencies in high field strength elements (HFSE; Ta, Nb, Zr; cf. Price et al., 1992, 1999, 2005). These, as well as other tr ace element characteristics, such as depletion of Nb relative to La, and enrichment of Pb and Sr relati ve to Ce, are typical arc-signatures (Fig. 1-5). Isotope studies carried out by Price et al. (1992; 1999; 2005 ) show that Mt. Taranaki rocks display a tight range of 143 Nd/ 144 Nd and 87 Sr/ 86 Sr ratios. However, 87 Sr/ 86 Sr ratios along the Taranaki lineament centres indica te a weak positive relationship with decreasing age although no relationship with SiO 2 is present. Lead and oxygen isotope data indicate that significant assimilation of crustal material or material rich in 18 O was not involved in the genesis of Mt. Taranaki magmas. Figure 1-5: Trace element patterns of selected Mt. Taranaki rocks. Warwicks/Staircase refers to lava flow groups of the main cone of Mt. Taranaki; Fanthams describes lava flows of the satellite vent Fanthams Peak. Data compiled from Price et al. (1992 , 1999 ). Normalisation after Sun and McDonough (1989 ). The enrichment of LILE and depletion of HFSE in Mt. Taranaki magmas is attributed to low-degree partial melts of a depleted mantle as well as a significant component of slab- derived fluids (Price et al., 1999). Parental magmas were th erefore interpreted to be relatively undersaturated, hydr ous, high-Mg basalts. Stewart et al. (1996) introduced a model based on Foden and Green (1992) wh ich suggests progressive underplating of these basalts at the base of the crust which then evolved to high-Al basalts. The high-Al basalts eventually intersect the amphibole stability field and interact with upper mantle/lower crust amphibolitic material, buffering the melt composition to basaltic Chapter 1 Introduction 18 andesite. These magmas ascend and erupt as dry, mainly plagioclase-, olivine-, pyroxene-bearing magmas. However, thes e magmas may re-enter the amphibole stability field at higher crustal levels to crystallise amphibole driving the melt composition through andesite and dacite. The incongruent melting of amphibole at depth and/or through dyke – wall rock interaction is probably the main source for K- enrichment in the melt phase (Stewart et al., 1996). In a comparison of andesite volcanism in th e Tongariro Volcanic Centre and Taranaki, Price et al. (2005) proposed th at andesite magmas are produced through the interaction of mantle-derived magmas and the lower crust. From this zone upward migrating magmas are crystal- and li thic-rich and their melt phase s of rhyolite and dacite composition. High level crustal storage and the presence of a dispersed, complex plumbing systems was thought to cause the observed geochemical and mineral textural variations within andesitic rock s through time (Price et al., 2005). Chapter 1 Introduction 19 1.5 References Abdurachman, E.K., Bourdier, J. - L. and Voight, B. 2000. Nuées ardentes of 22 November 1994 at Merapi volcano, Java, Indonesia. Journal of Volcanology and Geothermal Research 100, 345-3 61. Alidibirov, M. and Dingwell, D.B. 1996. Magma fragmentation by rapid decompression. Nature 380, 146- 148. Alidibirov, M. and Dingwell, D.B. 2000. Three fragmentation mechanisms for highly viscous magma under rapid decompression. Journal of Volcanology and Geothermal Research 100, 413- 421. Alidibirov, M., Dingwell, D.B., Stevenson, R.J., Hess, K.U., Webb, S.L. and Zinke, J. 1997. Physical properties of the 1980 Mount St. Helens cryptodome magma. Bulletin of Volcanology 59, 103-1 11. Allis, R.G., Armstrong, P.A. and Funnell, R.H. 1995. Implications of a high heat flow anomaly around New Plymouth, North Island, New Zealand. New Zealand Journal of Geology and Geophysics 38, 121- 130. Allis, R.G., Zhan, X., Evans, C. and Kroopnick, P. 1997. Groundwater flow beneath Mt Taranaki, New Zealand, and implications for oil and gas migration. New Zealand Journal of Geology and Geophysics 40, 137-1 49. Alloway, B., Neall, V.E. and Vucetich, C.G. 1995. Late Quaternary (post 28,000 year B.P.) tephrostratigraphy of northeast and central Taranaki, New Zealand. Journal of the Royal Society of New Zealand 25, 385-4 58. Alloway, B., McComb, P., Neall, V., Vucetich, C., Gibb, J., Sherburn, S. and Stirling, M. 2005. Stratigraphy, age, and correlation of voluminous debris-avalanche events from an ancestral Egmont Volcano: implications for coastal plain c onstruction and regional hazard assessment. Journal of the Royal Society of New Zealand 35, 229- 267. Anderson, S.W. and Fink, J.H. 1989. Hydrogen-isotope evidence for extrusion mechanisms of the Mount St Helens lava dome. Nature 341, 521 -52 3. Anderson, S.W., Fink, J.H. and Rose, W.I. 1995. Mount St. Helens and Santiaguito lava domes: the effect of short-term eruption rate on surface texture and degassing processes. Journal of Volcanology and Geothermal Research 69, 105 -11 6. Battaglia, M. 1993. On pyroclastic flow emplacement. Journal of Geophysical Research-Solid Earth 98, 2226 9-2 227 2. Blake, S. 1984. Volatile oversaturation during the evolution of silicic magma chambers as an eruption trigger. Journal of Geophysical Research-Solid Earth 89, 8237- 824 4. Blundy, J. and Cashman, K. 2001. Ascent-driven crystallisation of dacite magmas at Mount St Helens, 1980 -19 86. Contributions to Mineralogy and Petrology 140, 631-6 50. Chapter 1 Introduction 20 Blundy, J., Cashman, K. and Humphreys, M. 2006. Magma heating by decompression-driven crystallization beneath andesite volcanoes. Nature 443, 76-8 0. Boddington, T., Parkin, C.J. and Gubbins, D. 2004. Isolated deep earthquakes beneath the North Island of New Zealand. Geophysical Journal International 158, 972 -98 2. Boudon, G. and Lajoie, J. 1989. The 1902 Peléean deposits in the Fort Cemetery of St. Pierre, Martinique: a model for the accumula tion of turbulent nuées ardentes. Journal of Volcanology and Geothermal Research 38, 113- 129. Boudon, G., Camus, G., Gourgaud, A. and Lajoie, J. 1993. The 1984 nuée-ardente deposits of Merapi volcano, Central Java, Indonesia: stratigraphy, te xtural characteristics, a nd transport mechanisms. Bulletin of Volcanology 55, 327-3 42. Bourdier, J.L., Pratomo, I., Thouret, J.C., Boudon, G. and Vincent, P.M. 1997. Observations, stratigraphy and eruptive processes of the 1 990 eruption of Kelut volcano, Indonesia. Journal of Volcanology and Geothermal Research 79, 181- 203. Branney, M.J. and Kokelaar, P. 1992. A reappraisal of ignimbrite emplacement: progressive aggradation and changes from particulate to non-particulate flow during emplacement of high-grade ignimbrite. Bulletin of Volcanology 54, 504-5 20. Branney, M.J. and Kokelaar, P. 2002. Pyroclastic density currents and the sedimentation of ignimbrites. Geological Society of London Memoir 27. Briggs, R.M. 1983. Distribution, form, and structural cont rol of the Alexandra Vo lcanic Group, North Island, New Zealand. New Zealand Journal of Geology and Geophysics 26, 47-55. Briggs, R.M., Itaya, T., Lowe, D.J. and Keane, A.J. 1989. Ages of the Plio cene-Pleistocene Alexandra and Ngatutura Volcanics, western North island, New Zealand, and some geological implications. New Zealand Journal of Geology and Geophysics 32, 417-42 7. Camus, G., Gourgaud, A., Mossand - Berthommier, P.C. and Vincent, P.M. 2000. Merapi (Central Java, Indonesia): an outline of the structural and magmatological evolution, with a special emphasis to the major pyroclastic events. Journal of Volcanology and Geothermal Research 100, 139- 163. Carey, S. and Sigurdsson, H. 1985. The May 18, 1980 eruption of Mount St. Helens; II - modeling of dynamics of the Plinian phase. Journal of Geophysical Research-Solid Earth 90, 294 8-2 95 8. Carn, S.A., Watts, R.B., Thompson, G. and Norton, G.E. 2004. Anatomy of a lava dome collapse: the 20 March 2000 event at Soufriére Hills Volcano, Montserrat. Journal of Volcanology and Geothermal Research 131, 241 -26 4. Cas, R.A.F. and Wright, J.V. 1991. Subaqueous pyroclastic flows and ignimbrites: an assessment. Bulletin of Volcanology 53, 357-3 80. Cashman, K.V. and Hoblitt, R.P. 2004. Magmatic precursors to the 18 May 1980 eruption of Mount St. Helens, USA. Geology 32, 141 -14 4. Chapter 1 Introduction 21 Cashman, K.V. and McConnell, S.M. 2005. Multiple levels of magma storage during the 1980 summer eruptions of Mount St. Helens, WA. Bulletin of Volcanology 68, 57-75. Childs, C.W., Wells, N. and Downes, C.J. 1986. Kokowai Springs, Mount Egmont, New Zealand: chemistry and mineralogy of the ochre (ferrih ydrite) deposit and analysis of the waters. Journal of the Royal Society of New Zealand 16, 85-9 9. Choux, C.M. and Druitt, T.H. 2002. Analogue study of particle segregation in pyroclastic density currents, with implications for the emplacement mechanisms of large ignimbrites. Sedimentology 49, 907- 928. Clarke, E. de C. 1912. The geology of the New Plymouth Subdivision, Taranaki Division. New Zealand Geological Survey Bulletin 1 4. Cole, P.D., Calder, E.S., Druitt, T.H., Hoblitt, R., Robertson, R., Sparks, R.S.J. and Young, S.R. 1998. Pyroclastic flows generated by gravitational instability of the 1996-9 7 lava dome of Soufriere Hills Volcano, Montserrat. Geophysical Research Letters 258, 3425- 342 8. Cole, P.D., Calder, E.S., Sparks, R.S.J., Clarke, A.B., Druitt, T.H., Young, S.R., Herd, R.A., Harford, C.L. and Norton, G.E. 2002. Deposits from dome-collapse and fountain-collapse pyroclastic flows at Soufrière Hills Volcano, Montserrat. Geological Society, London, Memoirs 21, 231- 262. Couch, S., Sparks, R.S.J. and Carroll, M.R. 2003. The kinetics of degassing-induced crystallization at Soufrière Hills Volcano, Montserrat. Journal of Petrology 44, 1477- 150 2. Cowan, J. 1930. Fairy folk tales of the Maori. Whitcombe and Toombs. Auckland. Cronin, S.J. and Neall, V.E. 1997. A late Quaternary stratigraphic framework for the northeastern Ruapehu and eastern Tongariro ring plains, New Zealand. New Zealand Journal of Geology and Geophysics 40, 185-1 97. Cronin, S.J., Stewart, R.B., Neall, V.E., Platz, T. and Gaylord, D. 2003. The AD1040 to present Maero eruptive period of Egmont Volcan o, Taranaki, New Zealand. Abstract. Geological Society of New Zealand Miscellaneous Publication 116A, 43. Dade, W.B. and Huppert, H.E. 1996. Emplacement of the Taupo ignimbrite by a dilute turbulent flow. Nature 381, 509 -51 2. Devine, J.D., Rutherford, M.J., Norton, G.E. and Young, S.R. 2003. Magma storage region processes inferred from geochemistry of Fe-Ti oxides in andesitic magma, Soufrière Hills Volcano, Montserrat, W.I. Journal of Petrology 44, 1375-1 400. Dissen, R. van and Yeats, R.S. 1991. Hope fault, Jordan thrust and uplift of the Seaward Kaikoura Range, New Zealand. Geology 19, 393-3 96. Chapter 1 Introduction 22 Dobran, F. 1992. Nonequilibrium flow in volcanic conduits and application to the eruptions of Mt. St. Helens on May 18, 1980 , and Vesuvius in AD 79. Journal of Volcanology and Geothermal Research 49, 285- 311. Donoghue, S.L., Neall, V.E. and Palmer, A.S. 1995. Stratigraphy and chronology of late Quaternary andesitic tephra deposits, Tongariro Volcanic Center, New Zealand. Journal of the Royal Society of New Zealand 25, 115-2 06. Druce, A.P. 1966. Tree-ring dating of recent volcanic ash and lapilli, Mt Egmont. New Zealand Journal of Botany 4, 3-41. Fink, J.H., Malin, M.C. and Anderson, S.W. 1990. Intrusive and extrusive growth of the Mount St Helens lava dome. Nature 348, 435-4 37. Fink, J.H., Anderson, S.W. and Manley, C.R. 1992. Textural constraints on effusive silicic volcanism - beyond the permeable foam model. Journal of Geophysical Research-Solid Earth 97, 907 3-9 083. Fisher, R.V., Orsi, G., Ort, M. and Heiken, G. 1993. Mobility of a large-volume pyroclastic flow- emplacement of the Campanian ignimbrite, Italy. Journal of Volcanology and Geothermal Research 56, 205- 220. Foden, J.D. and Green, D.H. 1992. Possible role of amphibole in the origin of andesite: some experimental and natural evidence. Contributions to Mineralogy and Petrology 109, 479- 493. Freundt, A. 1998. The formation of high-grade ignimbrites, I: experiments on high- and low- concentration transport systems containing sticky particles. Bulletin of Volcanology 59, 414- 435. Freundt, A. 2003. Entrance of hot pyroclastic flows into the sea: experimental observations. Bulletin of Volcanology 65, 144-1 64. Freundt, A. and Bursik, M. 1998. Pyroclastic flow transport mechanisms. In: Freundt, A. and Rosi, M. (eds.) From magma to tephra: modelling physical processes of explosive volcanism. Elsevier. Amsterdam. pp. 151-2 22. Fujii, T. and Nakada, S. 1999. The 15 September 1991 pyroclas tic flows at Unzen Volcano (Japan): a flow model for associated ash-cloud surges. Journal of Volcanology and Geothermal Research 89, 159- 172. Funnell, R., Chapman, D., Allis, R. and Armstrong, P. 1996. Thermal state of the Taranaki Basin, New Zealand. Journal of Geophysical Research-Solid Earth 101, 25197 -25 215. Gardner, J.E., Thomas, R.M.E., Jaupart, C. and Tait, S. 1996. Fragmentation of magma during Plinian volcanic eruptions. Bulletin of Volcanology 58, 144- 162. Gibson, W. and Morgan, P.G. 1927. The geology of the Egmont Subdivision, Taranaki. New Zealand Geological Survey Bulletin 2 9. Gill, J.B. 1981. Orogenic andesites and plate tectonics. Springer Verlag. Berlin. Chapter 1 Introduction 23 Giordano, G. and Dobran, F. 1994. Computer simulations of the Tuscolano-Artemisio’s second pyroclastic flow unit (Alban Hills, Latium, Italy). Journal of Volcanology and Geothermal Research 61, 69-9 4. Gow, A.J. 1968. Petrographic and petrochemical studies of Mt. Egmont andesites. New Zealand Journal of Geology and Geophysics 11, 166-1 90. Grant - Taylor, T.L. 1964. Volcanic history of western Taranaki. New Zealand Journal of Geology and Geophysics 7, 78-8 6. Gründer, K. 2006. Petrography and mineral chemistry of a xenolith suite from Mt. Taranaki, New Zealand: insights into the sub-volcanic lithospher e of an arc volcano. Unpublished Diploma thesis. Johannes Gutenberg Universität, Ma inz, Bundesrepublik Deutschland. Grunewald, U., Sparks, R.S.J., Kearns, S. and Komorowski, J.C. 2000. Friction marks on blocks from pyroclastic flows at the Soufriere Hills volcano, Montserrat: implications for flow mechanisms. Geology 28, 827 -83 0. Harford, C.L., Sparks, R.S.J. and Fallick, A.E. 2003. Degassing at the Soufrière Hills Volcano, Montserrat, recorded in matrix glass compositions. Journal of Petrology 44, 1503- 152 3. Higgins, M.D. and Roberge, J. 2003. Crystal size distribution of plagioclase and amphibole from Soufrière Hills volcano, Montserrat: evidence for dy namic crystallization-textural coarsening cycles. Journal of Petrology 44, 1401- 141 1. Hobden, B.J., Houghton, B.F. and Nairn, I.A. 2002. Growth of a young, frequently active composite cone: Ngauruhoe volcano, New Zealand. Bulletin of Volcanology 64, 392-4 09. Hull, A.G. 1994. Past earthquake timing and magnitude along the Inglewood fault, Taranaki, New Zealand. Bulletin of the New Zealand National Society for Earthquake Engineering 27, 155-1 62. Hull, A.G. and Dellow, G. 1993. Earthquake hazards in the Tarana ki region. Institute of Geological and Nuclear Sciences Client Report 1993/0 3. Hutton, F.W. 1889. The eruptive rocks of New Zealand. Journal and Proceedings of the Royal Society of New South Wales 23, 102- 56. King, P.R. and Thrasher, G.P. 1996. Cretaceous-Cenozoic geology and petroleum systems of the Taranaki Basin, New Zealand. Institute of Geological and Nuclear Sciences Monograph 13. Klug, C. and Cashman, K.V. 1994. Vesiculation of May 18, 1980 , Mount St. Helens magma. Geology 22, 468- 472. Lacroix, A. 1904. La Montagne Pelée et ses éruptions. Masson et Cie. Paris. Le Maitre, R.W., Bateman, P., Dudek, A., Keller, J., Lameyre, J., Le Bas, M.J., Sabine, P.A., Schmid, R., Sorensen, H., St reckeisen, A., Woolley, A.R. and Zanettin, B. 1989. A classification of igneous rocks and glossary of terms. Blackwell Scientific Publications. Oxford. Chapter 1 Introduction 24 Lipman, P.W. and Mullineaux, D.R. (eds.) 1981. The 1980 eruptions of Mount St. Helens, Washington. U.S. Geological Sur vey Professional Paper 1250. Locke, C.A. and Cassidy, J. 1997. Egmont Volcano, New Zealand: three-dimensional structure and its implications for evolution. Journal of Volcanology and Geothermal Research 76, 149-16 1. Locke, C.A., Cassidy, J. and Macdonald, A. 1993. Three-dimensional structure of relict stratovolcanoes in Taranaki, New Zealand: evidence from gravity data. Journal of Volcanology and Geothermal Research 59, 121-1 30. Locke, C.A., Cassidy, J. and Macdonald, A. 1994. Constraints on the evolution of the Taranaki volcanoes, New Zealand, based on aeromagnetic data. Bulletin of Volcanology 56, 552-56 0. Lube, G., Huppert, H.E., Sparks, R.S.J. and Hallworth, M.A. 2004. Axisymmetric collapses of granular columns. Journal of Fluid Mechanics 508, 175- 199. Lube, G., Cronin, S.J., Platz, T., Freundt, A., Procter, J.N., Henderson, C. and Sheridan, M.F. 2007. Flow and deposition of pyroclastic granular flows: a type example from the 1975 Ngauruhoe eruption, New Zealand. Journal of Volcanology and Geothermal Research 161, 165 -18 6. Lyman, A.W., Koenig, E. and Fink, J.H. 2004. Predicting yield strengths and effusion rates of lava domes from morphology and underlying topography. Journal of Volcanology and Geothermal Research 129, 125- 138. Marshall, P. 1907. Geology of centre and north of North Island. Transactions and Proceedings of the Royal Society of New Zealand 40, 79-9 8. Massol, H., Jaupart, C. and Pepper, D.W. 2001. Ascent and decompression of viscous vesicular magma in a volcanic conduit. Journal of Geophysical Research-Solid Earth 106, 1622 3-1 624 0. McClelland, E.A. and Druitt, T.H. 1989. Paleomagnetic estimation of emplacement temperatures of pyroclastic deposits on Santorini, Greece. Bulletin of Volcanology 51, 16-2 7. Melnik, O. and Sparks, R.S.J. 2002. Dynamics of magma ascent and lava extrusion at Soufrière Hills Volcano, Montserrat. In: Druitt, T.H. and Kokelaar, B.P. (eds.) The eruption of Soufrière Hills Volcano, Montserrat, from 1995 to 1999. Geological Society, London, Memoirs 21, 153 -1 71. Mets, C. de, Gordon, R.G., Argus, D.F. and Stein, S. 1994. Effect of recent revisions to the geomagnetic reversal time scale on estimates of current plate motions. Geophysical Research Letters 21, 2191 -21 94. Minister, J.B. and Jordan, T.H. 1978. Present day plate motions. Journal of Geophysical Research- Solid Earth 83, 5331- 535 4. Mongillo, M.A. and Clelland, L. 1984. Concise listing of information on the thermal areas and thermal springs of New Zealand. Department of Scientific and Industrial Research. Wellington. DSI R geothermal report 9. Chapter 1 Introduction 25 Moore, J.G. and Melson, W.G. 1969. Nuées ardentes of the 1968 eruption of Mayon volcano, Philippines. Bulletin of Volcanology 33, 600- 620. Moore, J.G., Lipman, P.W., Swanson, D.A. and Alpha, T.R. 1981. Growth of lava domes in the crater, June 1980- January 1981. In: Lipman, P.W. and Mullineaux, D.R. (eds.) The 1980 eruptions of Mount St. Helens, Washington. U.S. Geological Survey Professional Paper 1250, 541 -54 8. Mortimer, N., Tulloch, A.J. and Ireland, T.R. 1997. Basement geology of Taranaki and Wanganui Basins, New Zealand. New Zealand Journal of Geology and Geophysics 40, 223-2 36. Mueller, S., Melnik, O., Spieler, O., Scheu, B. and Dingwell, D.B. 2005. Permeability and degassing of dome lavas undergoing rapid decompression: an experimental determination. Bulletin of Volcanology 67, 526-5 38. Nairn, I.A. and Self, S. 1978. Explosive eruptions and pyrocl astic avalanches from Ngauruhoe in February 1975. Journal of Volcanology and Geothermal Research 3, 39-60. Nakada, S., Miyake, Y., Sato, H., Oshima, O. and Fujinawa, A. 1995. Endogenous growth of dacite dome at Unzen Volcano (Japan), 1993 -1 994. Geology 23, 157-1 60. Nakada, S., Shimizu, H. and Ohta, K. 1999. Overview of the 1990-199 5 eruption at Unzen Volcano. Journal of Volcanology and Geothermal Research 89, 1-22. Navon, O., Chekhmir, A. and Lyakhovsky, V. 1998. Bubble growth in highly viscous melts: theory, experiments, and autoexplosivity of dome lavas. Earth and Planetary Science Letters 160, 763-7 76. Neall, V.E. 1971. Volcanic domes and lineations in Egmont National Park. New Zealand Journal of Geology and Geophysics 14, 71-81. Neall, V.E. 1979. Sheets P19, P20, and P21. New Plymouth, Egmont and Manaia. 1st ed. Geological map of New Zealand 1:50,000. 3 maps and notes. New Zealand Department of Scientific and Industrial Research. Wellington. Neall, V.E., Stewart, R.B. and Smith, I.E.M. 1986. History and petrology of the Taranaki volcanoes. In: Smith, I.E.M. (ed.) Late Cenozoic volcanism. Royal Society of New Zealand Bulletin 23, 251-2 63. Newnham, R. and Alloway, B. 2004. A terrestrial record of last interglacial climate preserved by voluminous debris avalanche inundation in Taranaki, New Zealand. Journal of Quaternary Science 19, 299- 314. Nicol, A., Stagpoole, V. and Maslen, G. 2004. Structure and petroleum potential of the Taranaki fault play. Abstract. 20 04 New Zealand Petroleum Conference Proceedings . Ministry of Economic Development, Wellington, New Zealand. Nodder, S.D. 1993. Neotectonics of the offshore Cape Egmont Fault Zone, Taranaki Basin, New Zealand. New Zealand Journal of Geology and Geophysics 36, 167-1 84. Pallister, J.S., Hoblitt, R.P. and Reyes, A.G. 1992. A basalt trigger for the 1991 eruptions of Pinatubo volcano? Nature 356, 426- 428. Chapter 1 Introduction 26 Papale, P. and Dobran, F. 1994. Magma flow along the volcanic conduit during the Plinian and pyroclastic flow phases of the May, 18, 198 0, Mount St. Helens eruption. Journal of Geophysical Research-Solid Earth 99, 4355 -43 73. Platz, T. 2001. Mapping and characterisation of the vol caniclastic Maero formation deposits on the northwestern sector of Egmont volcano (Mt. Taranaki), New Zealand. Unpublished Diploma mapping thesis. Ernst-Moritz-Arndt Universitä t, Greifswald, Bundesrepublik Deutschland. Power, J.A., Jolly, A.D., Nye, C.J. and Harbin, M.L. 2002. A conceptual model of the Mount Spurr magmatic system from seismic and geochemical observations of the 1992 Crater Peak eruption sequence. Bulletin of Volcanology 64, 206-21 8. Price, R.C., McCulloch, M.T., Smith, I.E.M. and Stewart, R.B. 1992. Pb-Nd-Sr isotopic compositions and trace element characteristics of young volcanic rocks from Egmont Volcano and comparisons with basalts and andesites from the Taupo Volcanic Zone, New Zealand. Geochimica et Cosmochimica Acta 56, 941-9 53. Price, R.C., Stewart, R.B., Woodhead, J.D. and Smith, I.E.M. 1999. Petrogenesis of high-K arc magmas: evidence from Egmont Vol cano, North Island, New Zealand. Journal of Petrology 40, 167- 1 97. Price, R.C. Gamble, J.A., Smith, I.E.M., Stewart, R.B., Eggins, S. and Wright, I.C. 2005. An integrated model for the temporal evolution of andesites and rhyolites and crustal development in New Zealand’s North Island. Journal of Volcanology and Geothermal Research 140, 1-2 4. Reyners, M., Eberhart-Phillips, D., Stuart, G. and Nishimura, Y. 2006. Imaging subduction from the trench to 300 km depth beneath the central North Island, New Zealand, with Vp and Vp/Vs. Geophysical Journal International 165, 565-5 83. Rodríguez-Elizarrarás, S., Siebe, C., Komorowski, J. - C., Espíndola, J.M. and Saucedo, R. 1991. Field observations of pristine block- and ash-flow deposits emplaced Ap ril 16-17, 1991 at Volcán de Colima, Mexico. Journal of Volcanology and Geothermal Research 48, 399-4 12. Rutherford, M.J. and Hill, P.M. 1993. Magma ascent rates from amphibole breakdown - an experimental study applied to the 1980- 198 6 Mount St Helens eruptions. Journal of Geophysical Research-Solid Earth 98, 1966 7-1 968 5. Rutherford, M.J. and Devine, J.D. 2003. Magmatic conditions and magma ascent as indicated by hornblende phase equilibria and reactions in the 1995-2 002 Soufriére Hills magma. Journal of Petrology 44, 1433 -14 53. Sato, H., Nakada, S., Fujii, T., Nakamura, M. and Suzuki-Kamata, K. 1999. Groundmass pargasite in the 1991-19 95 dacite of Unzen volcano: phase stability experiments and volcanological implications. Journal of Volcanology and Geothermal Research 89, 197- 212. Chapter 1 Introduction 27 Saucedo, R., Macias, J.L., Bursik, M.I., Mora, J.C., Gavilanes, J.C. and Cortes, A. 2002. Emplacement of pyroclastic flows during the 1998-19 99 eruption of Volcán de Colima, México. Journal of Volcanology and Geothermal Research 117, 129 -15 3. Scandone, R. and Malone, S.D. 1985. Magma supply, magma discharge and readjustment of the feeding system of Mount St. Helens during 1980. Journal of Volcanology and Geothermal Research 23, 239 - 2 62. Scandone, R., Cashman, K.V. and Malone, S.D. 2007. Magma supply, magma ascent and the style of volcanic eruptions. Earth and Planetary Science Letters 253, 513-5 29. Scanlan, A.B. 1961. Egmont - the story of a mount ain. A.H. & A.W. Reed. Wellington. Schmincke, H.-U. 1974. Volcanological aspects of peralkaline silicic welded ash-flow tuffs. Bulletin of Volcanology 38, 594-6 36. Scott, A.C. and Glasspool, I.J. 2005. Charcoal reflectance as a proxy for the empl acement temperature of pyroclastic flow deposits. Geology 33, 589 -59 2. Sherburn, S. and White, R.S. 2005. Crustal seismicity in Taranaki, New Zealand using accurate hypocentres from a dense network. Geophysical Journal International 162, 494-5 06. Sherburn, S., White, R.S. and Chadwick, M. 2006. Three-dimensional tomographic imaging of the Taranaki volcanoes, New Zealand. Geophysical Journal International 166, 957 -96 9. Simmons, J., Elsworth, D. and Voight, B. 2004. Instability of exogenous lava lobes during intense rainfall. Bulletin of Volcanology 66, 725- 734. Simmons, J., Elsworth, D. and Voight, B. 2005. Classification and idealized limit-equilibrium analyses of dome collapses at Soufiere Hills volcano, M ontserrat, during growth of the first lava dome: November 1995- March 1998. Journal of Volcanology and Geothermal Research 139, 241 -25 8. Smith, R.L. and Bailey, R.A. 1966. The Bandelier Tuff: a study of ash-flow eruption cycles from zoned magma chambers. Bulletin of Volcanology 29, 83-10 3. Sparks, R.S.J. 1976. Grain size variations in ignimbrites and implications for the transport of pyroclastic flows. Sedimentology 23, 147 -1 88. Sparks, R.S.J. 1997. Causes and consequences of pressurisation in lava dome eruptions. Earth and Planetary Science Letters 150, 177-1 89. Sparks, R.S.J. and Young, S.R. 2002. The eruption of Soufrière H ills Volcano, Montserrat (1995 -199 9). In: Druitt, T.H. and Kokelaar, B.P. (eds.) The eruption of Soufrière Hills Volcano, Montserrat, from 1995 to 1999. Geological Society, London, Memoirs 21, 45-6 9. Chapter 1 Introduction 28 Sparks, R.S.J., Young, S.R., Barclay, J., Calder, E. S., Cole, P., Darroux, B., Davies, M.A., Druitt, T.H., Harford, C., Herd, R., Ja mes, M., Lejeune, A.M., Loughlin , S., Norton, G. , Skerrit, G., Stasiuk, M.V., Stevens, N. S., Toothill, J., Wadge, G. and Watts, R. 1998. Magma production and growth of the lava dome of the Soufriere Hills Volcano, Montserrat, West Indies: November 1995 to December 1997. Geophysical Research Letters 25, 342 1-34 24. Sparks, R.S.J., Murphy, M.D., Lejeune, A.M., Watts, R.B., Barclay, J. and Young, S.R. 2000. Control on the emplacement of the andesite lava dome of the Soufriere Hills volcano, Montserrat by degassing-induced crystallization. Terra Nova 12, 14-2 0. Stagpoole, V. 1999. The Awhitu volcanic complex, an offshore Pliocene volcano in the northern Taranaki Basin, New Zealand. New Zealand Journal of Geology and Geophysics 42, 327-3 34. Stern, T.A. 1987. Asymmetric back-arc sp reading, heat flux and structur e beneath the central volcanic region of New Zealand. Earth and Planetary Science Letters 85, 265-2 76. Stewart, R.B., Price, R.C. and Smith, I.E.M. 1996. Evolution of high-K arc magma, Egmont volcano, Taranaki, New Zealand: evidence from mineral chemistry. Journal of Volcanology and Geothermal Research 74, 275-2 95. Sun, S. - S. and McDonough, W.F. 1989. Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes. Geological Society, London, Special Publications 42, 313-3 45. Swanson, D.A. and Holcomb, R.T. 1990. Regularities in growth of the Mount St. Helens dacite dome, 1980 -19 86. In: Fink, J.H. (ed.) Lava flows and domes: emplacement mechanisms and hazard implications. IAVCEI Proceeding in Volcanology 2, 3-24. Swanson, D.A., Casadevall, T.J., Dzur isin, D., Malone, S.D., Newhall, C.G. and Weaver, C.S. 1983. Predicting eruptions at Mount St. Hele ns, June 1980 thro ugh December 1982. S cience 221, 1369 - 1 376. The Taranaki Herald 1871. 25 March. Topping, W.W. 1972. Burrell Lapilli eruptives, Mount Egmont, New Zealand. New Zealand Journal of Geology and Geophysics 15, 476-4 90. Ui, T., Matsuwo, N., Sumita, M. and Fujinawa, A. 1999. Generation of block and ash flows during the 1990-19 95 eruption of Unzen Volcano, Japan. Journal of Volcanology and Geothermal Research 89, 123- 137. Voight, B. and Elsworth, D. 2000. Instability and collapse of hazardous gas-pressurized lava domes. Geophysical Research Letters 27, 1-4. Chapter 1 Introduction 29 Voight, B., Sparks, R.S.J., Miller, A.D., Stewart, R. C., Hoblitt, R.P., Clarke, A., Ewart, J., Aspinall, W.P., Baptie, B., Calder, E.S., Cole, P., Druitt, T.H., Hartford, C., Herd, R.A., Jackson, P., Lejeune, A.M., Lockhart, A.B., Loughlin, S.C., Luckett, R., Lynch, L., Norton, G.E., Robertson, R., Watson, I.M., Watts, R. and Young, S.R. 1999. Magma flow instability and cyclic activity at Soufriere Hills Volcano, Montserrat, British West Indies. S cience 283, 1138 -114 2. Voight, B., Linde, A.T., Sacks, I.S., Mattioli, G.S ., Sparks, R.S.J., Elsworth, D., Hidayat, D., Malin, P.E., Shalev, E., Widiwijayanti, C., Young, S.R ., Bass, V., Clarke, A., Dunkley, P., Johnston, W., McWhorter, N., Neuberg, J. and Williams, P. 2006. Unprecedented pressure increase in deep magma reservoir triggered by lava-dome collapse. Geophysical Research Letters 33, L0331 2. Walker, G.P.L., Wilson, C.J.N. and Froggatt, P.C. 1980. Fines-depleted ignimbrite in New Zealand: the product of a turbulent pyroclastic flow. Geology 8, 245- 249. Wallace, L.M., Beavan, J., McCaffrey, R. and Darby, D. 2004. Subduction zone coupling and tectonic block rotations in the North Island, New Zealand . Journal of Geophysical Research-Solid Earth 109, B1240 6. Wilson, C.J.N. and Walker, G.P.L. 1982. Ignimbrite depositional faci es: the anatomy of a pyroclastic flow. Journal of the Geological Society 139, 581 -59 2. Wilson, C.J.N., Houghton, B.F., McWillia ms, M.O., Lanphere, M.A., Weaver, S.D. and Briggs, R.M. 1995. Volcanic and structural evolution of Taupo Volcanic Zone, New Zealand: a review. Journal of Volcanology and Geothermal Research 68, 1-28. Wright, J.V. and Walker, G.P.L. 1981. Eruption, transport and deposition of ignimbrite - a case study from Mexico. Journal of Volcanology and Geothermal Research 9, 111-1 31. Zanella, E., Gurioli, L., Pareschi, M.T. and Lanza, R. 2007. Influences of urban fabric on pyroclastic density currents at Pompeii (Italy): 2. temperature of the deposits and hazard implications. Journal of Geophysical Research-Solid Earth 112, B05 214. Zernack, A.V., Procter, J.N. and Cronin, S.J. 2005. Cyclic growth and destruction of stratovolcanoes. Abstract. Geological Society of New Zealand Miscellaneous Publication 119A, 99-100. Chapter 2 Methodology 31 Chapter 2 Methodology 2 Chapter 2 In this chapter the overall strategy for the research is outlined and all methods that were used in this study and applied by the author are described. 2.1 Brief Outline In this study the following activities were undertaken toward fulfilling the aims outlined in Chapter 1: - Development of an overall stratigraphy of eruptive units during the Maero Eruptive Period, including mapping deposit distributio n, systematic sampling of bulk units, primary lithologies (clasts), and organic materi al between and within units that could be used for radiocarbon dating. - Concentration of studies on the depositi on record of two eruption episodes with contrasting eruptive styles from the Maer o Eruptive Period: the explosive Burrell episode and the effusive Pyramid lava dome-forming eruption. Deposits resulting from these episodes and the surrounding deposits were mapped in the field and sampled systematically. In order to complete the task of stratigraphi c correlation of separate eruptive units in the Maero Period, glass chemistry (using electr on microprobe) was analysed from ash samples of fall units, surge deposits and Bl ock-and-Ash Flow (BAF) matrix materials Chapter 2 Methodology 32 (Platz, 2001; S.J. Cronin, unpubl. data). In addition to mineralogical and geochemical analyses (bulk rock and mineral chemistry) , physical properties (density, porosity, permeability) of erupted rocks were investigated. Further, water contents in melt inclusions were analysed to examine the role of volatiles in the explosivity of past eruptions. This was supplemented by other means of estimating water contents of glasses (through microprobe and differential scanning calorimetry and thermal gravimetric analysis). 2.2 Field Studies Field studies carried out in the summers of 2004 and 2005 were focussed primarily on the fan of deposits making up the NW volcano flanks, expanding later to other areas, including the summit lava dome, and soil/tephra sections ar ound the upper flanks of the remaining volcano flanks. Sampling and descript ion of the main BAF units were carried out in order to later characterise their mineralogical and geochemical composition. Some of the samples from the NW sector of Mt. Taranaki used for mineralogical and geochemical studies were described by Platz (2001). All deposits were described and sampled following standard procedures, i.e. description of main characteristics of deposits at each location including thickn ess, clast componentry, maximum clast diameters for pumice and lithics, average diameters of about 10 clasts (pumice and lithics), bedding, grading, and sorting. Bulk samples and/or individual clasts were collected if necessary for grain size analysis. Similar techni ques were applied to pumice flow deposits. In addition, 109 pumice clasts of various textures and sizes were collected for porosity, permeability measurements and bulk rock chemical analysis; typically pumice clasts of fall deposits were too small and fragile for these methods. At some sites measurement of the natural remanent magnetisation (NRM) of larger primary clasts was carried out using a portable fluxgate magnetometer. This method is considered a rapid and reliable indicator of whether pyroclastic deposits were emplaced above Curie Point temperatures (Hoblitt an d Kellog, 1979). Ten oriented and marked clasts of at least 5 cm in diameter were taken out of each outcrop. With the portable fluxgate magnetometer the positive or negative alignment in 6 orientations (x, y, z) were measured. If more than 8 out of 10 clasts indi cated the same pattern in all six directions it was assumed that deposition occurred above the Curie Point (c.350 °C) or blocking temperature, where ferromagnesian microlites are able to orientate in a uniform Chapter 2 Methodology 33 direction (McClelland and Erwi n, 2003). If only seven or ei ght clasts showed the same pattern, less confidence could be placed in the result, but it could be equally valid since some clasts could be xenoliths entrained from the bed by the BAF, or represent already cooled outer parts of a collapsing dome. Wh ere less than seven clasts showed a common orientation in their magnetic directions, th e deposits were interpreted as random and therefore emplaced below Curie Point temperatures. The summit lava dome remnant was mapped in detail for internal-structures, geometry and related deposits such as blocky lava flows and rock-avalanche deposits. Samples were collected from various portions of the dome for comparison of petrographic and bulk chemical compositions. In addition, stra tigraphic relationships of lavas around the summit area were investigated and several potentially Maero-aged lava flows were sampled in the standard way for petrographic/geochemistry studies. Outcrop positions were collected using a Garmin hand-held G PS and all maps were constructed using the GIS program ArcView. Samples taken for fu rther laboratory processing were brush- cleaned or cleaned in a sonic bath prior to drying at 50-70 °C. 2.3 Mineralogy 2.3.1 Sample Preparation Dense rock clasts of lapilli size or greater were sliced into 1 cm thick tablets to fit petrographic slides (27×46 mm) using a Stru ers Discoplan-TS or a coarse rock saw. Progressively finer silicon carbide grits (200 to 1000 grit) were used to grind the basal surface of the rock samples. A two component epoxy (EpoTek) was used to impregnate the samples and, if porous, to glue them ont o the pre-roughened glass slide surface. After 24 hrs the tablets were cut close to the slide and the thin-section was ground to approximately 30 µm thickness using varyi ng sizes of silicon car bide. For EMPA and SEM studies, thin sections were polished us ing a Struers Planopol-3 and increasingly finer grades of diamond paste (6, 3, and 1 µm). Glass shards selected for EMPA were sepa rated using a Frantz isodynamic magnetic separator. The 125-250 µm grain size fracti on proved best for magnetic separation and was chosen for further processing. Grains for particle and glass compositional studies were first embedded in epoxy and if porous , air bubbles were removed by repeated vacuum extractions. The epoxy plugs were late r ground until the majority of clasts were Chapter 2 Methodology 34 intersected. The intersected grain surfaces were glued onto petrographic slides and thin- sectioned for microscopic studies and polished as above for EMP analysis. Pumice clasts received special treatment due to their fragility. For petrographic, mineral chemistry, and textural studies, pumice samp les were cut into cubes and impregnated with low viscosity methyl methacrylate ( ρ =0.943 gcm -3 at 20 °C a clear, colourless and low-viscosity polymer η=0.58 mPas at 50 °C) before thin sectioning. The samples were repeatedly evacuated until no further bubbles were released from the sample, and then the fluid was polymerised using Cobalt-60 gamma radiation. 2.3.2 Microscopic Studies Petrographic studies were carried out using a polarising microscope (NIKON Eclipse E600W POL) with integrated 5 Mega-Pixel Digital Camera. Proportions of minerals to groundmass are based on 500 counts per thin sect ion and are calculated on a vesicle-free basis. Microlites were cla ssified as groundmass. The following categories were used to classify crystal sizes: macrocrysts (> 1 mm), phenocrysts (1 mm to 150 µm), microphenocrysts (150-50 µm), and microlit es (<50 µm). Hornblende reaction rim thicknesses were only measured on crystals that were cut parallel to a, b or c-axes. Thickness analyses were carried out on 6-15 crystals per thin section with 10-23 measurements per crystal, depending on crys tal size. For particle textural studies a binocular microscope (Leica Wild MZ8) was also utilised. 2.4 Geochemistry 2.4.1 X -ray Fluorescence Spectroscopy Samples from individual units were collected to obtain a representative suite of clast types from one deposit and were then studied for their bulk rock geochemistry. Hence, 84 pumice and lithic clast specimens were crushed and ground using a tungsten carbide ring mill. Major element concentrations were determined by X-ray fluorescence (XRF) spectrometry (Siemens SRS 3000; The Univers ity of Auckland) on fused glass discs following a method of Norrish and Hutton ( 1969), and calibrated to a suite of 36 international standards at Victoria Univer sity of Wellington and The University of Auckland. In general, the precision is ±1% (1 σ). Trace elements were analysed using Chapter 2 Methodology 35 pressed powder pellets according to methods based on Norrish and Chappell (1977). Theoretical detection limits are 1-2 ppm and reproducibility ±5% (1 σ). Some samples of the Burrell episode and Pyramid eruptives show considerably lower Ba values. This discrepancy is the result of an analytical error, indicated by the low-Ba samples being confined to one XRF sample batch, with low values not shown by parallel laser ICP-MS analysis. 2.4.2 Electron Microprobe Analysis Mineral chemical studies were focussed on the pheno- and microcryst phases of clinopyroxene, hornblende, and Fe-Ti oxide s, and in addition to minor groundmass constituents of apatite, biot ite, olivine, and orthopyroxene. Inclusions of glass, Fe-Ti oxides and plagioclase in different host mine rals were also studied. The matrix glass within tephra clasts was also analysed. Plagioclase was only analysed as a microlite phase within some pumice samples; it was not analysed in phenocryst or groundmass phases because the distinction of plagioclase and groundmass as well as zonation within crystals were impossible to identify on backscatter electron imagery on the University of Auckland electron microprobe. Mineral and glass electron microprobe analyses (EMPA) were carried out with a JEOL JXA-840 equipped with an energy dispersive spectrometer at the University of Auckland. Analytical conditions were an acc elerating voltage of 15kV, a beam current of 600 pA and 100 seconds live-time. It has be en recognised that K and particularly Na are mobile in natural as well as synthetic glasses and can be strongly underrepresented in the analysis, especially if another oxide such as SiO 2 dominates the bulk glass composition (Nielsen and Sigurdsson, 1981). Fo r glass EMPA, a defocused beam of 20 µm diameter was hence used to minimise alkali loss. At least 10 glass shards per sample were measured and where possible, two anal yses per shard were collected. For mineral EMPA a focused beam was applied. Where possi ble, analysed glass shards and minerals were documented using back scatter electron imagery. All glass EMPA are normalised to 100%. Since there is no agreement among geochemists regarding the use of normalised vs. unnormalised glass compositional data, normalised data are used in this study for further calculations. If unnormalised glass data were used for calculations and the difference to 100% is solely attributed to the presence of H 2 O then the matrix used to calculate weight percentages of element oxides Chapter 2 Methodology 36 must be modified, i.e. the matrix effect difference is about 1% absolute in the 5% water range (J.J. Donovan, person. comm., 2006). Dependi ng on the difference of the totals to 100%, the effect of normalisation can be large, especially if one component is dominant (i.e. SiO 2 ). In this case, silica abundances can increase up to several wt.%. However, normalised and unnormalised sample populations still show a similar trend in bivariate diagrams. Detection limits for EMP analysis are given Table 2-1. As part of an inter-laboratory EMPA project (64 participating laboratories) basaltic glass was analysed with deviation from its accepted composition listed in Table 2-1 (R. Sims, unpubl. data). Table 2-1. Detection limit of element oxides measured at the EMP (University of Auckland) including the deviation from a reference glass composition. Detection limit Si O 2 TiO 2 Al2 O 3 FeO MnO Mg O CaO Na 2 O K 2 O P 2 O 5 SO 3 Cl Cr2 O 3 NiO 1 σ [wt.%] 0.1 1 0 . 08 0 . 0 6 0 .0 7 0 . 0 7 0 .0 7 0 . 04 0. 1 1 0 . 0 3 0 . 0 7 0 .0 6 0 .0 3 0 . 0 6 0 . 10 3 σ [wt.%] 0.3 2 0 . 23 0 . 1 8 0 .2 0 0 . 2 0 0 .2 1 0 . 13 0. 3 2 0 . 1 0 0 . 2 0 0 .1 9 0 .0 8 0 . 1 7 0 . 30 International Round-Robin Inter-lab EPMA project (6 4 participating labs) : Difference to reference values of re-fused basaltic glass composition Si O 2 TiO 2 Al2 O 3 FeO MnO Mg O CaO Na 2 O K 2 O P 2 O 5 - 0. 1 8 0 . 12 -0 . 05 - 0. 0 4 -0 . 01 0 .1 7 - 0 .0 2 0. 0 5 0 . 0 2 0 . 0 1 2.4.3 Laser Inductively Coupled Plasma Mass Spectrometry For trace element studies, 46 samples from the Maero Formation and four lava flows were chosen as a representative suite of the youngest eruptive period of Taranaki. Rock powders produced for XRF analysis were also used for trace element analyses using an Agilent 7500ce quadrupole inductiv ely-coupled plasma mass-spectrometer (ICP-MS) coupled with a NewWave Resear ch UP-213 laser-ablation unit at the Department of Geosciences, University of Ma inz, Germany. Instrumental and analytical procedures are described in Nehring et al., (in press). Glass beads were prepared by using an iridium strip heater equipped w ith an argon atmosphere chamber to prevent oxidation and volatile loss. Approximately 30 mg of powder was placed on the iridium strip and melted for 10 s using a 110-120 A cu rrent at 1100-140 0 °C. Three analyses per glass bead were performed using a 100 µm b eam with a 10 Hz frequency and 60 s count time. Plasma torch conditions were kept at an oxide ThO/Th ratio below 1%. As a carrier and nebulizer gas, argon was used. Detection limits are 0.001-0.5 µgg -1 (Nehring et al., in press). After nine analyses (i.e . three samples), certified glass reference Chapter 2 Methodology 37 materials NIST SRM 612, NIST SRM 610 or BCR-2 were measured with reproducibilities of <10% (1 σ ) and all values were within 10% of published analyses (Pearce et al., 1997). Data evaluation was carried out using Glitter software and 43 Ca and 44 Ca as internal standards. An initially observed Ta anomaly (four samples had elevated values) was later related to use of a tungsten grinder in the preparation process. Tantalum is an important high field strength element and indicative of magma mixing and fracti onation processes since it is compatible in biotite and hornblende. Three samples were re-analysed after preparation using an agate mill, and no elevated Ta con centrations were found. Due to this potential contamination in the XRF dataset, Ta wa s not used further for geochemical data interpretation from this suite of analyses. 2.5 Porosity and Permeability Porosity and permeability measurements were carried out at the Department of Geological Sciences, University of Ore gon, USA. Cores (approx. 2.54 cm diameter and height) of clasts were drille d, in two or three mutually pe rpendicular orientations where sample size allowed (Fig. 2-1). From the known volume of the core (V c), measured with a sliding calliper, and the mass of the cores (mc), the bulk density, ρbulk ( ρbulk =m c/V c) was determined. A Micromeritics multi-volum e He-pycnometer was used to determine the helium-accessible volume (V He ) of the sample by measuring the pressures in the sample chamber (P1) and the reference chamber (P2). The V He results in an estimate of skeletal density, ρskeletal ( ρskeletal =m c/V He ) and connected porosity, φconn, ( φconn=V He /V c). However, even rocks of dense appearance commonly possess non-accessible, isolated cavities and cracks, and vesiculated rock s usually have both isolated pores and interconnected pores, and therefore ρskeletal is always lower than the solid density, ρsolid. In order to determine ρsolid, representative samples were powdered and their volume (V p) determined by He-pycnometry ( ρsolid=m p/V p). The bulk vesicularity ( φbulk =1- ( ρbulk / ρsolid) can then be used to calc ulate the isolate porosity ( φiso= φbulk - φconn). Long drilled cores of pumice were cut in to two to examine reproducibility and homogeneity along a specific orientation. This showed that pumice samples were mostly homogeneous, with differences in φbulk of 0.1 to 8% ov er the range of φbulk =34- 84%. However, in single larger pumice clasts with up to seven drilled cores per sample in multiple orientations, differences in φbulk of up to 20% were noted. Chapter 2 Methodology 38 Figure 2-1: The specimen illustrates how pumice cores were obtained. As in this case, six cores in three mutually perpendicular orientations were drilled. Error analysis of φconn for the entire data set was assessed by evaluating V c using diameter and height variations, and V He by using pressure variations. The core diameter and height were measured five times each and then used for volume calculations resulting in 25 individual V c. The standard deviations vary between 0.005 and 1.171 cm3 . Pressures P1 and P2 were measured at least three times in consecutive runs giving at least 9 results for V He with standard deviations between 0.009 and 0.588 cm 3 . Standard deviations for V c and V He indicate a concentration of data points below 0.1 cm3 with most data points enveloped by the 0.3 cm 3 border (Fig. 2-2b). The latter border is also regarded as the maximu m accepted standard deviation for both V c and V He . The standard deviation of 225 individual φconn per sample (25 V c × 9 V He ) ranges between 0.12% and 6.20% (F ig. 2-3). Using the 0.3 cm 3 border, the upper range reduces to 3.07%. There is no relationship between stan dard deviation and vesicularity, although highest deviations are observed for φconn=44-73%. For each specimen a standard deviation of 1% for φconn is adopted (average 0.87%). As demonstrated in Fig. 2-3, two sources of error are apparent. The core drilling carried out in New Zealand was not as precise as that in Oregon. Furt her, the standard deviation of sample volumes measured by He-pycnomet ry varies especially for highly porous samples and could have been caused by, for exam ple, 1) incorrect purging of the sample chamber to remove all present air with He prio r to analysis, 2) residual water in vesicles which adds vapour pressure to the gas pressure, and 3) pr essures were not equilibrated Chapter 2 Methodology 39 either in the sample or reference chamber, or both. The latter can be excluded because the analysis was only performed after pressures did equilibrate. For future measurements, two changes are recommended based on experience gained in this study. Firstly, cores to be analysed should be stored in a desiccator to avoid water absorption due to varying room temperatures and/or air moisture. This potentially inhibits the development of any vapour pressure during analysis. Secondly, a minimum of five sample runs is recommended. In this study it is observed that the first of three runs record (in most instances) the highest pressures with runs 2 and 3 being similar. This strongly implies that pulsatory purging of the sample chamber prior to analysis was insufficient. Over five runs, data from th e first or the first two runs should be disregarded, and only the last three should be averaged and used for further calculations. Errors made during pycnometry measuremen ts at the University of Oregon were avoided when additional samples were analysed using the recently available He- pycnometer (Quantachrome U ltrapycnometer 1000) at Masse y University. The density of powderised samples from XRF-analysis (n=3 2) was determined by measuring at least 5 runs where only the last three were averag ed. Averaged analyses were only accepted if the standard deviation was below 0.005 vol.%. Figure 2-2: Standard deviations for diameter and length of the cores (a) and for volumes V c and V He (b). It is noted that one sample in a) is off the chart at a standard deviation of 0.1321 cm. b) Samples are differentiated into those drilled in Oregon and at Massey. Oregon samples show smaller variations for both V c and V He . Chapter 2 Methodology 40 Figure 2-3: Standard deviation for connected porosities. The limit of 0.3 cm3 for Vc and VHe in Fig. 2-2b is used as maximum limit. Cores drilled in Oregon are shown for comparison. The gas permeability of cores was measured in a capillary flow porometer (Porous Materials, Inc.) with air as the working gas. Permeability of the sample core is measured by applying successively increasing flow rates and measuring the differential pressures at each flow rate increment. Different flow rates were applied depending on the vesicularity (i.e. low flow ra te for less vesicular samples). Three to six runs per sample were performed to demonstrate reproducibility (Fig. 2-4). Figure 2-4: Permeability measurements of individual cores were performed three to six times, partially using multiple flow rates, in order to assess reproducibility. In this case, the red grap h suggests higher flow rates compared to the other three runs and was excluded from further calculations. Chapter 2 Methodology 41 Data were calculated as de scribed by Wright et al. ( 2007) and Rust and Cashman (2004). A modified form of Darcy’s Law know n as the Forchheimer’s equation is used to account for the energy loss in non-laminar flows. The viscous (Darcian) permeability, k 1 , and the inertial (non-Darcian) permeability, k 2 , can be calculated by: 2 21 2 0 2 2 ssC i kkPL PP νρνμ +=− [ E q. 2-1] where Pi and P0 are the entrance and exit pressures at the sample core, respectively. In addition, the viscosity ( μ) and density ( ρ) of air, and the core length (L c) are needed. The superficial velocity ( υs) is calculated by dividing the flow rate by the cross- sectional area. The pressure P refers to the fluid pressure at which the velocity and viscosity are measured; in the Forchheimer equation P=P 0 so that μ and ρ correspond to values at atmospheric pressure. 2.6 Scanning Electron Microscopy Scanning electron images were obtained fr om carbon coated thin and thick section samples at HortResearch, Palmerston North (Cambridge 250 Mark III) and the University of Oregon (JEOL 6300). Operating conditions were an accelerating voltage of 20 kV and 10 kV, respectively. Secondary electron images and backscatter images were obtained in analogue and digital form. For textural studies, SEM images were partially binarised; white represents th e groundmass glass and minerals and whereas black constitutes void space (i.e. vesicles and cavities). 2.7 Fourier Transform Infrared Spectroscopy Fourier Transform Infrared Spectroscopy (F TIR) was used for studies of dissolved water contents in andesite melt inclusion hosted in mineral phases. The three major mineral phases present in andesite rocks, pl agioclase, clinopyroxe ne and hornblende, all have good to very good cleavage which makes them less suitable as an inclusion host. During decompression (i.e. magma ascent) volatil es dissolved in melt inclusions within these phases may escape along cleavage planes, and hence any measured volatile content may or may not reflect the melt volatile content at the time of capture. Despite this constraint, clinopyroxene phenocrysts were targeted since, bei ng translucent, shape Chapter 2 Methodology 42 and locations of inclusions could easily be seen and those appearing intact focussed upon. In addition, the clinopyroxene was more robust than plagioclase and hornblende. Selected pumice and lithic samples of the Burrell Lapilli eruption were crushed in an agate mortar and clinopyroxene crystals were hand picked. Crystals were then checked for glass inclusions using a polarising microscope. Selected crystals were glued on a petrographic slide using crystal bond (a resin that is worked and reworked by melting at about 100 °C), and gently ground using 600 grit paper until glass inclusions were intersected on one side. The surface was polished using 6 µm and 1 µm diamond paste. The wafer was then turned and glued with the polished surface onto the slide by heating the crystal bond. The exposed side was also ground and polished. Most melt inclusions, however, were not intersected on both sides since they were too small. The double-polished wafers were analysed using a Thermo Nicolet Nexus 670 FTIR spectrometer attached to an IR microscope at the Department of Geological Sciences, University of Oregon. Infrared spectra were obtained in the mid-infrared and near- infrared area of 650-6000 cm -1 with a resolution of 0.09 cm -1 . Water and CO 2 species were investigated at the absorbance bands 1630 cm -1 , 2350 cm -1 , 3550 cm -1 , 4520 cm -1 and 5230 cm -1 (Table 2-2). Due to the size of melt inclusions, apertures of 10-20 µm by 10-20 µm were often applied, which is ne ar the minimum limit of instrumental detection, however, analysis time was increase d from 256 s for larger inclusions to 1024 s for the smallest inclusions. Because melt inclusions in clinopyroxene are small and irregular in shape, they were mostly exposed only on one face. In thes e cases, their thicknesses were estimated optically using a polarising microscope, and thus total H 2 O concentrations are approximations. Where the inclusion was inters ected at both sides, the thickness was determined by using a micrometer. The error involved in the analysis is difficult to determine but it is assumed that water estimates may have an error of up to 20%. Infrared spectra of intersected and partially truncated melt inclusions only show a broad asymmetric peak at the infrared band 3550 cm -1 , which is assigned to total H 2 O absorbance (molecular H 2 O plus OH - ). Although the molecular water absorbance peak at 1630 cm -1 is present, peaks at 5230 cm -1 and 4520 cm -1 are in most cases absent. In order to determine peak heights, appropriate baselines had to be defined by applying a straight baseline for Abs3550 and a curved line (using a hand-fitted curve) was applied for Abs1630 , Abs 4520 and Abs5230 , where present (Silver et al., 1990; Luhr, 2001). Total water concentrations are determined using the Beer-Lambert Law: Chapter 2 Methodology 43 0152.18 3550 3550 2 ρεdC Abs OH= [ E q. 2-2] 3550 3550 0152.18 2 ρεd Abs C OH = [ E q. 2-3] where d is the thickness [cm] of the melt inclusion, ρ is the density of glass [gL -1 ] at 20 °C and 1 bar, ε3 55 0 is the molar absorption coefficient [Lmol -1 cm-1 ], Abs 3550 is the absorbance at the wavelength 3550 cm -1 and CH2O is the weight fraction of molecular water. Table 2-2. Water and carbon dioxide peaks on the FTIR spectra including their bonds within glasses. The glass composition of most inclusions was measured by EMPA and their density calculated as described in Best (2003). Data for partial molar volume of oxides, their thermal expansivity and compressibility ar e from Lange and Carmichael (1990), Lange (1997), and Ochs and Lange (1997). Since the dissolved water content affects the density of melt, the density is iterat ively determined. Molar absorptivities ε [Lmol -1 cm- 1 ] for IR absorption bands of Fe-bearing ande sites were applied (M andeville et al., 2002). 2.8 Thermal Analysis Simultaneous Differential Scanning Calori metry and Thermal Gravimetric Analysis (DSC-TGA) was carried out on pumice matrix glasses using a SDT Q600 (TA Instruments Waters-LLC). This device measur es heat flow and weight changes with transitions and reactions in materials to temperature of up to 1500 °C. Reactions and transitions comprise endothermic and exotherm ic reactions with changes in weight (e.g. degradation) or without weight changes (e.g . melting and crystallisation). For this study DSC-TGA was used to determine whether it was possible to measure total volatile Chapter 2 Methodology 44 concentrations by weight difference since FTIR spectroscopy cannot be applied to highly vesicular samples. In order to obtain pure matrix glass, singl e grey and brown pumice clasts were crushed using a mortar and sieved to obtain the <125 µm fracti on and afterwards ground to coarse silt. The glass fraction was isolated using sodium polytungstate ( ρ=2.52 gcm -3 ) and centrifuged, then dried at 70 °C for at least 12 hours. The glas s fractions (<<5 g) were checked for purity using a polarising microscope (Fig. 2-5). Prior to DSC-TGA analysis, glass samples were powdered using an agate mortar and 20-30 mg were placed into the sample holder. For glass DS C-TGA analysis the following method was progressively developed: the sample wa s first heated to 110 °C at 20 °Cmin -1 and kept isothermal at 110 °C for 15 min, then h eated to 850 °C or 1000 °C at 5 °Cmin -1 and then cooled. The experiments were carried out in a nitrogen atmosphere to prevent iron oxidation, however, some samples e xperienced some degree of oxidation. Figure 2-5: Separated groundmass glass fraction from a pumice clast (SD32 ). The output of the analysis is a graph showing the weight difference (in %), heat flow (Wg -1 ) and the temperature difference (°Cmg -1 ) over the temperature range (Fig. 2-6). A sharp peak is observed at about 105-110 °C for heat flow and temperature difference. These peaks are evidence of an endothermic reaction and are related to the release of water from the surface of the glass. A drop in weight at this temperature is clearly observed. With increasin g temperature the weight continuously decreases, fi rst at a high rate and then the slope gradually flattens with higher temperatures. The difference in weight was measured between the maximum and minimum point of the graph in the Chapter 2 Methodology 45 interval of c.110 °C to the maximum temp erature using the software TA Universal Analysis 2000. The graphs for heat flow a nd temperature difference proceed evenly and show no sign for further endothermic or exothermic reactions. One sample was analysed using a Stanton Redcroft DTA 673-674 coupled with infrared H 2 O and CO 2 detectors using the procedures described by Morgan (19 77) and Lindgren et al. (2002). The sample was heated at 5 °Cmin -1 in a gas flow of 300 ml N 2 min-1 . The analysis was carried out by H.B. Lindgren (Geological Survey of Denmark and Greenland). Figure 2-6: Thermal analysis of volcanic glass to 850 °C. The change in weight (green axis) was measured after the isothermal break at 110 °C. Chapter 2 Methodology 46 2.9 References Best, M.G. 2003. Igneous and metamorphic petrology. Blackwell Science. Malden. Fine, G. and Stopler, E. 1985. The speciation of carbon dioxide in sodium aluminosilicate glasses. Contributions to Mineralogy and Petrology 91, 105- 121. Hoblitt, R.P. and Kellog, K.S. 1979. Emplacement temperature of unsorted and unstratified deposits of volcanic rock debris as determined by paleomagnetic techniques. Geological Society of America Bulletin 90 , 633 -64 2. Lange, R.A. 1997. A revised model for the density and thermal expansivity of K 2 O-Na 2 O-CaO-MgO- Al2 O 3 -SiO 2 liquids between 713 and 1896 K: extension to crustal magmatic temperatures. Contributions to Mineralogy and Petrology 130, 1-1 1. Lange, R.A. and Carmichael, I.S.E. 1990. Thermodynamic properties of silicate liquids with emphasis on density, thermal expansion, and compressibility. In: Nicholls, J. and Russell, J.K. (eds.) Modern methods of igneous petrology: understanding magmatic processes. Reviews in Mineralogy 24, 25-6 4. Lindgren, H., Drits, V.A., Sakharov, B.A., Jakobsen, H.J., Salyn, A.L., Dainyak, L.G. and Krøyer, H. 2002. The structure and diagenetic transforma tion of illite-smectite and chlorite-smectite from North Sea Cretaceous-Tertiary chalk. Clay Minerals 37, 429-4 50. Luhr, J.F. 2001. Glass inclusions and melt volatile contents at Parícutin Volcano, Mexico. Contributions to Mineralogy and Petrology 142, 261 -28 3. Mandeville, C.W., Webster, J.D., Rutherf ord, M.J., Taylor, B.E., Timbal, A. and Faure, K. 2002. Determination of molar absorptivities for infrared absorption bands of H 2 O in andesitic glasses. American Mineralogist 87, 813-8 21. McClelland, E. and Erwin, P.S. 2 003. Was a dacite dome implicated in the 9,500 B.P. collapse of Mt Ruapehu? A palaeomagnetic investigation. Bulletin of Volcanology 65, 294-3 05. Morgan, D.J. 1977. Simultaneous DTA-EGA of mineral and natural mineral mixtures. Journal of Thermal Analysis 12, 245-2 63. Nakamoto, K. 1978. Infrared and Raman spectra in inorganic and coordination compounds. 3rd ed. Wiley. New York. Nehring, F., Jacob, D.E. , Barth, M.G. and Foley, S.F. (in press). Laser-ablation ICP-MS analysis of siliceous rock glasses fused on an iridium strip heater using MgO dilution. Microchimica Acta . Nielsen, C.H. and Sigurdsson, H. 1981. Quantitative methods for el ectron microprobe analysis of sodium in natural and synthetic glasses. American Mineralogist 66, 547 -5 52. Norrish, K. and Hutton, J.T. 1969. An accurate method for the analysis of a wide range of geological samples. Geochimica et Cosmochimica Acta 33, 431- 451. Chapter 2 Methodology 47 Norrish, K. and Chappell, B.W. 1977. X-ray fluorescence spectroscopy. In: Zussman, J. (ed.) Physical methods in determinative mineralogy. Academic Press. New York. pp. 201-27 2. Ochs, F.A., III and Lange, R.A. 1997. The partial molar volume, thermal expansivity, and compressibility of H 2 O in NaAlSi 3 O 8 liquid: new measurements and an internally consistent model. Contributions to Mineralogy and Petrology 129, 155 -16 5. Pearce, N.J.G., Perkins, W.T ., Westgate, J.A., Gorton, M.P ., Jackson, S.E., Neal, C.R. and Chenery, S.P. 1997. A compilation of new and published major and trace element data for NIST SRM 610 and NIST SRM 612 glass reference materials. Geostandards Newsletter 2 1 , 115-1 44. Platz, T. 2001. Mapping and characterisation of the vol caniclastic Maero formation deposits on the northwestern sector of Egmont volcano (Mt. Taranaki), New Zealand. Unpublished Diploma mapping thesis. Ernst-Moritz-Arndt Universitä t, Greifswald, Bundesrepublik Deutschland. Rust, A.C. and Cashman, K.V. 2004. Permeability of vesicular silicic magma: inertial and hysteresis effects. Earth and Planetary Science Letters 228, 93-1 07. Silver, L.A., Ihinger, P.D. and Stolper, E. 1990. The influence of bulk composition on the speciation of water in silicate glasses. Contribution to Mineralogy and Petrology 104, 142 -16 2. Stolper, E. 1982. Water in silicate glasses: an infrared spectroscopic study. Contributions to Mineralogy and Petrology 81, 1-17. Wright, H.M.N., Cashman, K.V., Rosi, M. and Cioni, R. 2007. Breadcrust bombs as indicators of Vulcanian eruption dynamics at Guag ua Pichincha volcano, Ecuador. Bulletin of Volcanology 69, 281- 300. Chapter 3 The Maero Eruptive Period 49 Chapter 3 The Maero Eruptive Period 3 Chapter 3 The Maero Eruptive Period This chapter describes the distribution and nature of the diverse eruptives and related deposits emplaced during the Maero Er uptive Period. Lava domes were the predominant extrusive product of this period; hence associated volcaniclastic units include Block-and-Ash-Flow and pyroclastic-su rge deposits. The previous stratigraphy of the Maero Eruptive Period is revised to include four new eruption episodes. A new tephrostratigraphy of the Maero Formation is introduced, its reference sections are defined and the type section described. 3.1 Introduction In the past four decades the tephrostratigra phy of Mt. Taranaki has been studied in detail, with the predominant focus being on medial and distal deposits of the largest sub-Plinian eruptions (Neall, 1972; Toppi ng, 1972; Franks, 1984; Alloway et al., 1995). As the chronology of tephra eruptions and intervening lahar and debris avalanche events became increasingly better understood, atten tion has now turned to improving the dating of lava flow and dome-forming episode s. The focus on mappable, mainly distal fall deposits excluded detailed studies of the small, apparently more-frequent eruptions, whose deposits are only preserved on or near the edifice itself (Turner et al., 2007). This is particularly true for the Maero Eruptive Pe riod, where many of the eruptions were of small scale or low-explosivity and hence te phra deposits are only preserved in proximal areas (Druce, 1966). Typically during this period, Block-and-Ash Flow (BAF) and surge deposition occurred on the NW sector of the mountain, whereas thin ash layers associated with these events were predominantly deposited on the northern, eastern, and Chapter 3 The Maero Eruptive Period 50 southern flanks (Neall, 1979). These thin ash layers are impossible to distinguish from one another based on their physical properties and appearances (Druce, 1966). Hence geochemical correlation techniques are required. Glass chemistry has proven to be one of th e most useful tools for correlating rhyolite tephra units within New Zealand (e.g. Lowe, 1988; Shane, 2000). Hence, this method was also applied to correlating proximal andesite tephras on Mt. Taranaki. In this instance glass chemistry of fall deposits from the northern, eastern and southern flanks were compared to fall, surge and BAF deposits on the NW sector. Early attempts to use glass chemistry for tephra correlation at Mt . Taranaki were unsuccessful (May, 2003; Platz et al., 2004) mainly due to the poor knowledge of the te phra sequence and the apparent broad compositional diversity in glass within individual units. The problems in studying the glass chemistry of andesitic tephras at Taranaki are described further in Chapter 4, along with a procedure of optimising this method by identifying and removing contaminated data points from the analyses. This chapter introduces the range of volcanic deposits erupted in the last 1000 yrs and summarises type and reference sections representing this. Emphasis will be given to deposits of pyroclastic density currents (in particular BAF deposits) and their distribution and flow paths. These are volumetrically the predominant deposits representative of the eruptive period. By means of field studies and glass chemical analyses, a revised stratigraphy is introduced for the last c.1000 yrs of volcanic activity at Mt. Taranaki. For a better comprehension of the volcanic history of Mt. Taranaki, two new terms are presented here. Previously, volcaniclastic de posits in the latest 500 years of volcanic activity at Mt. Taranaki were mapped under the informal term “Maero debris flows” (Neall, 1979). Investigations since have shown that these units are primarily Block-and- Ash Flow deposits. Additionally, older eruptions have also been shown to belong to this mapping unit due to their stratigraphic position, similar composition, style and distribution (Platz, 2001; Croni n et al., 2003). As a result of this, the term “Maero Eruptive Period” is here introdu ced to describe the latest period of volcanic eruptions at Mt. Taranaki, where since c.1000 yrs B.P. ac tivity has been of a similar style and affected primarily the NW sector of the vol cano. Associated with this is the “Maero Formation” – here defined to represent a ll volcanic products erupted during the Maero Eruptive Period. Chapter 3 The Maero Eruptive Period 51 3.1.1 Previous Studies The first person to realise that Mt. Tarana ki was comparatively recently active was Mr A.W. Burrell who found scoria clasts in tree forks in 1883 (Oliver, 1931). The discovery of a Maori umu (earth oven) bur ied by volcanic deposits on the upper eastern slope of Mt. Taranaki was al so described by Oliver (1931) . It was covered by a c.38 cm pumice layer (later named Burrell Lapilli). This discovery clearly testified that Mt. Taranaki was active during the colonisation of New Zealand by Maori. Further, other recent ash layers were described as Newa ll Ash [recognised by L.I. Grange and N.H. Taylor in the 1930s (Druce, 1966)], Burrell Ash [described by Taylor (1953)] and Puniho Grit [described by Grant-Taylor (1964)]. The detailed stratigraphic and botanical work of Druce (1966) showed that Burrell Ash and Newall Ash could be further subdivided into multiple events, and hence revealed that more eruptions occurred during the volcano’s recent history. Druce (1966) recognised that considerable variations in vegetation and soil development related to the distribution of volcanic deposits from one side of the volcano to the other. Although he described fundamental differences in the character of deposits between eastern and the northwestern slopes of Mt. Taranaki, he desc ribed all volcanic deposits as tephra layers. Most deposits on the NW sector contained char coal fragments as well as a poorly sorted mix of lapilli to block sized clasts set in an ash matrix, according to Druce (1966). He also recognised that these particular units were confined to the W-NW flank within c.14 km from the volcano’s summit. On the basis of three soil-forming breaks, Druce (1966) grouped 10 “tephra” layers into three Formations: Newall Formation, Burrell Formation, and Tahurangi Formation (Table 3-1). The Burrell Lapilli eruption was dated by Druce at AD 1655 using dendrochronology of trees presumably killed and/or damaged by the eruption. The Tahurangi Ash was determined by dendrochronology (of possibly damaged trees) and pe at accumulation rates more loosely around the age of AD 1755 (Druce, 1966). The discovery of another Maori umu on the SE flank revealed further age details. Charcoal from the oven was dated at 470 ± 55 yrs B.P. (i.e. c. AD 1480), which is indicative for the presence of Maori prior to the Newall eruptions (Topping, 1974). In addition, two ash layers previously not recognised were found between the Kaupokonui Tephra and Newall Ash (Topping, 1974). Chapter 3 The Maero Eruptive Period 52 Neall (1979) was the firs t to interpret the charcoal-containing units on the NW sector as being related to nuées ardentes1 . The type section for a succession of these units is located in the Maero Stream (Neall, 1979). Here at least 14 debris and pyroclastic flow deposits considered to be emplaced over the past c.500 yrs were collectively mapped and termed the Maero Debris Flows (Neall, 1979 ). These deposits overl ie a peat dated at c.1975 yrs B.P., which Neall (1979) used to interpreted a 1500 yrs period of relative quiescence in the NW sector of the volcano (Neall, 1979). Table 3-1. Stratigraphy of the youngest deposits of Mt. Taranaki (Druce, 1966). Studies on the destruction and recovery of vegetation around the volcano during individual eruptions, as well as pollen anal ysis, added further details to the eruptive history, particularly in determining the lengt h of some of the repose periods (Clarkson, 1990; McGlone, 1982; McGlone et al., 1988; Lees and Neall, 1993; McGlone and Neall, 1994; Table 3-2). McGlone et al. (1988) also implied that the youngest eruption, the Tahurangi Ash, could have occurred at any time between AD 1700 and AD 1900. This was further refined by studies of Lees and Neall (1993) who found Pinus pollen beneath the Tahurangi Ash indirectly indicating that the age of the Tahurangi eruption could have been closer to AD 1860. Platz (2001) confirmed that the deposits exposed on the northwestern sector were primarily deposited by pyroclastic flow deposits and confirmed that some units travelled more than 13.5 km from source, and henc e well beyond the Nati onal Park boundary. In addition, a poorly sorted blas t-like deposit was mapped exte nding at least 14 km from source with a sharp lateral confinement to the area W-NW of the summit of Mt. Taranaki (Cronin et al ., 2003). Fall deposits in the NW sector are mainly thin ash layers, 1 Nuée ardente (French: incandescent cloud) is described as a fast moving, highly destructive ground- hugging pyroclastic density current (cf. Lacroix, 1904). Chapter 3 The Maero Eruptive Period 53 Table 3-2. Stratigraphy of the youngest deposits at Mt. Taranaki until 2003 (Druce, 1966 ; Neall, 1972 ; Neall, 1979 ; McGlone et al., 1988; Lees and Neall, 1993). which are mostly genetically related to pyroclastic density currents (i.e. co-ignimbrite ashes). Due to the difficulties in reliably correlating the fall sequence on the east and south flanks with deposits of pyroclastic density currents on the W-NW sector, Cronin et al. (2003) made a first attempt to dis card the partially confusing nomenclature for eruptive units and introduced a new stratigraphy on the basis of fall layers and pyroclastic flow deposits. According to this and new radiocarbon dates, six eruptive episodes (M1-M6; M=Maero) were defined which occurred after AD 1040 (Table 3-3; Cronin et al., 2003). Cronin et al. (2003) thus defined this period as being characterised by effusive to explosive events producing lava domes, lava flows and pumice fall deposits. Chapter 3 The Maero Eruptive Period 54 Table 3-3. Current stratigraphy of the last 1000 yrs of activity at Mt. Taranaki (modified after Cronin et al., 2003). 3.1.1.1 Previous Work on Lava Flow Stratigraphy The interrelation of tephra fall sequences expos ed on the eastern to southern flanks and the successions of pyroclastic density currents in the W-NW sector of Mt. Taranaki is poorly known. In addition, the formation of lava flows of the upper cone of Mt. Taranaki including Fanthams Peak is also not well understood or correlated to other eruptive episodes. Paleomagnetic studies imply the formation of several of the summit lava flows in two periods approximately 400 and 700 yrs ago (Downey et al., 1994), which corresponds with the young coverbed sequences on many of the flows. The succession of lava flows forming the upper portion of Mt. Taranaki (Fig. 3-1) was subdivided into five age/geographic groups: 1. The Warwicks Castle Group, representing the oldest lava flows that form the present upper cone of Mt. Taranaki. They include di stinct landscape features around the edifice such as Warwicks Castle and Bobs Ridge, an d extend between a majo r break in slope at Chapter 3 The Maero Eruptive Period 55 about 1000 m up to 2000 m elevation. This gr oup has an inferred maximum age of 8 ka (Stewart et al., 1996). 2. Neall (2003) introduced the Peters Group, the extent of which is limited to the W- NW sector. The age of this group is infe rred to be between 3.3-7 ka (Neall, 2003). Figure 3-1: Distribution of lava flows on the Mt. Taranaki main cone and Fanthams Peak. Sampled lava flows and scoria-and-ash flow units are highlighted. The thick solid line represents the Opua amphitheatre scarps. The map has been modified after Neall (2003 ) ; map details were kindly provided by J.N. Procter. Chapter 3 The Maero Eruptive Period 56 3. The Fanthams Peak Group comprises lava flows solely emitted from the satellite cone to the south of the main summit. The minimum age of these lava flows is not yet well constrained but Neall et al. (1986) a nd Downey et al. (1994) showed that the youngest flows are 3.3 ka old or less. Later work on their coverbed stratigraphy would suggest that some lava flows on the east side of Fanthams Peak could be as young as 1400 yrs B.P. (Rosenthal, 2005). 4. The Staircase Group is exposed on the easte rn to northeastern slopes overlying the Warwicks Castle Group and partially filli ng an amphitheatre in the south that was created by a sector-collapse that formed the Opua Debris Avalanch e (Neall et al., 1986). Stratigraphic and paleomagnetic evidence suggest an age of 1.7 ka for this group (McGlone et al., 1988; Downey et al., 1994). 5. The youngest lava flows form the uppermost cone and are known as the Summit Group (Stewart el al., 1996). Neall (2003) fu rther subdivided this group into Skeet Group and Summit Group, where the latter only comprises the present summit dome and the Turtle. These flows directly overlay the Staircase Group a nd predominantly fill the Opua amphitheatre in the south, but also extend to the North and East. No Summit lava flows are exposed on th e west to northwest flank. The exception may be the remnant thick lava flow or coulée (know n as the Turtle) of unknown age which is partially collapsed on its southern and upstream side. From paleomagnetic and stratigraphic evidence the Su mmit Group is younger than 1.7 ka and some flows have been estimated at 400 and 700 years old (Downey et al., 1994). The youngest lava flows could be correlatives to events of the Newall Formation (Ste wart et al., 1996). Chapter 3 The Maero Eruptive Period 57 3.2 Results 3.2.1 Stratigraphic Type and Reference Sections Representative sections around Mt. Taranaki were chosen to highlight the diversity of volcanic deposits at different sites of the mountain and within each locality (Fig. 3-2). A new chronostratigraphy for the Maero Eruptive Period and a new lithostratigraphy for the respective Maero Formation are here established with new type and reference sections being defined. At least 10 eruptive episodes occurred during the Maero Eruptive Period producing a diversity of volcan iclastic deposits around the flanks of Mt. Taranaki. In order to avoid confusion, prev iously established and used stratigraphic terms are partially substituted by new names. The Maero Eruptive Period started with the Hooker eruptive episode followed by the episodes of Te Popo, Waingongoro, Newall, Waiweranui, Puniho, Burrell, Ma ngahume, Tahurangi, and Pyramid. The formal members of the Maero Formation maintain the same nomenclature, along with the broad lithostratigraphic terms of Ash, Lapilli or Breccia. Figure 3-2: Location of type and reference sections around the edifice of Mt. Taranaki. TS-type section, P-Pembroke Road, W-Waingongoro Stream, M-Manaia Road. Numbered black squares refer to reference sections. Black diamonds mark distal locations of pyroclastic flow deposits in the area of Saunders Road – Wiremu Road – Waiweranui Stream. Chapter 3 The Maero Eruptive Period 58 3.2.1.1 Type Section of the Maero Formation A new type section for the Maero Formation is defined (Fig. 3-3). It is located in Maero Stream near the former type section of the Maero debris flows of Neall (1979) which is now obscured by colluvium. The new type section comprises a succession of pyroclastic flow deposits of at least seven eruptive episodes. It can be correlated to nearby reference sections in Pyramid Stream (Fig. 3-3; Fi g. 3-4) and Hangatahua River (Fig. 3-3). Individual exposed units can be grouped in to sequences based on the occurrence of paleosols and were correlated with the new chronosequence of the Maero Eruptive Period. A detailed description of the type section is given below (cf. Platz, 2001). The Maero Formation overlies the Oakura Te phra (<6970 ± 76 yrs B.P.; Neall, 1973), a cumulative unit of 1.4 m thickness, that repres ents cumulative accretion of fine-grained medial ash fall. A >2.5 m thick bouldery BAF deposit is exposed below this unit. Directly above the Oakura Tephra occur units of the Te Popo eruptive episode. A lower pyroclastic surge deposition sequence is overlain by a BAF deposit (Te Popo Breccia). The pyroclastic surge deposition sequence is composed of a pocketing 0-5 cm thick grey medium ash, up to a 10 cm thick layer of pale brown and pale olive brown ash with wavy bedding, up to a 10 cm layer of coarse to medium ash with rip-up clasts of fine brown ash, overlain by a 2-5 cm thick unit of pa le grey medium ash with 4 cm diameter rip-ups of fine brown ash, and capped by a 1- 5 cm pale brown ash containing charcoal fragments. The charcoal fragments were dated at 878 ± 39 yrs B.P. (Cronin et al., 2003). The pyroclastic surge sequence is overlaid by a 1.1 m thick brownish grey and strong brown and red-oxidised BAF deposit (Te Po po Breccia). It consists of approx. 70% clasts with maximum diameter of 20-30 cm set in a medium to coarse ash matrix of the same lithology. The clasts were subdivided into grey (60%), ora nge-coloured (30%), and greyish brown and pale grey (10%) a ndesite clasts. The clast-rich unit (c.70%) shows magnetic alignment indicating deposition temperature above 350 °C (after the method of Hoblitt and Kellog, 1979). The upper portion of the unit is overlain by a 4-5 cm dark brown humic-stained ash soil. The overlying sequence was deposited duri ng the Newall eruptive episode. It is composed of a pyroclastic surge and a BA F deposit (Newall Breccia). The pyroclastic surge unit overlies the paleosol, is about 4.5 cm thick, and comprises four sub-units: a c.1 mm thick fine pale brown ash passing up to coarse to medium, poorly sorted ash (c.4 cm) that normally grades into pale grey medium ash (5 mm) containing common fine charcoal fragments. The charcoal fragments were dated at 387 ± 43 yrs B.P. (Cronin et Chapter 3 The Maero Eruptive Period 59 Figure 3-3: Type section of the Maero Formation located in Maero Stream, NW sector of Mt. Taranaki. Reference sections Nos. 1 and 2 are located in Pyra mid Stream and Hangatahua River, respectively. The outcrops comprise deposits of BAFs, surges and pumice flows as well as co-ignimbrite ashes. Debris flow-, lahar-, and fluvial deposits are also exposed but not always differentiated. For further details see text and appendices. Labels Py7 etc refer to sample numbers. See also Fig. 3-4 for field photographs. Chapter 3 The Maero Eruptive Period 60 al., 2003). The uppermost unit (5 mm), above a gradational contact consists of pale brown and pale greyish-brown pocketing fine ash. The pyroclastic surge unit is overlain by a 68 cm thick BAF deposit (Newall Brecci a), which is composed of a grey ash matrix supporting c.30% lapilli-sized, grey a ndesite clasts with maximum diameters of 4 cm. The unit is massive or with weak planar fabric in places. Large charcoalised logs are aligned parallel to bedding. The lower portion of the BAF deposit is clast- poor with orange-stained clasts concentrated near th e lower erosive contact. The unit is capped by a 5 cm thick dark humic-stained soil. Overlying this soil are three individual units that were deposited during the Waiweranui eruptive episode. The lower 1.1 m thick, matrix-supported BAF deposit (Waiweranui Breccia) is composed of a grey ash (50%) and cobble-sized clasts of dominantly grey and minor pale-grey andesite. The clasts show magnetic alignment and the largest clasts occur near the top. The upper 15 cm of the un it shows pervasive dark-red oxidation of matrix and clasts. A weak planar fabric of cl ast strings is observed and charcoal logs lie parallel to this, approx. 15 cm above the ba sal contact. These were dated at 468 ± 42 yrs B.P. (Cronin et al., 2003). The lower portion of the BAF unit (30-40 cm) is characterised by the absence of coarse clasts (maximum diameter 4 cm) and a grey medium ash matrix. The lower erosive contact is sharp and wavy. The next unit above is a 30 cm thick, massive, matrix-supported lapi lli-breccia with a grey ash matrix and embedded fine lapilli clasts (up to 6 cm in diameter). Clast lit hologies include grey (60%), orange (30%), and pale grey (10%) andesite. The grey andesite clasts show magnetic alignment. The basal 5 cm is fine s-poor and clast-supported whereas the upper 20 cm is a grey-ash matrix diamicton with la pilli-sized clasts. A weak soil is developed on top of this unit. The next unit above was emplaced during the Puniho eruption episode. The deposit (Puniho Breccia) is approx. 60 cm thick, is matrix-supported and clast poor (20%) comprising grey and up to 15% orange-stain ed andesite clasts. The upper 10 cm has pervasive red oxidation (matrix and clasts). The sequence above this consists of four individual units and was deposited during the Burrell eruption episode ( AD 1655). The lower unit pinches and swells between 0 and 10 cm thick. Grey (90%) and orange-stained (up to 10%) lapilli- to cobble-sized clasts are the major constituents of this ash-poor un it. The major unit [Burrell Breccia (A)] is interpreted to represent a BAF deposit asso ciated with the explosive removal of the Chapter 3 The Maero Eruptive Period 61 Burrell dome (cf. Chapter 6). It is 1.6 m thic k and consists of a grey medium ash matrix, with up to boulder size clasts (maximum 40 cm diameter) near the top. The upper 30-40 cm show pervasive red oxidation of matrix and clasts. The unit is clast-supported in places, with grey and pale-grey andesite clasts being most common lithologies, along with minor pumice (2%) and orange-stained (3-5%) clasts. Grey andesite clasts show magnetic alignment and many of them are jig-saw fractured in situ. The basal 10 cm of the deposit is clast poor. The Burrell Breccia (A ) is overlaid by a c.15 cm thick strong brown and red-coloured medium to coarse as h with interspersed rare boulder clasts. The uppermost unit in this sequence comprises a 1- 3 cm thick pale brow n to pale brownish grey medium to fine ash, and represents a fall deposit either directly from the sub- Plinian eruption column or from ash clouds associated with pyroclastic pumice flows which are deposited further downstream. The next major unit overlying the Burrell sequen ce, is a direct correlative to Tahurangi Ash of AD 1755 (cf. Chapter 5). The Tahurangi Breccia (a) is a bouldery 1.4 m thick unit: unusual in having a matrix-supported upper portion and a clast-supported base. A total of 30% ash matrix is estimated in the upper portion with the majority of clasts being highly angular, black to dark grey andesite (98% ) and minor pale grey and orange-stained andesite. The major clast constituents show magnetic alignment. The upper 30-40 cm of the unit is oxidised to a pervasive reddish and purplish colour. The basal 5 cm is characterised by brown-stained ash matrix with fine tussock and charcoal fragments. Charcoal from this unit proved too young for radiocarbon dating (<250 yrs B.P.; Cronin et al., 2003). A weak soil is developed on top of the unit with weak weathering extending into th e upper 10 cm of deposits. The uppermost unit of the type section is a rock-avalanc he deposit (Pyramid Rock Avalanche). This deposit represents most of the missing portion of the remnant summit lava dome, which was erupted post- AD 1839 (Pyramid eruptive episode; cf. Chapter 5). The unit here is 60 cm thick and consists of brownish grey medium ash. It is matrix- supported with up to cobble-sized, subangular to subrounded predominantly grey and brownish grey andesite clasts. Rare (<5%) strong brown and orange clasts are noted. Chapter 3 The Maero Eruptive Period 62 Chapter 3 The Maero Eruptive Period 63 Figure 3-4: Field photographs of deposits from the NW sector of Mt. Taranaki. a) NW sector of Mt. Taranaki as viewed from the summit. Note the depression in vegetation caused by an avalanche (see Chapter 5). H. R. - Hangatahua River . b) exposure in the upper right Pyramid fork. Cliff section is approx. 30 m high and shows mainly lahar and hyperconcentrated flow deposits. Note Shane Cronin for scale (arrow). c) outcrop on the true left side of the Maero Stream near the intersection of Puniho and Holly Hut track. The upper unit shows the Tahurangi Block-and-Ash Flow deposit (unit a) with its upper matrix-supported and lower clast-supported zones. d) reference section No.1, lower Pyramid Stream, true right side. Only major units are labelled. Discolouration of units IV-Bb and II-B is caused by a raised iron-rich water table. Note the black bar has same height as the spate (arrow). e) close-up of c); basal portion of Tahurangi BAF (unit a) which consists of fine to medium ash. Dashed line marks the boundary between main body of the BAF and its basal portion. f) close-up of d); finely laminated and cross-bedded fine to medium ash surge deposit. Some pumice clasts (a rrows) are present. g) exposure on the true right side of Maero Stream. Major units are labelled. The rework ed section is 2.5 m thick and represents predominantly fluvial deposits. h) close-up of g); unit IV-Bb. Noteworthy are boulder trains and weak reverse grading of the lower to middle por tion. i) distal blast deposit [Newall Breccia (a)] showing its typical pocketing appearance. j) stream exposure of a BAF deposit with pervasive red coloured top portion and grey bottom portion (Waiwera nui Stream). k) in situ charcolised tree at a distal BAF exposure [Burrell Brecci a (A)] near Saunders Road. l) de gassing pipe originating at a charred log within the BAF deposit (same unit as in k). Note that the pipe branches at the centre of the photograph. Same unit as in k). Labelled units in c, d and g refer to the lithostratigraphic code. See discussion of Chapter 3 for further details. Detailed description of reference section No.1 can be found in the appendix. Photographs of f) and i) were taken by Shane Cronin. Chapter 3 The Maero Eruptive Period 64 3.2.1.2 Reference Sections of the Maero Formation Tephra fall sequences on the eastern flank of Mt. Taranaki were well described by Druce (1966) and Topping (1974) . Reference sections (No. 3 and 4) in this sector are located on Pembroke Road and near the in tersection of the upper Round the Mountain Track (RTMT) and Waingongoro Stream (Fig. 3-5). The two secti ons are only about 600 m apart, but between them they exhibit differing sedimentological features which give valuable information about transport and eruption mechanisms. Figure 3-5: Reference sections Nos. 3 and 4 located on the east flank of Mt. Taranaki. Reference site: No.3–Pembroke Road cutting; No.4–intersection of Wain- gongoro Stream and Round-The-Mountain-Track. E02-7 2 etc are sample numbers. At the Pembroke Road site, three units are distinctive: (1) the poorly sorted Puniho Ash which is preserved as a pyroclastic surge deposit, with wavy cross- and planar-bedding; (2) Newall Ash, which consists of twin ash layers; and (3) the oldest (unknown) tephra layer composed of a single layer of pumice and lithic lapilli clasts (Hooker Lapilli). All other units comprise single fall deposits of varying thicknesses. The Waingongoro section at similar altitude contains a buried Maori oven (umu), which is covered by Chapter 3 The Maero Eruptive Period 65 Newall Ash. Puniho Ash is thicker than at the previous section but shows no bedding. Waiweranui Ash is preserved only as grey lithic fragments. Newall Ash is apparently composed of two ash units, a and b; with gr ey lithic lapilli being scattered within the lower unit (a). The correlation of the two loca lities reveals a composite event record of eight erupted units. Charcoal in the umu was radiocarbon dated at 470 ± 55 yrs B.P. (Topping, 1974). Consequently, at these s ites between the deposition of Kaupokonui Tephra and Newall Ash three eruptive events are recorded, whereas since 470 yrs B.P. at least five events are shown (Fig. 3-5). Two closely spaced sections on the south flank of Mt. Taranaki form additional useful reference sections (Nos. 5 and 6) since they preserve additional eruptive units of this period. Here, near the site of the old Ma ngahume Hut, deposits from three eruptive events are preserved above the Burrell Lapilli (Fig. 3-6). A pale grey, fine to coarse ash correlates to Tahurangi Ash ( AD 1755; Druce, 1966). The unit above it constitutes a single layer of dark grey, highly crystal line lithic lapilli (P yramid Lapilli) and corresponds to the Pyramid eruption (cf. Chap ter 5). The unit below Tahurangi Ash also comprises a single layer of lithic lapilli (Mangahume Lapilli) and was erupted some time between AD 1655 and AD 1755. The tephra sequence upslope of the Old Mangahume Hut directly overlies one of th e younger summit-group lava flows implying (together with the paleomagnetic age determ inations of Downey et al., 1994) a possible emplacement during the Maero Eruptive Period . The next reference section (No. 7) is situated c.5 km NNW of the summit with in the Ahukawakawa Swamp and contains deposits of tephra fall and pyroclastic flows (Fig. 3-7). Th e lower section is composed of a >1 m thick BAF deposit which is overl ain by a pyroclastic surge unit. Four fall units are recognised which are separated by peat or peaty silt loams2 (Tahurangi Ash, Waiweranui Ash, Newall Ash, and Te Popo Ash) . The units comprise fine to medium ash; only one unit (Newall Ash; see reference section No. 3; Fig. 3-5) is composed of a pumice-rich surge deposit over- and u nderlain by a 2 cm thick fine ash. West of the summit, two sec tions are exposed, at medial 3 (uphill of Kahui Hut; reference section No. 8) and distal 3 (Parihaka Road; reference section No. 9) locations 2 Brownish clay to silty loams of varying thickness are derived from medial volcanic ash fall and/or colluvium. 3 Proximal, medial and distal reach es describe distances from the present summit of Mt. Taranaki. The term proximal is used for areas <3 km from source, (i.e. in the range for ballistics), medial for a radius of 3-10 km from source (marked by the National Park boundary and also the normal limit for most pyroclastic flow and surge units), and distal at a radius >10 km, which only the rarer larger event reached. Chapter 3 The Maero Eruptive Period 66 (Fig. 3-7). Deposit types contrast with th e previous sections since they were predominantly emplaced by pyroclastic density currents. The reference section exposed near Kahui Hut shows from bottom to top a se quence of three pyroclastic surge deposits followed by an alternating BAF-surge-BAF deposition sequence. In the upper portion of the section there are two ash layers, presum ably correlatives to Newall Ash. Another ash layer (Tahurangi Ash) occurs at the t op of the exposure which is overlain by the present-day topsoil. Figure 3-6: Reference sections Nos. 5 and 6 located on the south flank of Mt. Taranaki. Reference site: No.5–at the site of the Old Mangahume Hut; No.6–near the Old Mangahume Hut, upslope of No.5. Labels E03-35 etc are sample numbers. The other reference section (No. 9) in this se ctor is located 14 km west of the summit and is a lateral blast-type deposit (Fig. 3-7) . It comprises a poorly sorted, pocketing unit of firm grey and pale greyish-brown, medi um to coarse ash with banding on a 1 cm scale. Fine lapilli lithic clasts and dense a nd vesicular ash occurs within two-thirds of the layers. Grey, blade-like a ngular dense clasts up to 4 cm long are also common. The correlation of this deposit to sections on the mountain is difficult due to an absence of other related beds, but it appears most likely to represent a correlative of a pyroclastic surge deposit sequence at the lower Ka hui section (Waingongoro Ash). However, stratigraphic relation remains unknown. Chapter 3 The Maero Eruptive Period 67 Figure 3-7: Reference sections Nos. 7, 8 and 9 located on the west and north sector of Mt. Taranaki. Reference site: No.7–Ahukawakawa Swamp; No.8–Marupakoko Stream near Kahui Hut; No.9–Parihaka Road cutting. Labels E02-4 etc refer to sample numbers. Two locations in Pyramid Stream and Hanga tahua River near the type section are defined as medial reference sites (Nos. 1 and 2) on the NW sector since they expand the record from the type section by two further units (Fig. 3-3). Distal pyroclastic flow deposits in the NW sector are found outside the National Park in the area defined by Saunders Road – Wiremu Road – Waiweranui Stream (Fig. 3-2). Although other isolated exposures of pyroclastic flow deposits are known (cf. Platz, 2001), their correlation is difficult based on stratigraphic context alone. Even radiocarbon dating of two distal deposits did not elucidate their stratigraphic position due to their being more than one potential correlative within the standard error range of the radiocarbon method. Chapter 3 The Maero Eruptive Period 68 3.2.2 Block-and-Ash Flow Deposits 3.2.2.1 Distribution and Flow Paths Block-and-Ash Flows (BAFs) generated by collapses from lava domes during the Maero Eruptive Period were predominantly directed towards the NW sector. Currently a continuous crater wall extends around the NE -E-SW sectors. The former W-SW crater rim was composed of a similar succession of at least four lava flows as exposed at the entrance of Okahu Gorge (Fig. 3-8). The distribution of BAF deposits allow the identification of three crater exit areas a nd flow paths for BAFs during Maero Eruptive Period. These are, in increasing order of im portance the N, the SW (Okahu Gorge), and the W-NW sectors (Fig. 3-9, Fig. 3-10). Figure 3-8: Crater rim stratigraphy as exposed on its SW side at the entrance of Okahu Gorge. Four lava flows are identifiable with the youngest flow (4) being known as South Flow. North sector The northern exit area includes the tribut aries of Peters, Minirapa, and Kokowai Streams (Fig. 3-10). Exposures in these show BAF deposits and re lated debris flow deposits of Maero age, however , their detailed st ratigraphy in relation to other units in the Maero catchment cannot be well constr ained. In addition, coverbed sequences on ridges and lava flows contain numerous tephra layers; some of them are interpreted to be pyroclastic surge deposits related to BAFs. Chapter 3 The Maero Eruptive Period 69 Figure 3-9: Composite photograph of the upper NW flank of Mt. Taranaki showing named and described morphological features. Southwest sector The present-day Okahu Gorge begins at a cliff on the SW side of the crater and deepens and widens downslope (Fi g. 3-10). Many of the Maero Period dome-collapse related rockfalls and BAFs travelled this path. At about 1700 m altitude, the gorge turns sharply from W to SW. At this location BAFs either fu lly or partly flowed straight ahead (W) to enter the upper Maero/Turehu catchment, or th ey followed the SW course of the gorge. Those BAFs travelling W out of the gorge sp lit further around Turehu Hill (remnants of two lava flows) and entered the upper catchme nts of one or more of the Turehu, Maero, Marupakoko or Kapoaiaia Streams. Puni ho Hill does not appear to have been overtopped by BAFs. Farther downstream, in the Okahu Stream valley (at c.700 m altitude) some BAFs overtopped valley sides over a length of c.700 m and travelled up to 3 km through the forest. This deposition pa ttern is also indicated by vegetation zones seen in aerial photographs. Northwest sector This was the major travel path for BAFs during the Ma ero Eruptive Period. The upper NW flank forms a natural slightly concave sh aped chute which is currently bordered by the Turtle to the north and by a low ridge to the south (Fig. 3-9). At the bottom end of the chute at c.1700 m altitude, BAFs were at times split to the NNW (W of the Big Pyramid) into the Upson Stream-Skinner Hill area, but mainly descended a steep bluff (Fig. 3-9). At the base of the bluff (1500 m a ltitude), local slope drops to 17°, from the typical 35° upslope. Block-a nd-Ash Flows descending the bl uff were deflected off the “Big Pyramid” hill, traversed a broad low -relief surface before being channelled into Chapter 3 The Maero Eruptive Period 70 areas occupied by the present Pyramid, Unnamed, Turehu or Maero Streams. The bulk of passing BAFs were channelled into the Ma ero Stream area where they were confined on the southwest by a lava ridge. Only rarely could the basal dense avalanche parts of the BAFs overcome the lava barrier to form small, steeply inclined lobes. Block-and- Ash-Flow related surges, by c ontrast, travelled farther we stward beyond the lava ridge. Figure 3-10: Flow paths and distribution of BAF and surge deposits on the N to W sectors. Crater exit areas: 1-north sector, 2-northwes t sector, 3-southwest sector. A fan containing the volumetric bulk of BAF deposits from the Maero Eruptive Period is bordered by Pyramid Stream, Maero Stream and Hangatahua River. This region as well as the area farther westward towards Kahui Hut, is characterised by the depression of current altitudinal vegetation zones by up to 300 m. The best exposures of these BAF deposits are found in Maero Stream where i ndividual units can be traced over several km. Two breaks in slope occur at the 1000 m and 600 m contours, where the overall slope drops to 9° and 3°, respectively. The present day Maero Stream valley has an immature box-shaped form, although from deposits preserved in the sides it appears that the valley may have remained in the same general position over the last 1000 years. Hence, there are sub tle variations in the paths Chapter 3 The Maero Eruptive Period 71 of individual flows. Burrell-ag ed and older deposits (i.e. pre- AD 1655) are exposed in the top section of a c.20 m high cli ff at the true left side of the Hangatahua River, indicating the BAF travel path at that time was closer to the true left valley margin than the present river location (Fig. 3-10). The youthful forest ve getation of this area is still recognisable on aerial photographs. Initially, BAFs were di verted in a left-hand bend in the lower Maero Stream, before flowing, either parallel to the Hanga tahua River, or entering the Waiweranui Stream at c.540 m altitude. Be yond the present National Park boundary, at the 400 m contour line, BAFs st alled in the relatively flat area defined by Waiweranui Stream – Saunders Road – Wiremu Road. Po st Burrell-aged BAFs appear to have travelled down the Maero Stream and into the Hangatahua River valley. Deposits of these units are exposed in the Blue Rata Re serve, north of the Hangatahua River at a distance of 13 km from source. The distribution of ash-cloud surges is generally bound to the travel paths and deposition areas of BAFs (Watanabe et al., 1999). This meant that ash-cloud surges were restricted to the W to NNE sector of Mt. Taranaki. Under some conditions, for example sharp valley bends, th e turbulent ash cloud of a BAF is able to decouple from its basal avalanche (Fisher, 1995). A characte ristic triangular region on the NW flanks shows evidence of being repeatedly devastated by ash-cloud surges, despite being protected by topography from the densest portion of BAFs (Fig. 3-10). The ash-cloud surges of Maero BAFs could overcome th e morphological barrier and repeatedly destroy the vegetation in this region. 3.2.3 Lava Flows Selected lava flows of the upper cone of Mt . Taranaki (Fig. 3-1) were studied, because cover bed stratigraphy and paleomagnetic studies suggested their emplacement at c.400 and 700 yrs ago (Downey et al., 1994), th us within the Maero Eruptive Period. Consequently, seven lava flows were sampled – Minirapa and Lizard lava flows to the north, two unknown lava flows to the west, th e Turtle to the northwest, and two lava flows on the south flank within th e Opua amphitheatre (Table 3-4). The Lizard lava flow (short Lizard) originates on the north sector of the crater where the crater wall is collapsed. It splits at its uppermost portion above 2200 m altitude. The main portion flowed within the valley of the main tributary of Kokowai Stream down to an altitude of 1260 m, whereas a smaller l obe followed the Minirapa Stream valley. The Chapter 3 The Maero Eruptive Period 72 cover bed sequence on the Lizard at its terminat ion is 50 cm thick and consists of a c.25 cm thick dark, greyish brown silt loam with increasing content of angular lava pebbles and cobbles with depth. It is overlain by two grey, fine to medium and medium ash layers of 2 and 3 cm thickness. The top unit is a dark brown sandy soil of 20 cm thickness. The current valley-confined flow path of the Lizard, along with the coverage of only two ash layers would suggest a relatively young emplacement age. The Minirapa lava flow (short Minirapa) has a particularly distinctive-appearing. It is a low-viscosity flow that also originated from the north sector of the crater and followed the current Minirapa Stream valley (Fig. 3-1) . It overlies the Lizard flow. The lava flow came to rest at 1200 m and formed a lobe up to 155 m wide: at its origin the flow is c.80 m wide and it narrows downhill to between 40 and 50 m. In contrast to the Lizard, cover beds with a total thickness of at least 1 m are preserved on the Minirapa flow. The oldest unit is a diamicton containing greyish-brown pumice and grey and red lithics set in a sandy matrix, which appears most likely to be a pyroclastic flow deposit. This pumice-b earing unit may correspond to the youngest known pre-Maero pumice-producing even t, the c.1350 ± 150 yrs B.P. Kaupokonui Tephra (McGlone et al., 1988), however, th is unit is only known from the S and E flanks. Recent studies of tephra on the NE fl anks have confirmed that the Kaupokonui is not present, but a younger pumice tephra and pyr oclastic flow deposit does occur (M.B. Turner, unpubl. data). While not directly dated, this unit can be geochemically correlated to dated units within a fall sequenc e in Lake Rotokare, c.35 km SE of Mt. Taranaki. Ages of either 777 ± 36 or 885 ± 36 yrs B.P. (M.B. Turner, unpubl. data) relate to this Rotokare sample which in tu rn give a minimum age for the Minirapa flow. This implies that both Minirapa and Lizard flows could have been erupted only within the early Maero Eruptive Period prior to c .800 yrs B.P. Although Neall (2003) assigned the Lizard an age between 1.4 and 0.5 ka, the Mi nirapa was allocated to the Staircase Group (3.3-1.4 ka). In Stewart et al. (1996) both lava flows are classed as Summit Group lava flows, less than 0.7 ka old. The Turtle is an up to c.15 m thick erode d lava coulée on the upper NW flank. Thick columnar joints are exposed at its southe rn and northern collapsed faces and a thin carapace occurs on its top. The only cover beds found on the flow surface, including within deep cracks, include rare dense grey, jig-saw jointed bombs a nd blocks (up to 20 cm in diameter). The absence of, for ex ample, Burrell Lapilli pumice clasts on its surface could be evidence for post- AD 1655 emplacement, although its steep upper Chapter 3 The Maero Eruptive Period 73 Chapter 3 The Maero Eruptive Period 74 surface and the high altitude may have prevented the preservation of this and earlier tephra deposits. The distinct shape of the Turtle is that of a lava coulée and its emplacement was assumed to have been part of the former summit lava dome (Cronin, 1991). Other studies allocated the Turtle an age of 0.7 ka or younger (Stewart et al., 1996; Neall, 2003). Two further (previously undescribed) lava fl ows are also exposed in the upper central fork of Pyramid Stream between 1100 and 1200 m (Fig. 3-1). Loose rockfall or BAF deposits are plastered onto and overlie the lava flows in places. The location and elevation of these lava flows suggests a stratigraphic affiliation to the Warwicks Castle Group, although, after Neall (2003) these flow s would be assigned to Peters Group. On the south flank between Bobs Ridge a nd Fanthams Peak two lava flows were sampled (Fig. 3-1). In addition, two distin ct scoriaceous, black diamictons were observed which are interpreted to be proximal scoria-and-ash flow deposits. The deposits are poorly sorted and contain monolithologic scoria lapilli and blocks/bombs up to 25 cm in diameter that are set in medium to coarse black ash matrix. They are exposed on Bobs Ridge and east of it within the Opua amphitheatre. The oldest sampled lava flow corresponds to unit 2 of the crater rim (see Fig. 3-8). It is overlain by the South Flow (lava flow unit 4; Fig. 3-8), which at 2100 m altitude forms two small bluffs before bifurcating into two lobes that extend down to c.1100 m. On the upper flanks only Burrell Lapilli-aged pumice pyr oclastic flow deposits directly overlie the South Flow. This (and the dates of Down ey et al., 1994) imply that South Flow extruded during the Maero Eruptiv e Period, possibly only 400 yrs ago. The scoria-and-ash flow deposit that lies we st of South Flow (s till within the Opua amphitheatre; Fig. 3-1) is al so overlain by Burrell-aged py roclastic flow deposits. The stratigraphic relationship of the scoria-and-ash flow unit on B obs Ridge is more difficult to interpret. It is overlain by only a thin (c.20 cm) poorly sorted lithic diamicton. Its exposed location on a ridge a nd its thickness of more than 20 cm in relation to its distance of about 900 m to the crater rim may suggest a larger eruptive event, perhaps one of the larger Stair case lava flow eruptions. Chapter 3 The Maero Eruptive Period 75 3.2.4 Glass Chemistry Various sites around Mt. Taranaki described by Platz (2001), S. Cronin (unpubl. data), and this study were selected for glass compositional studies in order to chemically characterise the diversity of Maero-aged deposition types and to test the correlation of established units (Fig. 3-2). The dataset of glass chemical compositions comprises 68 samples with a total number of 1016 analyses. Contaminated or hybrid analyses (i.e. EMP analyses comprising mixed proportions of glass and mineral ph ases) were identifie d by comparing intra- variations of two or more glass shard analyses and by applying the glass evaluation procedure presented in Chapter 4. A broa d range in major element composition was observed: SiO 2 contents vary between 57.2 to 76.2 wt.% and K 2 O from 2.3 to 7.2 wt.%. A near Gaussian distributi on curve is observed for SiO 2 values between c.64-77 wt.%. with the mode at 68-69 wt.% SiO 2 and 5.5-6 wt.% K 2 O (Fig. 3-11). Howe ver, there are also two weak secondary modes in the low- silica range. These modes should also occur in K 2 O abundances if potassium behaves as an incompatible element, although the K 2 O distribution curve does not have such distinct modes, and instead the distribution is skewed toward the low-K 2 O range (Fig. 3-11). Figure 3-11: Histogram of all glass chemical analyses for SiO 2 (a) and K 2 O (b). Chapter 3 The Maero Eruptive Period 76 Intra-sample variations can be larg e, for example up to 11.2 wt.% in SiO 2 , although the majority are between 1-6 wt.% SiO 2 . The largest intra-sample variations are observed for bimodal samples whereas the lowest often occur within tephra samples with rhyolitic glass compositions. Standard st atistical methods (e.g., using standard deviations and 95% confiden ce intervals) as well as can onical discriminant function analysis (DFA) were used to group individual units to single eruptive events (cf. Shane and Froggatt, 1994; Cronin et al., 1996; 1997). Although DFA used seven independent variables, grouping proved diffi cult since the variance within samples is too strong. The probability of samples being classified into unique groups were in most cases <<75% (Platz et al., 2004). This means that the gl ass chemistry of individual units could only rarely be uniquely identifi ed using DFA and other statis tical methods. However, glass chemical characteristics (e.g. bimodality, andesitic vs. rhyolitic composition) of individual units were of support for field- based correlation of tephra and pyroclastic flow deposits. 3.2.4.1 Special Characteristics of Some Erupted Units Intra-sample variations can be large and can have various reasons. Firstly, two or more glass populations (andesitic to rhyolitic) can be present w ithin one tephra layer (Fig. 3-12; see also WW19 in Fig. 3-13). The compositional differences are often also reflected in different glass shard shapes (F ig. 3-12). However, even with similar glass compositions, the texture of glass shards also varied strongly. Even without distinctive populations being present, analys es on the same glass shard in some units showed silica variations of several wt.% (i.e. WW6: Δ SiO 2 = 3.5 wt.%; SR41: Δ SiO 2 = 3.5 wt.%; SR40: Δ SiO 2 = 2.8 wt.%). In addition, one of the units was contaminated by foreign glass shards (E03-35; Fig. 3-12). This tephra la yer (of Burrell Lapilli) located at the old Mangahume Hut on the southern flank contains rhyolite glas s shards with a chemistry typical of Taupo volcano (Taupo Volcanic Zone). These exotic shard compositions are a good match for the AD 186 Taupo tephra or older Holocene Taupo tephras (P.A.R. Shane, pers. comm., 2007). Howe ver, it is impossible that these glass shards represent primary fall, and hence they either represen t reworked Taupo tephra (unlikely since this unit was not deposited in this direction), or rare Taranaki glasses that coincidentally have the same compositions in major elements as Taupo. Chapter 3 The Maero Eruptive Period 77 Clasts of scoria-and-ash flow deposits, bot h within the Opua amphitheatre and on Bobs Ridge (Table 3-4), have glass compositions that contrast with dome-related and sub- Plinian eruptions of the Maero Er uptive Period. These have lower SiO 2 and higher TiO 2 , MgO and partially higher FeO abundances (Fig. 3-14). Some tephra layers contain individual glass shard(s) that are similar in composition to scoria-and-ash flow producing eruptions (e.g. shards in samples E03-37, E02-28 , E02-40 , E02-41 , E02- 74/75/76, Py20, E03-78; Fig. 3-14). Figure 3-12: Backscatter-electron microscopy images of i ndividual glass shards. Different glass shard shapes within individual samples are illustrated (top: vesicular, partially deformed; bottom: dense and angular). Label in images gives an alysis number and approximate beam location (black dot). Grey=glass; light grey=minerals; black=epoxy. a-b) Tahurangi Ash sample (T04-98 ) with chemically homogenous glass shards but of different texture; a) deformed vesicles, b) dense angular. Note that the presence of large minerals (bottom) may alter vesicle distribution. c-d) Newall Ash sample (E02-7 5) with distinct shard textures; glass chemistry of shard (d) is similar to those of scoria-and-ash flow glass shard (c). Note regular to irregular vesicles in c). e-f) Burrell Lap illi sample (E03-3 5) with Taranaki glass shard (f) and Taupo volcano-derived glass shard (e), which shows a deformed vesicle. 3.2.4.2 Correlation of Tephra and Pyroclastic Flow Deposits Deposits of the Burrell Lapilli eruption are used to illustrate how glass chemistry of various deposition types can be used as a back up tool for field-based correlation. Burrell-related fall, pumice flow, BAF, and pyroclastic surge deposits are discussed in Chapter 6; their range in glass co mposition is presented in Fig. 3-13. Chapter 3 The Maero Eruptive Period 78 Figure 3-13: Glass chemistry of Burre ll-aged deposits. The SiO 2 vs. K 2 O diagram shows a general positive correlation. Highest mean silica and potassium contents are observed for BAF and surge deposits. The crosses represent the 95% confidence interval of the sample mean. The tephra sample E03-35 is bimodal containing glass shards with a signature similar to Taupo volcano (or atypical of Taranaki). If foreign glass shard analyses are exclude d the sample mean is located within the field of BAF/surges (=E03-3 5’ ). Also incl uded are sample means of a surge deposit pre-dating the Maero Eruptive Period showing a rhyolitic glass chemistry. It is noted that one sample (WW19) shows large variations and a bimodal sample population. Individual BAF and surge deposits of the NW sector show compositional heterogeneities (e.g. Py31), however, their mean values are within 1.8 wt.% SiO 2 and 0.5 wt.% K 2 O. Here the glass chemistry of medial as well as distal deposits supports the correlation based on field observations. Gla ss chemistry is, however, not always an effective tool as shown in two examples where two closely-spaced outcrops in Waiweranui Stream have comparable field characteristics, but their glass chemistry differs (WW19-WW8, WW 20- WW6; Fig. 3-13). Correlation of tephra deposits across different sectors of Mt. Taranaki is only partially possible, mainly due to the small-volume a nd restricted distribution of thin fall and surge deposits. Hence, sequences of fall and pyroclastic surge units within soils around Mt. Taranaki typically consist of thin fine to medium ash layers that are visually indistinguishable and highly variably preserved on scales of tens of metres. Correlation of some units using glass chemistry is possible, e.g. the youngest preserved tephra deposit exposed near Kahui Hut and in Ahukawakawa Swamp, both represent fall deposits of the Tahurangi Ash erupted in AD 1755. Also correlated with this fall event Chapter 3 The Maero Eruptive Period 79 are surge deposits below a BAF unit of the sa me episode (Fig. 3-15). Similarly, tephra layers exposed on Pembroke Road and Ahukawakawa Swamp are also correlated to a BAF-related surge deposit in the Maero Str eam area dated at 878 ± 39 yrs B.P. (Fig. 3-15). This correlation genera tes another important link between the BAF-dominated sequences of the NW flanks, with the fa ll-dominated sequences on northern, eastern, and southern flanks. Figure 3-14: Comparison of mean sample values of scoria- and-ash flow units (T04-53, T04-56) to pyroclastic pumice flow deposits of the Burrell episode (units 1-3) and Puniho Ash, and other Maero deposits (crosses). Individual glass shards (small black squares) within deposits, other than T04-53 and T04-56, that have compositions similar to scoria-and-ash flows. a) SiO 2 vs. K 2 O, b) CaO vs. FeO. Chapter 3 The Maero Eruptive Period 80 Figure 3-15: Correlation of individual tephra and pyroclastic flow units based on field studies and glass chemistry (SiO 2 vs. K 2 O). a) Waingongoro and Waiweranui episodes, b) Newall and Puniho episodes, c) Tahurangi and Te Popo episodes. Chapter 3 The Maero Eruptive Period 81 3.3 Discussion 3.3.1 Eruption Frequency of the Maero Eruptive Period Although not each individual tephra layer or pyroclastic flow deposit is uniquely characterised by its glass chemical composition, this met hod however greatly assisted the correlation of single eruptive units across large distances. Despite the broad glass compositional range within individual samples, tephra fall layers and pyroclastic flow and surge deposits can be correlated if field observations and stratigraphic positions within sequences are included. On this basi s, tephra sequences on the northern, eastern, and southern flanks are first discussed and are then compared to pyroclastic flow and surge deposits of the western to northwestern sector of Mt. Taranaki. On the northern, eastern and southern volcano flanks, 10 eruptive events are identified during the Maero Eruptive Period. These are represented by specific tephra units that are clearly separated by nine time breaks in the form of paleosols and/or variably clay/silt loams. The latest eruptive event is recorded within the present-day soil. The beginning of the Maero Eruptive Period is herein defined by the first occurrence of a tephra unit stratigraphically above the Kaupokonui Tephr a (on the SE flank). At reference section 3 on Pembroke Road (east sector), a single 1 cm-thick tephra layer (Hooker Lapilli) occurs 3-4 cm above Kaupokonui Tephra within a greyish-brown, weathered, fine medial ash. This tephra layer is considered representative for the first eruption of the Maero Eruptive Period. Th e next youngest tephr a, Te Popo Ash, separated by 7 cm of greyish brown medial ash was geochemically correlated (see above) to a pyroclastic surge deposit in Ma ero Stream dated at 878 ± 39 yrs B.P. (Fig. 3-15). Upward toward the visually di stinctive Burrell Lapilli pumice fall of AD 1655, four intervening separate eruptive events (Waingongoro Ash, Newall Ash, Waiweranui Ash, and Puniho Ash) are preser ved as ash fall units. In on e instance, two consecutive tephra layers (Newall Ash [b ], (E02-74) and Newall Ash [c], (E02- 7 5 ) ) have no intervening medial ash and hence are interpreted as two eruptive events belonging to a single episode (i.e. Newall epis ode). Above the Burrell Lapilli fall unit at this location, only one tephra is observed, the up to 10 cm-thick Tahurangi Ash erupted in AD 1755. At a few other locations on the southern to southwestern sector of Mt. Taranaki, two layers comprising single layers of dense lithic lapilli occur above and below the Tahurangi ash. The youngest layer is attributed to the Pyramid eruptive episode (cf. Chapter 5). Chapter 3 The Maero Eruptive Period 82 The succession of tephra layers and their allocation into the stratigraphic sequence is summarised in Table 3-5. The dating of indi vidual events has been approximated by combining results of botanical studies with thicknesses of paleosols and medial ash. The break between deposits of the previous memb ers of Burrell Ash and Burrell Lapilli is marked by a 3 cm thick layer of medial ash with evidence of soil development (a paleosol), interpreted to represent a period of c.70 yrs (Lees and Neall, 1993). A similar period of repose is thought to occur between the previous members of Burrell Ash and Waiweranui Ash due to a 3 cm thick medial ash paleosol (Lees and Neall, 1993). Using a similar approach, the thicknesses of paleos ols/medial ash accumulations at Pembroke Road can be used to give a rough idea of the period of repose between events. These estimates correlate well to those from botanical studies, at least for the last six eruptions. By adding the approximat ed periods of repose an age of c.AD 860 results for Kaupokonui Tephra, which is in broad agreem ents to the mid point of calibrated calendar age range of AD 400-105 0 (Table 3-5). The correlation of these fall and pyroclastic surge units around the volcano to BAF and surge units on the main NW depositional axis is more complicated. This is partly due to their restricted distribution and tight confinement near, or within valleys. While closely spaced outcrops can be correlated successfully, the correlation of BA Fs over catchment boundaries and across distances of >400 m ca n be impossible, since their deposit characteristics may vary considerably, especi ally in distal reaches. In addition, soil development and accumulation of medial ash between eruptive events appears to have occurred much more slowly than on the eastern flanks and these horizons may be easily eroded by the violent and erosive BAFs. Hence, periods of repose may not always be recognised in these coarse BAF sequences. By consideration of field observations, glass chemical compositions and available 14 C dates, seven eruptive episodes can be identified from deposits in the medial area on the NW sector of Mt. Taranaki (cf. Maero type section). Deposits from four of these are recognised at distal reaches. How the emplacement of lava flow s fits into this sequence are not yet considered in this stratigraphy. Their effusion is probably not accompanied by significant ash fall. The generation of scoria-and-ash flows may be associated with lava flow eruptions or represent single eruptive events. ‘Foreign’ glass shards we re found within blast/surge and tephra fall deposits (see above), which show silica compositions intermediate to the scoria-and-ash flow produci ng events and lava dome-pr oducing eruptions. Individual Chapter 3 The Maero Eruptive Period 83 Chapter 3 The Maero Eruptive Period Chapter 3 The Maero Eruptive Period 85 analyses were compared and arranged into two groups4 . The tephra marker bed at 878 ± 39 yrs B.P. (Te Popo Ash) contains group 1 ‘foreign’ shards whereas the next two younger units at the same location (Waingongor o Ash and Newall Ash) contain group 2 glass shards. If those analysed glass shards are juvenile and their chemistry is representative for their respective eruptions, then at least two additional eruptions must have occurred within the Maero Eruptive Period. The first event occurred at or just prior to 878 ± 39 yrs B.P. (i.e. Te Popo eruptive episode), the second one at or preceding the Waingongoro eruptive episode. Those ‘foreign’ shards within the next younger unit (Newall Ash c; sample E02-74) do not necessar ily need to represent a separate effusive, lava flow-producing event and were likely pi cked up by this event. The two scoria-and- ash flow deposits on the upper south flank as well as the So uth Flow cannot be clearly integrated into the stratigraphic sequence, but field observations and glass chemical studies indicate that only the deposits within the Opua amphitheatre are related to the Maero Eruptive Period, possibly a correl ative to the Waingongoro eruptive episode (post 878 ± 39 yrs B.P.). Since at high alt itudes above the scrub line the scoria-and-ash flow deposits and the South Flow are only ove rlain by Burrell-aged deposits, lava flow emplacement may have occurred just prior to the sub-Plinian event [Burrell Breccia (B)]. By combining the frequency of eruptions in each sector, a tota l of at least 10, and possibly 12-14, events result. Ten tephra layers occur in the eastern to southern sector, which in parts can be correlated to the western to northwestern sector. In addition, at least two lava flow-producing eruptions a nd possibly two further eruptions producing scoria-and-ash flows may have occurred. This interpretation is based on the assumption that each dome-forming and explosive erupti on produced a tephra layer(s), either as a single lithic layer (e.g. Pyramid Lapilli), as th in, and layered fine to medium ash close to source (e.g. Puniho Ash) or as widespread pumice tephra layer (Burrell Lapilli). Although 12-14 individual events could be identified, it is currently not known whether transitions occurred between lava flow- and lava dome-forming eruptions. Also, deposits of small events such as the Pyramid eruption may have not been preserved or found. 4 Note the grouping is based only on a total of 19 analyses. Chapter 3 The Maero Eruptive Period 86 3.3.2 Comparison to the Previously Known Stratigraphy The new versus old stratigraphy of the Maero Erup tive Period (Table 3-2) coincidentally implies no difference in the total number of events, if the minimum number of 12 eruptive events is used. Th ere was a lack of knowledge of volcanic phenomena at the time of Grant-Taylor ( 1964) and Druce (1966). Hence, pyroclastic flow and surge deposits with/w ithout carbonised wood were often described as lapilli beds and regarded as fall deposits. Consequent ly, some of those lapilli units only found in the NW sector were designated as indivi dual eruptive events. The best examples are the previously known Puniho Lapi lli 1 and 2 (see Table 3-1; Table 3-2) with the type section located within the triangular area NW of Puniho Hill (see Fig. 3-10; Druce, 1966). This type section is reinterpreted as a succession of pyroclastic surge deposits, which are predominantly related to BAFs. For instance, Puniho Lapilli 2 of Druce (1966) is reinterpreted to be related to a BAF deposit of the Ta hurangi eruption (see Chapter 5). As described above, this tria ngular area was only inundated by pyroclastic surges due to its morphological characteristics. Therefore, this area is unsuitable as type locality for individual eruptions. It is difficult to correlate individual pyroclastic surge units found in this area to BAF deposits. Info rmation such as componentry, glass shard morphology and composition and/or radiocarbon dating are needed. Similar approaches led to the recogniti on of Newall and Waiweranui Lapilli as individual beds and hence eruptions, a lthough there was only limited support by the tephrostratigraphic sequence known from the eas t flank of Mt. Tarana ki. Type sections for those units again refer to pyroclastic surge and BA F deposits on the NW sector. Identification and correlation of deposit sets of very similar appearance generated an artificially high eruption frequency, e.g. P uniho Lapilli 1 and 2. As a consequence of Druce’s (1966) established stratigraphy, eruption names were maintained. For example, Topping (1974) described a tephra sequence on the east flank of Mt. Taranaki (see Fig. 3-5) and correlated individua l layers to those of Druc e (1966). According to his correlations, a 0-10 mm thick layer of grey lithic fragments was identified as Newall or Waiweranui Lapilli. Yet 5 cm below it a si milar lithic layer appears but was ignored in his interpretation (Topping, 1974). In the episode-by-episode appr oach of Cronin et al. (2003), the results of botanical studies (cf. McGlone et al., 1988; Lees and Neall, 1993) were not considered leading to inconsistencies in their reconstruction of the eruptive sequence. For example, Burrell Chapter 3 The Maero Eruptive Period 87 Lapilli and Burrell Ash were grouped into one eruptive episode (see Table 3-3) although it was earlier interpreted that a period of c.70 years separates the events (see Table 3-2). Paleomagnetic studies suggested the emplacement of lava flows on the NE sector at 700 and 400 yrs B.P. (Downey et al., 1994). These may be imprecise due to unavailability of a paleosecular variation curve available for New Zealand at that time (Downey et al., 1994), although it has sin ce been found that the relevant curve is very similar to the one used in the analysis (R .B. Stewart, pers. comm., 2007). There may also be inconsistencies in the lava flow stratigraphy used in their studies because their classified 400/700 yrs B.P. lava flows of the NE flank are older than their Summit lava flow group. In addition, it was interpreted by Cr onin (1991) that the youngest lava flow of the present intact crater rim is located in the south. Despite this, some lava flows were considered as Newall-equivalent eruptives (c f. Stewart et al., 1996; Cronin et al., 2003). Based on the current study, only the South Flow and its underlying flow of the southern crater rim (i.e. units 1 and 2; see Fig. 3-8) appear to have been erupted with certainty during the Maero Eruptive Period. Further, no data are av ailable which may indicate lava flow emplacement on the western flank in early Maero times because the western crater rim has been removed. The Turtle is the only remnant lava preserved here, likely representing part of a former coulée. 3.3.3 Tephrostratigraphy of the Maero Eruptive Period During the Maero Eruptive Peri od, at least 10 eruptive epis odes occurred that produced tephra fallout and/or pyroclastic BAF and surge deposits. Glass shard chemistry indicates that probably two eruptions producing scoria-and-ash flows may have occurred prior to the Newall eruptive ep isode. The tephrostratigraphy of the Maero Eruptive Period is presented in Table 3-5 with individual eruptive episodes. Former names have been retained where appropriate. From the observed 10 tephra layers found on th e eastern to southern flanks of Mt. Taranaki seven can be correlated to the NW sector. The oldest eruption of the Maero Eruptive Period, the Hooker 5 eruptive episode, is only preserved as a single incomplete layer of pumice and lithic fragments (Hooker Lapilli) on the eastern flank. A possible 5 Named after L.O. Hooker (1873 -19 50 ) ; this lapilli layer was also found on the track towards Hooker Shelter. Chapter 3 The Maero Eruptive Period 88 correlative on the medial NW sector was not found, however, due to the limited extent and volume of the fall, surge and/or BAF depos its associated with this event may have been buried or eroded by subsequent events. Te Popo 6 Ash of the Te Popo eruptive episode is the most important marker bed for the early Maero Eruptive Period since it can be correlated to pyroclastic surge and BAF deposits on th e NW sector dated at 878 ± 39 yrs B.P. These deposits directly overlie a >1 m thick brown clay loam soil, its upper portion dated at 1975 ± 50 yrs B.P. (Neall, 1979). The next youngest Waingongoro 7 eruptive episode produced a tephra layer (Waingongoro Ash) that was dated at 470 ± 55 yrs B.P. (Topping, 1974). The fine ash has been found on the eastern flank only in medial reaches; no correlative deposits were identified on the NW flank. The Newall eruptive episode is of importance since it initiates the period of high magma discharge in this and subsequent eruptions . On the eastern flank this eruption is recognised as two medial fine to medium ash layers [Newall Ash (b) and (c)] partially overlying a single lithic layer [Newall Ash (a)]. On the NW sector a blast deposit [Newall Breccia (a)] is found up to 14 km from the summit. This event is likely to be the trigger for the Newall eruptive episode during which a major pa rt of the western crater rim appears to have been removed. One major BAF sequence [Newall Breccia (b)] and several surge deposits were identifie d at medial to distal locations on the NW sector. The two tephra layers preserved on the east flank could be related to any of the event deposits of the NW and since no no ticeable time gap is recorded they are considered as one eruptive episode. Radio carbon dates suggest a calendar age between AD 1270-1420 for the onset of the Newall eruptive episode. The following Waiweranui eruptive episode is often represented by a bedded fine to medium ashfall unit (Waiweranu i Ash) on the eastern flanks of Mt. Taranaki. The NW equivalent deposits comprise one major BA F deposit (Waiweranui Br eccia) and several surge deposits preserved at medial to distal reaches. Although a brief period of repose between Newall Lapilli and Waiweranui Lapi lli was indicated by Druce (1966), a much longer time break (90 yrs?) is interprete d from accumulation of medial ash at the Pembroke Road reference section (east sector). According to 14 C dates, an eruption time for Waiweranui episode between AD 1390-1520 is suggested. About 70 yrs later at c. AD 1585 ( 14 C date: AD 1440-1660) the Puniho eruptive episode began; the observed deposit on the east flank comprises me dial fine to medium ash 6 Named after Te Popo Stream on the east flank of Mt. Taranaki. 7 Named after Waingongoro Stream where at its upper reaches a Maori oven was found (Topping, 1974 ). Chapter 3 The Maero Eruptive Period 89 (Puniho Ash), often related to a pyroclastic surge. One medial pyroclastic pumice flow deposit, a distal BAF (Puniho Breccia) and pyroclastic su rge deposits found on the NW sector can also be related to this event. The Puniho episode was preceded and succeeded by periods of repose of approximately 70 yrs (McGlone et al., 1988; Lees and Neall, 1993). The only sub-Plinian eruption within the Maer o Eruptive Period is recorded as Burrell eruptive episode, which occurred in AD 1655. Widespread pumice lapilli fallout (Burrell Lapilli) is observed on the NE-E-SE fla nks including three dist al pyroclastic pumice flows [Burrell Breccia (B) units 1-3; cf. Ch apter 5]. On the NW sector this eruptive episode is recorded as a major BAF [Burre ll Breccia (A)], pyrocla stic surge and three pyroclastic pumice flow deposits [Burrell Brecc ia (B) units 1-3]. No fallout deposits are observed here. Pyroclastic flows are preserved >13 km from source. Burrell Lapilli is overlain by a single lithic layer found near the old Mangahume Hut site (see Fig. 3-6). This proximal fa ll deposit is termed Mangahume Lapilli 8 . However, it cannot be correlated to any deposit found on the NW sector. It is possible that this eruptive event may only represent a minor lava dome-producing eruption similar to the youngest eruption of Mt. Taranaki (cf. Chapter 5). Due to an anticipated small erupted volume, associated deposits are probably only found in proximal reaches either represented by syn-eruptive rock falls or small BAFs, or post-eruptive by rock avalanche, debris flow or lahars. The Tahurangi eruptive episode began at c.AD 1755 based on tree ring dating and peat accumulation rates (Druce, 1966). On the eastern site of the mountain Tahurangi Ash is preserved in distal reaches as thin ash pockets ; at medial locations it is composed of up to 10 cm thick, fine to medium ash, often constituting the thickest ash layer found. This may be explained by the observation that a distal and a medial BAF deposit [Tahurangi Breccia (a) and (b)] can be correlated to the Tahurangi episode indicating a possible higher total magma discharge. The latest eruption of Mt. Taranaki, the Pyramid 9 eruptive episode (cf. Chapter 5), produced a lava dome (Pyramid Dome) partially preserved in the present summit crater. This episode is recorded in the tephrostratigraphic sequence on the southern flank as proximal single lithic layer (Pyramid Lapilli). In historic time (pre- AD 1885), a cold lava 8 Named after Mangahume Stream on the south flank of Mt. Taranaki. 9 Named after the morphological features Big and Little Pyramid on the NW flank of Mt. Taranaki. Chapter 3 The Maero Eruptive Period 90 dome collapse occurred producing a rock-avala nche deposit (Pyramid Rock Avalanche) on the NW sector. Although the eruption date is poorly constrained, it is concluded that the eruption presumably took place between AD 1840 and AD 1866 based on dome cooling calculations and historical reports (cf. Chapter 5). Chapter 3 The Maero Eruptive Period 91 3.4 References Alloway, B., Neall, V.E. and Vucetich, C.G. 1995. Late Quaternary (post 28,000 year B.P.) tephrostratigraphy of northeast and central Taranaki, New Zealand. Journal of the Royal Society of New Zealand 25, 385-4 58. Clarkson, B.D. 1990. A review of vegetation developmen t following recent (<450 years) volcanic disturbance in North Island, New Zealand. New Zealand Journal of Ecology 14, 59-71. Cronin, S.J. 1991. The geology of the summit of Mt. Egmont. Unpublished BSc Honours thesis. Massey University, Palmerston North, New Zealand. Cronin, S.J., Neall, V.E., Stewart, R.B. and Palmer, A.S. 1996. A multiple-par ameter approach to andesitic tephra correlation, Ruapehu volcano, New Zealand. Journal of Volcanology and Geothermal Research 72, 199- 215. Cronin, S.J., Neall, V.E., Palmer, A.S. and Stewart, R.B. 1997. Methods of identifying late Quaternary rhyolitic tephras on the ring plains of Ruap ehu and Tongariro volcanoes, New Zealand. New Zealand Journal of Geology and Geophysics 40, 175-1 84. Cronin, S.J., Stewart, R.B., Neall, V.E., Platz, T. and Gaylord, D. 2003. The AD1040 to present Maero eruptive period of Egmont Volcan o, Taranaki, New Zealand. Abstract. Geological Society of New Zealand Miscellaneous Publication 116A, 43. Downey, W.S., Kellett, R.J., Smith, I.E.M., Price, R.C. and Stewart, R.B. 1994. New paleomagnetic evidence for the recent eruptive activity of Mt. Taranaki, New Zealand. Journal of Volcanology and Geothermal Research 60, 15-2 7. Druce, A.P. 1966. Tree-ring dating of recent volcanic ash and lapilli, Mt Egmont. New Zealand Journal of Botany 4, 3-41. Fisher, R.V. 1995. Decoupling of pyroclastic currents: hazards assessments. Journal of Volcanology and Geothermal Research 66, 257- 263. Franks, A.M. 1984. Soils of Eltham County and the tephrochronology of central Taranaki. Unpublished PhD thesis. Massey University, Pa lmerston North, New Zealand. Grant - Taylor, T.L. 1964. Volcanic history of western Taranaki. New Zealand Journal of Geology and Geophysics 7, 78-8 6. Hoblitt, R.P. and Kellog, K.S. 1979. Emplacement temperature of unsorted and unstratified deposits of volcanic rock debris as determined by paleomagnetic techniques. Geological Society of America Bulletin 90, 633 -64 2. Lacroix, A. 1904. La Montagne Pelée et ses éruptions. Masson et Cie. Paris. Lees, C.M. and Neall, V.E. 1993. Vegetation response to volcanic eruptions on Egmont Volcano, New Zealand, during the last 1500 years. Journal of the Royal Society of New Zealand 23, 91-1 27. Chapter 3 The Maero Eruptive Period 92 Lowe, D.J. 1988. Stratigraphy, age, composition, and correla tion of late Quaternary tephras interbedded with organic sediments in Waikato lakes, North Island, New Zealand. New Zealand Journal of Geology and Geophysics 31, 125-1 65. May, D.J. 2003. The correlation of recent tephra with la va flows on Egmont volcano, Taranaki, New Zealand using evidence of minera l chemistry. Unpublished Master s thesis. The University of Auckland, Auckland, New Zealand. McGlone, M.S. 1982. Modern pollen rain, Egmont National Park, New Zealand. New Zealand Journal of Botany 20, 253- 262. McGlone, M.S. and Neall, V.E. 1994. The late Pleistocene and Holocene vegetation history of Taranaki, North Island, New Zealand. New Zealand Journal of Botany 32, 251- 269. McGlone, M.S., Neall, V.E. and Clarkson, B.D. 1988. The effect of recent volcanic events and climatic changes on the vegetation of Mt Egmont (Mt Taranaki), New Zealand. New Zealand Journal of Botany 26, 123- 144. Neall, V.E. 1972. Tephrochronology and tephrostratigraphy of western Taranaki (N108 -1 09) , New Zealand. New Zealand Journal of Geology and Geophysics 15, 507-5 57. Neall, V.E. 1973. Some aspects of western Taranaki geology and pedology. Unpublished PhD thesis. Victoria University of Wellington, Wellington, New Zealand. Neall, V.E. 1979. Sheets P19, P20, and P21. New Plymouth, Egmont and Manaia. 1st ed. Geological map of New Zealand 1:50,0 00. 3 maps and notes. New Zealand Department of Scientific and Industrial Research. Wellington. Neall, V.E. 2003. The volcanic history of Taranaki. Ins titute of Natural Resources, Massey University, Soil & Earth Sciences Occasional Publication 2. Neall, V.E., Stewart, R.B. and Smith, I.E.M. 1986. History and petrology of the Taranaki volcanoes. In: Smith, I.E.M. (ed.) Late Cenozoic volcanism. Royal Society of New Zealand Bulletin 23, 251-2 63. Oliver, W.R.B. 1931. An ancient Maori oven on Mount Egmont. Journal of the Polynesian Society 40, 73-8 0. Platz, T. 2001. Mapping and characterisation of the vol caniclastic Maero formation deposits on the northwestern sector of Egmont volcano (Mt. Taranaki), New Zealand. Unpublished Diploma mapping thesis. Ernst-Moritz-Arndt Universitä t, Greifswald, Bundesrepublik Deutschland. Platz, T., Cronin, S.J., Smith, I.E.M., Foley, S.F., Neall, V.E. and Stewart, R.B. 2004. Glass chemistry as a tool for tephra correlation: a new statistical approach on <1000 year old tephra deposits on Egmont volcano, Mt. Taranaki, New Zealand. Abstract. Geological Society of Australia 73, 284. Rosenthal, A. 2005. Mapping and characterisation of the eastern Fanthams Peak lavas, Egmont Volcano, Taranaki, New Zealand. Unpublished Diploma mapping thesis. Technische Universität Bergakademie Freiberg, Freiberg, Bundesrepublik Deutschland. Chapter 3 The Maero Eruptive Period 93 Shane, P. 2000. Tephrochronology: a New Zealand case study. Earth Science Reviews 49, 223-2 59. Shane, P.A.R. and Froggatt, P.C. 1994. Discriminant function analysis of glass chemistry of New Zealand and North American tephra deposits. Quaternary Research 41, 70-81. Stewart, R.B., Price, R.C. and Smith, I.E.M. 1996. Evolution of high-K arc magma, Egmont volcano, Taranaki, New Zealand: evidence from mineral chemistry. Journal of Volcanology and Geothermal Research 74, 275-2 95. Taylor, N.H. 1953. The ecological significan ce of the central North Island ash showers: the soil pattern. Proceedings of the New Zealand Ecological Society 1, 11-12. Topping, W.W. 1972. Burrell Lapilli eruptives, Mount Egmont, New Zealand. New Zealand Journal of Geology and Geophysics 15, 476-4 90. Topping, W.W. 1974. A 1480 A.D. Maori oven from Mount Egmont, New Zealand. New Zealand Journal of Science 17, 119- 122. Turner, M.B., Cronin, S.J., Bebbington, M.S. and Platz, T. 2007. Developing probabilistic eruption forecasts for dormant volcanoes: a case st udy from Mt Taranaki, New Zealand. Bulletin of Volcanology (DOI: 10.100 7 /s0044 5-0 07- 015 1-4 ). Watanabe, K., Ono, K., Sakaguchi, K., Takada, A. and Hoshizumi, H. 1999. Co-ignimbrite ash-fall deposits of the 1991 eruptions of Fugen-dake, Unzen Volcano, Japan. Journal of Volcanology and Geothermal Research 89, 95-1 12. Wellman, H.W. 1962. Holocene of the North Island of New Zealand: a coastal reconnaissance. Transactions of the Royal Society of New Zealand Geology 1 , 29-99. Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 95 Chapter 4 Improving the Re liability of Microprobe- based Glass Analyses 4 Chapter 6 This chapter addresses problems in correlating andesitic tephra units using glass chemistry. A simple evaluation procedure is presented to examine glass compositional datasets for contaminated hybr id glass-mineral analyses. Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 96 Chapter 4 represents the full version of a pub lished journal paper. Its format has been altered to match the overall thesis. Title Improving the reliability of microprobe-b ased analyses of andesitic glasses for tephra correlation. Authors Thomas Platz, Shane J. Cronin, Ian E.M. Smith, Michael B. Turner, and Robert B. Stewart. Status published in The Holocene 2007, vol. 17, 573-583. Principal investigator Thomas Platz Carried out - field description and sampling, - laboratory preparation of samples, - optical microscopy, - electron microprobe analysis. - manuscript preparation, writing and submission Co-investigators Shane J. Cronin, Robert B. Stewart, Ian E.M. Smith, and Michael B. Turner Aided the study by - providing Ruapehu glass analyses (Shane J Cronin) - providing access to and assistance with microprobe analysis - discussing results - ed iting and discussion of manuscript Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 97 4.1 Abstract Andesite tephras have not been used in tephrostratigraphic studies to the same extent as rhyolitic units because they are seen as being chemically more complex than the latter. They are particularly difficult to “fingerprint ” owing to apparent heterogeneity in glass analyses from single andesitic layers. Us ing case study tephras from two andesite volcanoes in New Zealand, Mts. Taranaki and Ruapehu, glass chemical heterogeneity is shown to be due predominantly to (1) differi ng particle types generated during closed- and open-vent phases of single eruption sequen ces, resulting in a broad range of glass compositions and (2) contaminated glass mi croprobe analyses by various proportions of microlite phase(s). A simple evaluation procedure using least-squares mixing calculations is presented to classify glass datasets for hybrid analyses and to estimate the proportions of the main contaminant microlite phase. By employing particle morphology studies as well as the glass-analys is evaluation procedure, variations in andesitic glass compositions can be significantly reduced. Hence this approach shows promise for enabling the use of some andesitic tephras for tephrostratigraphic correlation. This may facilitate the addition of a new degree of resolution in tephrostratigraphic records. 4.2 Introduction Volcanic systems show behaviour that is fundamentally linked to the physico-chemical state of erupting magma. The variety of such states enables a petrological approach to be used to identify eruption sequences, as we ll as individual eruptions, by analyses of the products of eruptions. However, the erup tive history of a volcano is not represented equally by its different facies . The proximal or edifice part of the volcano represents accumulated lava flows erupted with little explosive violence. More widely distributed volcanic products are produced during expl osive eruptions and transported up to hundreds of kilometres from the central vol cano. These distal deposits are petrologically more variable (Shane and Hoverd, 2002), but they record the largest and most hazardous eruptions. Deriving a complete erup tive history requires in tegration of both proximal and distal records. Typically the distal deposits provide the most complete temporal record to link various parts of volcanic systems and their principal investigative tool is tephrostratigraphy. Tephra is the collective term for solid particles Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 98 ranging in size from ash to block that were erupted from a volcano and deposited from the air. Tephrostratigraphy has underpinned our understanding of volcanic systems since Thorarinsson (1944) showed that single widesp read tephra layers can be used as marker beds, and when well dated, as isochrons. The a pplication of such isochrons as a dating tool is known as tephrochronol ogy. Precise age determination and correlation of tephra layers are important not only for the establishment of eruption recurrence intervals at volcanoes, but are also essential for dati ng events such as earthquake ground motion (Kelsey et al., 1998), tsunami inundation (Nanayama et al., 2003), global/regional climate changes (Björck et al., 1992), human history (Hogg et al., 2003), and the petrological evolution of volcanic systems (e.g. Gamble et al., 1999; Andreastuti et al., 2000). In the North Island of New Zealand, the se quence of rhyolitic tephras erupted from volcanoes of the Taupo Volcanic Zone (TVZ) is generally we ll established for the past c.60 ka (Froggatt and Lowe, 1990; Shane et al., 2002). Major and minor element compositions of glass and minerals have been successfully applied to the routine identification and correlation of distal eruptive units to their sources (Lowe 1988a; Eden and Froggatt, 1996; Shane, 1998). Some large volume (>>1 km 3 ) rhyolite tephras show homogeneous glass compositions making this characteristic one of the easiest means of correlation. In others, however, bimodal glas s compositions have been identified in a number of rhyolite tephras (Shane et al., 2002; Smith et al., 2005). Andesitic tephra units are interbedded with the rhyolite tephras in many sequences in New Zealand. In sequences younger than c.60 ka, these units erupted from either the Tongariro Volcanic Centre (TgVC, co mprising Mts. Ruapehu, Tongariro and Ngauruhoe), or Mt. Taranaki (Fig. 4-1). Dete rmining the source of these tephras as well as unequivocally correlating them to dated proximal units would add a new degree of resolution to the tephrostratigraphic record of the North Island. Shane (2005) renewed the focus on distal andesitic tephra layers, but correlating these with proximal deposits has been difficult because of the heterogeneity of the glass major element compositions and because there are few published reliable analyses of glass for proximal tephras (partly due to the problems outlined below). The present limited use of andesite tephra layers for correlation purposes can be attributed to three prime factors. Firstly, andesitic glass rapi dly alters to allophanic clay and is often poorly preserved, especially in soil-forming environments (Lowe, 1986). Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 99 Figure 4-1: Locations of Holocene vo lcanic centres in the North Island of New Zealand: Auckland Volcanic Field, Taupo Volcanic Zone (TVZ), Tongariro Volcanic Centre (TgVC), Mt. Taranaki. Numbers 1-4 refer to distal andesite tephra sites: 1-Onepoto Basin/Pukaki Lagoon/Lake Pupuke (Sandiford et al., 2001; Shane and Hoverd, 2002 ; Shane, 2005 ), 2-Waikato lakes (Lowe, 1988b), 3-Kaipo Bog (Lowe et al., 1999 ), 4-Lake Tutira (Eden and Froggatt, 1996 ), 5-Lake Poukawa (Shane et al., 2002 ), 6-Kaimanawa Mts. and Ruahine Ranges (Froggatt and Rogers, 1990 ). Secondly, andesitic volcanoe s produce relatively small volume tephras (mostly <0.1 km 3 ) compared with those from rhyolitic cent res. These low eruption volumes lead to narrowly distributed fallouts, often of very low thickness, and hence, poor preservation potential (Cronin et al., 1998). Thirdly, geoc hemical studies of andesitic glasses have commonly shown heterogeneous compositions. These variations are predominantly attributed to a strong influence of late-stage magma mixing/mingling and/or fractionation/crystallisation processes in the complex conduits of andesitic systems (Venezky and Rutherford, 1997; Hammer et al., 2000; Price et al., 2005). Some studies have shown that andesitic units can, in some cases, be distinguished, sourced and identified on the basis of their mineral chemistry (Donoghue and Neall, 1996), but Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 100 canonical discriminant function analysis is necessary (Cronin et al., 1996a,b). In these studies Fe-Ti oxides have proved especially useful for proximal to medial tephra distributions. However, gravit y settling and winnowing within the atmosphere reduces the dispersion of denser Fe-Ti oxides and mine ral particles to distal sites. As a result only glass and light minerals are typically present in distal andesitic tephras. Generally, their sources (TgVC vs. Taranaki) can be distinguished (Lowe, 1988a; Shane and Hoverd, 2002), but individual, discrete erup tives have been difficult to identify uniquely. Distinction between sources is ba sed on the observation that Mt. Taranaki magmas have generally (a lthough not always) high potassium contents, whereas Ruapehu eruptives have generally (with excep tions), lower K contents (Price et al., 1999). Because of these limitations, there have been very few attempts to match glass compositions to specific dated andesitic tephras (Lowe, 1988b; Eden and Froggatt, 1996; Shane, 2005). In this paper we sound a note of warning a nd suggest a new way forward for the use of glass compositions in tephra correlation studies that are based on andesitic systems. The main reasons for heterogeneous glass populations in andesites are explored and, through case studies of specific tephra units from two well studied volcanoes, the difficulties of identification and correlation using glass composition are demonstrated. A new simple procedure is introduced to enable electron microprobe analysis of andesitic glass to be critically evaluated so that spurious data can be detected and discarded. Our aim is to develop sound, glass-analysis based datasets useful for correlating andesite tephras. Using the criteria proposed here, we see gr eat potential for robust proximal-proximal and proximal-distal correlat ion of andesitic tephra. 4.2.1 Andesitic Volcanism, Tephr a Generation and Dispersal Tephra layers are produced at andesitic volcanoes through a wide a variety of eruptive styles, which can be summarised into two main types: (1) coeval ash associated with the effusion of lava flows and lava domes, a nd (2) explosive eruptions. A combination of lava emission and pumice production during the course of an eruption sequence is a common feature among andesite volcanoes (e.g. Santa Maria, Guatemala; Bezymianny, Kamchatka). Lava flow emplacement is often associated with very small eruption columns and only minor ash is produced as thin layers, mos tly on the upper flanks of the volcano (Hobden Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 101 et al., 2002). Such tephras are commonly of little use for correlation because they are poorly preserved, difficult to identify and not widely distributed. Du ring extrusion of lava domes, tephras are produced through phreatic explosions and during dome- collapses where Block-and-Ash Flows (BAF s) are generated. BAFs are pyroclastic flows, consisting of a dense, coarse grained basal avalanche overlain by a billowing cloud of elutriated ash and gas that can decouple from the flow and travel as a separate ash cloud surge (Fisher, 1995). This elutriated cloud of ash may rise several kilometers vertically to produce regional thin fallout deposits of “co-ignimbrite” ash (Gardner et al., 1994). Eruptions involving the explos ive expansion of gas-charge d magma produce the largest volumes of tephra and highest eruption columns. These types of events at NZ andesitic volcanoes have typically generated >10 km-h igh plumes with fall deposits covering several thousand square k ilometres (Alloway et al., 1995; Donoghue, et al., 1999). Often such highly explosive events at ande sitic volcanoes are immediately preceded by the emplacement of a lava dome, such as at Merapi, Indonesia (Boudon et al., 1993) and Redoubt volcano, Alaska (Gardner et al., 1994). The style of eruption is governed by the physical properties of erupted magma (i.e. viscosity, temperature, melt composition, wate r content, vesicularity) as well as the range of processes that occur as andesitic magma approaches the surface. The most important of these processes is the mixing or mingling of magma batches, enabled by the relatively low viscosity of andesitic magmas in comparison to rhyolitic ones (Murphy et al., 1998; Roman et al., 2006). Mixing or ming ling may be of distinct magmas, but because there are commonly ma ny small batches of melt accumulating under similar conditions in andesitic edifices (Gamble et al., 1999; Price et al. 2005), mixed compositions may only have subtle differences. In addition, the hot and low- viscosity melts enable high rates of fractionation/crystallisation during magma ascent (Blundy and Cashman, 2001), along with syn-eruptive degassing (Wolf and Eichelberger, 1997). Pre- and syn-eruptive crystallisation, especi ally during the extrusio n of lava domes, has a pronounced impact on melt composition and viscosity. As the magma progressively ascends and degasses, groundmass microlite cr ystallization is induced (Cashman, 1988). Most commonly, needle-like plagioclase cr ystals form (<10 µm in length). Their number density and distribution in the upper conduit and the lava dome are dependent on local temperature and pressure gradients, and as a result th e behaviour of the Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 102 remaining melt is altered by varying degrees. As well as these processes, because andesitic magmas are generally erupted at higher temperatures than those of rhyolitic composition, post-eruptive crystallisation and microlite formation may also occur and generate fine-scale heterogeneity, for exam ple, as a result chemical diffusion and concentration gradients in areas near (<100 µm from the solid-liquid interface) growing crystals (Best, 2003). All of these processes are active to varying extents during the er uption of individual tephras, with rapid ascent rates and short periods of high-level pre-eruption storage imparting the lowest degrees of glass heterogeneity. If these processes operate on longer time scales, major differences in glass co mposition, vesicularity and even bulk rock composition can be produced (e.g. Hammer et al., 2000; Platz et al., 2007). Hence magmatic/eruptive factors may produce vari able glass compositions in andesite eruptions, and also fine-scale microtextures of andesitic glass may give rise to analytical errors in its measurement by the most common method of electron microprobe analysis. The difficulty in achieving reliable glass electron microprobe results lies in being able to distinguish clearly between groundmass glass and microlites. Plagioclase and Fe-Ti oxide minerals are the most common late-stage crystallising phases in andesitic glasses. On backscatter electron images (BSE) there is litt le contrast between the very similar grey-tones of tiny plagioclase microlite s and glass, because both are composed predominantly of light elements (i.e. Si, Al , Ca, Na). In additi on, despite the clear contrast between Fe-Ti oxide phenocrys ts/microphenocrysts and groundmass-glass on BSE images, minute, nanolite oxide-crystals (200-10 0 0 nm) are also hardly resolvable (Schlinger et al., 1986). Particles with a hypocrystalline groundmass contain abundant needle-like plagioclase microlites of <10 µm lengths. To find a pure glass area within the groundmass for a defocused 20 µm electron beam can be highl y challenging. Similar problems occur in highly vesicular samples where bubble walls can be as thin as 6 µm, and the intersection of bubble walls are often occupied by microlites and microphenocrysts. If the electron- beam diameter is only slightly less than the thickness of bubble walls, analyses often have low totals. The interpretation and evaluation of glass compositional data also proves difficult because most tephras, especially andesites, show large variations commonly in SiO 2 , Al2 O 3 , CaO and alkali (Sandiford et al., 2001; Shane et al., 2002). Shane (2005), for example, reported a range in silica content of up to 10 wt.%, and consequently a high Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 103 range in values for other element oxides with in a single tephra layer. Such wide ranges were attributed to magma fractionation/crystallization processes, a compositionally zoned magma chamber, or mixing magma batc hes prior to and during the eruption. The potential for contamination of glass analyses by microcrystalline phases has been considered by many scientists (e.g. Lowe, 1988a, b) but no thorough evaluation of such contamination has been made, nor has a proc edure been developed to identify hybrid glass data. Hunt and Hill ( 2001) provided guidelines for electron microprobe analysis evaluation, but our study consid ers further analytical as well as volcanological aspects. 4.2.2 Sample Sites We chose tephras from two well studied New Zealand andesite volcanoes, Mts. Ruapehu and Taranaki. One individual ande site tephra from each of these volcanoes was chosen to demonstrate the problems inherent in using andesitic tephras for stratigraphic correlation and to introduce a glass data evaluation procedure that can circumvent some of these problems. The selected tephra units are: (1) the Burrell Lapilli sub-Plinian eruption ( AD 1655) of Mt. Taranaki (Platz et al., 2007), and (2) the 14 October 1995 tephra of Mt. Ru apehu (Cronin et al., 2003). 4.3 Results 4.3.1 Contrasts in Particle Texture Thin-sectioning and microscopic analysis of tephras allows differentiation of petrographic textures within individual part icles. Such a study is demonstrated in a sample from the sub-Plinian, Burrell Lapilli eruption of Mt. Tarana ki. A reconstruction of the eruptive sequence of th e Burrell Lapilli eruption ( AD 1655) shows that there were two phases to the eruption: (1) extrusion of a lava dome, followed by (2) an explosive, sub-Plinian event. Tephra fallout from th e sub-Plinian eruption column was first deposited towards the SE and then as the wi nd shifted, the dispersa l axis progressively shifted north-eastwards. Distal ash from this event has been found as far as Lake Tutira, 250 km east of Mt. Taranaki (Eden and Froggatt, 1996). Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 104 In grain-mount thin sections of Burrell tephra collected from medial locations, three types of ash particle text ures were distinguished: Type 1 - pumice particles that are highly vesicular with a glassy groundmass (Fig. 4-2a). Type 2 - dense lithic particles that have a hypocrystalline textured groundmass dominated by plagioclase microlites and tiny Fe-Ti oxide crystals (Fig. 4-2b). Type 3 - semi-vesicular pa rticles with a pale brown glassy groundmass containing fewer microlites compared with type 2 (Fig. 4-2c). These particles are least common. Type 2 and 3 particles were generated during the initial extrusive phase of the eruption and represent parts of the dome (type 2) and the upper conduit (type 3). The highly vesicular pumice clasts are clearly associated with the subsequent explosive phase. Despite the youth of these er uptive products, many of the pu mice particles are slightly weathered and show a characteristic pale brown to brown appearance. Figure 4-2: Particle textures found in Burrell Lapilli sub-Plinian fall deposits. a) pumice clast, type 1, with clear to pale brownish glass; b) hypocrystalline groundmass texture of type 2 clast showing plagioclase, Fe-Ti oxide and minor clinop yroxene microlites; c) semi-vesicular type 3 clast with brown groundmass glass, large cr ystals are hornblende; d) for comparison, hypocrystalline groundmass of the present summit dome of Mt. Taranaki with abundant microlites of plagioclase, Fe-Ti oxides and minor clinopyroxene. Scale bars are in µm. Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 105 4.3.2 Glass Chemistry and Data Evaluation Compared to andesitic glass (Table 4-1) , plagioclase has higher abundances of Al 2 O 3 and CaO, and to some degree Na 2 O, whereas SiO 2 is consequently lower (Table 4-2). If plagioclase contamination is suspected, the analyses should therefore have elevated values in Al2 O 3 , CaO, and possibly Na 2 O, but slightly lower values of SiO 2 . In addition, the glass data population should display a broad range in these oxi des compared with the restricted and complementary variation imposed by the stoichiometric limitations arising from the crystallographic structure of minerals. St andard deviations in glass compositions are typically highest for SiO 2 , Al 2 O 3 and CaO (Fig. 4-3), where high standard deviations for these three oxides show that contamination by plagioclase is evident. Figure 4-3: Standard deviations of all major oxides are shown for the original glass EMPA dataset and the glass dataset classed as uncontaminated by plagioclase for the Taranaki (a) and Ruapehu (b) samples. A clear reduction in SiO 2 , Al 2 O 3 and CaO variations is observed. See text for further details. All Fe expressed as FeO. Two variation diagrams illustrate that glass compositions of both tephras show a generally negative correlation of Al2 O 3 with increasing SiO 2 and a positive relationship between K 2 O and SiO 2 (Fig. 4-4). It is apparent that in both examples the glass population is generally oriented towards the mean compositions of the respective associated plagioclase. In both examples, data points with lowest silica contents show a strong linear relationship and plot on a glass-plagioclas e mixing line (Fig. 4-4). If contamination of glass EMPA data by a mi neral phase is suspected, the best way to test the dataset is to use bivariate compositional diagrams, where an incompatible and a compatible oxide of the mineral phase are plot ted. Because plagioclase consists of only Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 106 Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 107 Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 108 Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 109 four major oxides, all other anal ysed oxides could be considere d. In practice, it is best to use those that are well above detection limits, associated with low analytical errors and that are concentrated in the glass phase. In this case K 2 O and FeO are considered as incompatible oxides although minor iron (as Fe 3+ ) can also be substituted for Si 4+ in the plagioclase lattice. Figure 4-4: Bivariate plots of Al 2 O 3 and K 2 O vs. SiO 2 for Burrell Lapilli (Taranaki) and Ruapehu glass data. Data points with lowest SiO 2 show linear relationship towards mean plagioclase compositions (dashed lines). Using this empirical approach, the Burrell da taset can be classified into contaminated and uncontaminated analyses (Fig. 4-5a-d). Iron vs. aluminium oxides proved most helpful in classifying the data and an upper threshold value of 17.1 wt.% Al 2 O 3 appears to represent the limit for uncontaminated analysis. Analyses with higher alumina contents plot off the bulk population and towards the mean plagioclase composition (Fig. 4-5a). This threshold valu e is also notable on plots of K 2 O vs. SiO 2 , CaO and Na 2 O with contaminated data mostly converging toward the mean plagioclase composition. In the same way, upper threshol d values can be identified for CaO (3.5 wt.%) and K 2 O (4.4 wt.%) corresponding to th e threshold value found for Al2 O 3 (17.1 wt.%). Agreement of the data points classed as contaminated using all three threshold values gives confidence that the contamination is due to plagioclase. Hence these data must be erased from the dataset. After applying this technique there are two points slightly off the plagioclase-glass mixing line (Fig. 4-5d-c). These likely represen t a case of overlapping contamination, in this instance probably by minute Fe-Ti oxide nanolites. Data offsets from the ideal plagioclase-glass mixing line can also be due to variations in plagioclase microlite compositions, i.e. variations, especially in SiO 2 and Al2 O 3 , between rim and centre of the microlites caused by substitution of Al3+ for Si 4+ in tetrahedral position (Table 4-2). Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 110 Using this approach to identifying contamin ated glass EMPA, the variation within the Burrell dataset can be significantly reduced as shown by the decrease in standard deviations for SiO 2 , Al 2 O 3 , CaO and K 2 O (Fig. 4-3a). Figure 4-5: Bivariate oxide plots for Burrell Lapilli eruption, Taranaki (a-d) and 14. October 1995 eruption, Ruapehu (e-f) demonstrate how contaminated glass analyses were identified. Open symbols are classed uncontaminated, closed symbols represent hybr id glass-plagioclase anal yses. Dashed lines point towards mean plagioclase compositions. Tephra erupted during the major phase of the 1995 Ruapehu eruption on 14 October (Cronin et al., 2003) also show similar trends (Fig. 4-5e-f), despite representing a simple type of eruption event without the major differences in style identified for the Burrell event. In this case an upper threshold value of 16 wt.% Al 2 O 3 appears to identify the limit of uncontaminated data. This threshold value is supported by other major oxides Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 111 with the exception of Na 2 O. In this case Na 2 O shows no trend towards the mean plagioclase composition. The applied procedure reduced the variation, especially in Al2 O 3 , however, a large range in SiO 2 still remains (Fig. 4-3b). High standard deviations for CaO, FeO and MgO remain which may be due to contamination from microlites and nanolites of clinopyroxene, orthopyroxene and/or Fe-Ti oxide . These are common groundmass constituents in Ruapehu tephras (Cronin et al., 1997) and common microlites of orthopyroxene, although showing mo re contrast than plagioclase, are still difficult to distinguish from glass. 4.3.3 Estimating Plagioclase Proportions in Contaminated Analyses Once hybrid glass-plagioclase analyses are identified, the proporti on of plagioclase within the composite analysis should be calculable. In the Taranaki example a procedure is demonstrated by using four data points that plot closest to the assumed plagioclase-glass mixing line (Fig. 4-5b-c). To calculate mixing proportions requires the assumption of a composition where mixing betw een glass and plagioclase is thought to start. In this case, the Al 2 O 3 threshold value of 17.1 wt.% is expected to equal the starting composition, and the oxides of all four contaminated data points were linearly extrapolated back to Al 2 O 3 = 17.1 wt.% (Table 4-3). Mass balance calculations are then used to best fit the mixing proportion. In this example, the maximum portion of plagioclase within the hybrid analysis is estimated to be 13.7%. There is a strong linearity displayed by contaminated data points that can be used to constrain their mixing proportions. Low standard deviations (<0.21) for all oxides indicate very good agreement between analysed and calculated data points (Table 4-3). Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 112 Table 4-3. Calculation of various plagioclase mixing proportions within Taranaki hybrid glass EMPA. EMPA data in wt.%. See text for explanation. 4.4 Discussion In our experience, before tephra deposits are analysed by EMPA for their glass composition, the bulk sample should be examined for particle textures. The three types of particles identified in the Burrell Lapi lli fall deposit (Taranaki) occurred during different phases of the eruption, with lithic and poorly vesicular types 2 and 3 from initial dome-forming events, and pumice clasts from a climactic final sub-Plinian phase (Platz et al., 2007). The proportions of highest-d ensity type 2 and 3 clasts reduce rapidly with distance from source in the fall deposit, however, small fragments are still commonly found in distal reaches. This range of particle types is not just confined to this event, but has also been recognised in other eruptions of Taranaki (Turner et al., in press) and at other centre s (Hammer et al., 2000). The mineralogical composition of distal andesite tephras is commonly reported to aid in correlation (e.g. Lowe, 1988a, b; Eden and Froggat, 1996). However, detailed descriptions of particle types in the tephra fall deposits have been conspicuously absent, despite the fact that these can be simply determined from SEM or optical microscopy methods. Analysing glass without knowing the context of part icle types in a tephra unit Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 113 may lead to wide ranges in composition that are related to different phases of an eruption and are difficult to use for correlation purposes. By analysing each particle type as a separate sub-population for any given uni t, correlation studies can proceed with “like” being compared to “like”. Similarly, previous authors have stated the possibility of plagioclase contamination (e.g. Lowe, 1988a, b; Shane, 2005). Our study has shown that checking glass compositional data for impurities, especially plagioclas e, can tighten the compositional range for individual tephra units significantly. This in turn improves their potential for correlation purposes. If analyses of the contaminati ng phase are available, the proportion of contamination can also be evaluated, and this may be of use in determining the physical characteristics, such as microlite content, of individual tephras. The evaluation of broad glass populations must not only take into account particle type and potential microlite contamination in EMPA data, but there are additional features of some types of andesitic eruptions that make their tephras unsuitable for correlation. Tephras that result from pyroclastic surges, phreatic/phreatomagmatic blasts or BAFs generated from a lava dome collapse are dominated by type 2 and 3 li thic particles (Fig. 4-2b-c). In these cases, variable crystal lisation/cooling rate environments (i.e. temperature, water content, pressure) in di fferent parts of the lava dome and upper conduit gives rise to variable glass compositions (Hammer et al., 2000; Platz et al., 2004). In addition, lava dome rocks are char acterised by very high abundances of plagioclase microlites as well as Fe-Ti ox ides in the groundmass (see Fig. 4-2d). The very late stage crystallising microlites may lead to strong variations in surrounding melt composition on a micro-scale. Hence analysis of glasses in such dome-derived particles can lead to a broad compositional range for any single tephra unit. The strongest variation is still expected to be a sub-linear relationship between late-stage plagioclase- microlite and melt compositions. Using an example from the Burrell Lapilli eruption, the glass compositions of type 1 and 2 partic les in a BAF deposit of this sequence (Fig. 4-6) show a broad compositi onal range including hybrid analyses. In this instance, upper threshold values could be set at 17.9 wt.% Al 2 O 3 . This is in contrast to the value of 17.1 wt.% Al 2 O 3 used for the type 1 particles of the stratigraphically equivalent Burrell pumice-rich fall. The analyses betw een these two thresholds, show a further degree of transition, and could arguably be also excluded from the “true” glass population. However, this may not be appropr iate on grounds of contamination, instead they show that many of the points analysed in type 2 and 3 particles are from glass that Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 114 is fractionated to higher degrees than in type 1 shards (i.e. by a gr eater degree of late- stage microlite formation). In addition, the presence of more-common minute Fe-Ti oxides may also contribute to creating thes e transitional compositions. Hence, precise mean values and tight ranges of true “glass ” compositions for type 2 or 3 particles can be difficult to derive, and may not relate at all to correlative deposits made up mostly of type 1 pumice shards. Figure 4-6: The glass evaluation procedure cannot be directly applied to BAF deposits as demonstrated for the Burrell Lapilli equivalent BAF deposit. Although data points above a threshold value of 17.9 wt.% Al 2 O 3 clearly embrace contaminated glass analyses, the transitional data between 17.1 and 17.9 wt.% Al 2 O 3 cannot be uniquely classified. Dashed line points toward mean plagioclase composition. A review of all andesite glass compositions used for correlation purposes is required. In New Zealand, the published data from TgVC and Taranaki centres may be unreliable because they include both microlite-contamin ated analyses as well as particles from different cooling/eruption histories. The published mean compositions of Al2 O 3 and K 2 O (Lowe, 1988b; Lowe et al., 1999; Eden and Froggatt, 1996 and Shane and Hoverd, 2002) for a range of distal Tg VC and Taranaki tephras, when taken with their standard deviations (Fig. 4-7), show no possibility for discrimination of individual units, and difficult discrimination even between individual eruptive centres. The data from the Burrell Lapilli eruption (Taranaki) have been plotted for comparison (in grey, Fig. 4-7). Although the previously stated authors acknowledged difficulti es in glass analyses and the shortfall in application, it is appa rent that the large variation in Al2 O 3 is a clear indicator for plagioclase contamination in the published datasets, and that the removal of hybrid data points could significantly improve the use for correlation purposes. Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 115 A further problem in published glass analyses is the use of glass analyses with low total oxide sums. Shane and Hoverd (2002) and Sa ndiford et al. (2001) reported mean total oxide sums as low as 91.6 wt.% with indi vidual analyses lower than 86.5 wt.%. These differences clearly cannot be accounted for by water dissolved in glass. Even though secondary hydration of glass takes place af ter deposition (water absorption starts immediately as hot tephra particles come in contact with air), this cannot explain the very low totals. Therefore, these data should be discarded from the dataset. Figure 4-7: Glass compositions of Taranaki and TgVC tephras (Lowe 1988b; Lowe et al., 1999 ; Eden and Froggatt, 1996 and Shane and Hoverd, 2002 ) show large variations, here shown for the means and standard deviations of K 2 O and Al 2 O 3 . Contaminated plagioclase- glass analyses and/or the analysis of two or more particle types may have caused the apparent glass compositional heterogeneity. The small variation in the unmodified Burrell Lapilli dataset (Taranaki) is shown for comparison (in grey). Estimates of glass water contents using the weight-by-difference (WBD) method is commonly used and at times can be highly questionable, for instance, 13.5 wt.% H 2 O in andesite glass (as described above) is unr ealistic. Hydrogen cannot be measured by EMPA and the difference in analyses tota ls from 100% is commonly attributed to dissolved H 2 O in glass, despite the fact that other volatiles can contribute significant proportions, for instance SO 3 and F (Luhr et al., 1984; Croni n et al., 1998). In addition, Roman et al. (2006) demonstrated that water contents in glass inclusions estimated by WBD are approximately 1 wt .% higher than water cont ents measured by Fourier Transform Infrared analysis. Hence, low analysis totals for glass are more likely to represent measurement error which could result from a number of sources, including: (1) the exposed glass surface Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 116 not being as large as the beam spot; (2) the beam being placed incorrectly - e.g. overlapping the edge of a grain; (3) poor polish and scratc hes on the grain; (4) glass vesicles within the beam spot, or (5) errors in the detection/count ing of one or more oxides. Given these possibilities, covering the missing fractions of the analyses by simply calculating ratios from each of the components will produce questionable results at best. This type of practice is more likel y to lead to broader compositional ranges and consequent greater difficultie s in unique correlations. 4.5 Conclusions Although glass EMPA are carefully execute d by avoiding visible inclusions (i.e. mineral phases, vesicles) and common guideli nes are followed (cf. Froggatt, 1983; Hunt and Hill, 2001) a sensible error analysis is a final step that should be taken, one that has often been neglected. Also, previous studies on andesite eruptions commonly have lacked particle textural studi es which are considered a major factor for contributing to large variations in glass compositional datasets. This study highlights the fact that hybrid mineral-glass EMP analyses are common in datasets that are thought to contain “pur e” glass compositional analyses. By thoroughly reviewing those datasets by using bivariate oxide diagrams (comprising a compatible and an incompatible oxide of the most common groundmass mineral phase) hybrid data points can be identified and excluded from th e set. Proportions of the mineral phase(s) in hybrid analyses can be estimated by extra polating the classed contaminated data back to the identified threshold value. While working on andesite tephra sequences, it should always be kept in mind that tephra units are deposited through a variety of processes, an d single eruptive events can switch between eruptive styles producing various particle types, and hence, different glass compositions. Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 117 4.6 References Alloway, B., Neall, V.E. and Vucetich, C.G. 1995. Late Quaternary (post 28,000 year B.P.) tephrostratigraphy of northeast and central Taranaki, New Zealand. Journal of the Royal Society of New Zealand 25, 385-4 58. Andreastuti, S.D., Alloway, B.V. and Smith, I.E.M. 2000. A detailed tephrostratigraphic framework at Merapi Volcano, Central Java, Indonesia: implications for eruption predictions and hazard assessment. Journal of Volcanology and Geothermal Research 100, 51-67. Best, M.G. 2003. Igneous and metamorphic petrology. Blackwell Science. Malden. B j örck, S., Ingólfsson, Ó., Haflidason, H., Hallsdóttir, M. and Anderson, N.J. 1992. Lake- Torfadalsvatn: a high-resolution record of the No rth Atlantic ash zone I and the last glacial- interglacial environmental changes in Iceland. Boreas 21, 15-22. Blundy, J. and Cashman, K. 2001. Ascent-driven crystallisation of dacite magmas at Mount St Helens, 1980 -19 86. Contributions to Mineralogy and Petrology 140, 631-6 50. Boudon, G., Camus, G., Gourgaud, A. and Lajoie, J. 1993. The 1984 nuée-ardente deposits of Merapi volcano, Central Java, Indonesia: stratigraphy, te xtural characteristics, an d transport mechanisms. Bulletin of Volcanology 55, 327-3 42. Cashman, K.V. 1988. Crystallization of Mount St. Helens 1980-198 6 dacite: a quantitative textural approach. Bulletin of Volcanology 50, 194-20 9. Cronin, S.J., Neall, V.E., Stewart, R.B. and Palmer, A.S. 1996a. A multiple-parameter approach to andesitic tephra correlation, Ruapehu volcano, New Zealand. Journal of Volcanology and Geothermal Research 72, 199- 215. Cronin, S.J., Wallace, R.C. and Neall, V.E. 1996b. Sourcing and identif ying andesitic tephras using major oxide titanomagnetite and hornblende chemis try, Egmont volcano and Tongariro Volcanic Centre, New Zealand. Bulletin of Volcanology 58, 33-4 0. Cronin, S.J., Neall, V.E., Palmer, A.S. and Stewart, R.B. 1997. Methods of identifying late Quaternary rhyolitic tephras on the ring plains of Ruap ehu and Tongariro volcanoes, New Zealand. New Zealand Journal of Geology and Geophysics 40, 175-1 84. Cronin, S.J., Hedley, M.J., Neall, V.E. and Smith, R.G. 1998. Agronomic impact of tephra fallout from the 1995 and 1996 Ruapehu Volcano eruptions, New Zealand. Environmental Geology 34, 21-3 0. Cronin, S.J., Neall, V.E., Lecointre, J.A., Hedley, M.J. and Loganathan, P. 2003. Environmental hazards of fluoride in volcanic ash: a cas e study from Ruapehu volcano, New Zealand. Journal of Volcanology and Geothermal Research 121, 271-2 91. Donoghue, S.L. and Neall, V.E. 1996. Tephrostratigraphic studies at Tongariro volcanic centre New Zealand: an overview. Quaternary International 34-6 , 13-20. Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 118 Donoghue, S.L., Palmer, A.S., McClelland, E., Hobs on, K., Stewart, R.B., Neall, V.E., Lecointre, J.A. and Price, R.C. 1999. The Taurewa Eruptiv e Episode: evidence for climactic eruptions at Ruapehu volcano, New Zealand. Bulletin of Volcanology 60, 223-2 40. Eden, D.N. and Froggatt, P.C. 1996. A 6500-year-old history of tephra deposition recorded in the sediments of Lake Tutira, eastern North Island, New Zealand. Quaternary International 34/36 , 55-64. Fisher, R.V. 1995. Decoupling of pyroclastic currents: hazards assessments. Journal of Volcanology and Geothermal Research 66, 257- 263. Froggatt, P.C. 1983. Toward a comprehensive upper Quaternary tephra and ignimbrite stratigraphy in New Zealand using electron-micropr obe analysis of glass shards. Quaternary Research 19, 188- 200. Froggatt, P.C. and Lowe, D.J. 1990. A review of late Quaternary silicic and some other tephra formations from New Zealand - their stratigraphy, nomenclature, distribution, volume, and age. New Zealand Journal of Geology and Geophysics 33, 89-10 9. Froggatt, P.C. and Rogers, G.M. 1990. Tephrostratigraphy of high-altitude peat bogs along the axial ranges, North Island, New Zealand. New Zealand Journal of Geology and Geophysics 33, 111-1 24. Gamble, J.A., Wood, P., Price, R.C., Smith, I.E.M., Stewart, R.B. and Waight, T. 1999. A fifty year perspective of magmatic evolution on Ruapehu volcano, New Zealand: verification of open-system behaviour in an arc volcano. Earth and Planetary Science Letters 170, 301 -31 4. Gardner, C.A., Neal, C.A., Waitt, R.B. and Janda, R.J. 1994. Proximal pyroclastic deposits from the 1989-19 90 eruption of Redoubt Volcano, Alaska: stratigraphy, distribution, and physical characteristics. Journal of Volcanology and Geothermal Research 62, 213- 250. Hammer, J.E., Cashman, K.V. and Voight, B. 2000. Magmatic processes revealed by textural and compositional trends in Merapi dome lavas. Journal of Volcanology and Geothermal Research 100, 165- 192. Hobden, B.J., Houghton, B.F. and Nairn, I.A. 2002. Growth of a young, frequently active composite cone: Ngauruhoe volcano, New Zealand. Bulletin of Volcanology 64, 392-4 09. Hogg, A.G., Hi gham, T.F.G, Lowe, D.J., Palmer, J.G., Reimer, P.J. and Newnham, R.M. 2003. A wiggle-match date for Polynesian settlement of New Zealand. Antiquity 77, 116-1 25. Hunt, J.B. and Hill, P.G. 2001. Tephrological implications of b eam size - sample-size effects in electron microprobe analysis of glass shards. Journal of Quaternary Science 16, 105–1 17. Jarosewich, E., Nelen, J.A. and Norberg, J.A. 1980. Reference samples for electron microprobe analysis. Geostandards Newsletter 4, 43-4 7. Kelsey, H.M., Hull, A.G., Cash man, S.M., Berryman, K.R., Ca shman, P.H., Trexler, J.H. and Begg, J.G. 1998. Paleoseismology of an active reverse fault in a forearc setting: the Poukawa fault zone, Hikurangi forearc, New Zealand. Geological Society of America Bulletin 110, 112 3-1 148. Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 119 Lowe, D.J. 1986. Controls on the rates of weathering and clay mineral genesis in airfall tephras: a review and New Zealand case study. In: Coleman, S.M. and Dethier, D.P. (eds.) Rates of chemical weathering of rocks and minerals. Academic Press, Dordrecht. pp. 265-330. Lowe, D.J. 1988a. Late Quaternary volcanism in New Zealand: towards an integrated record using distal airfall tephras in lakes and bogs. Journal of Quaternary Science 3, 111-1 20. Lowe, D.J. 1988b. Stratigraphy, age, composition, and correlation of late Quaternary tephras interbedded with organic sediments in Waikato lakes, North Island, New Zealand. New Zealand Journal of Geology and Geophysics 31, 125-1 65. Lowe, D.J., Newnham, R.H. and Ward, C.M. 1999. Stratigraphy and chronology of a 15 ka sequence of multi-sourced silicic tephras in a montane peat bog, eastern North Island, New Zealand. New Zealand Journal of Geology and Geophysics 42, 565-5 79. Luhr, J.F, Carmichael, I.S.E. and Varekamp, J.C. 1984. The 1982 eruptions of El Chichon Volcano, Chiapas, Mexico - mineralogy and petrology of the anhydrite-bearing pumices. Journal of Volcanology and Geothermal Research 23, 69-1 08. Murphy, M.D., Sparks, R.S.J., Barclay, J., Carroll, M.R., Lejeune, A.M., Brewer, T.S., Macdonald, R., Black, S. and Young, S. 1998. The role of magma mixing in triggering the current eruption at the Soufriere Hills volcano, Montserrat, West Indies. Geophysical Research Letters 25, 3433- 343 6. Nanayama, F., Satake, K., Furukawa, R., Shimokawa, K., Atwater, B.F., Shigeno, K. and Yamaki, S. 2003. Unusually large earthquakes inferred from tsunami deposits along the Kuril trench. Nature 424, 660 -66 3. Nielsen, C.H. and Sigurdsson, H. 1981. Quantitative methods for el ectron microprobe analysis of sodium in natural and synthetic glasses. American Mineralogist 66, 547 -5 52. Platz, T., Cronin, S.J., Smith, I.E.M., Foley, S.F., Neall, V.E. and Stewart, R.B. 2004. Glass chemistry as a tool for tephra correlation: a new statistical approach on <1000 year old tephra deposits on Egmont volcano, Mt. Taranaki, New Zealand. Abstract. Geological Society of Australia 73, 284. Platz, T., Cronin, S.J., Cashman, K.V., Stewart, R.B. and Smith, I.E.M. 2007. Transition from effusive to explosive phases in andesite eruptio ns - a case-study from the AD1655 eruption of Mt. Taranaki, New Zealand. Journal of Volcanology and Geothermal Research 161, 15-34. Price, R.C., Stewart, R.B., Woodhead, J.D. and Smith, I.E.M. 1999. Petrogenesis of high-K arc magmas: evidence from Egmont Volc ano, North Island, New Zealand. Journal of Petrology 40, 167- 1 97. Price, R.C., Gamble, J.A., Smith, I.E.M., Stewart, R.B., Eggins, S. and Wright, I.C. 2005. An integrated model fro the temporal evolution of andesites and rhyolites and crustal development in New Zealand’s North Island. Journal of Volcanology and Geothermal Research 140, 1-2 4. Chapter 4 Improving the Reliability of Microprobe-based Glass Analyses 120 Roman, D.C., Cashman, K.V., Gardner, C.A., Wallace, P.J. and Donovan, J.J. 2006. Storage and interaction of compositionally heterogeneous magmas from the 1986 eruption of Augustine Volcano, Alaska. Bulletin of Volcanology 68, 240- 254. Sandiford, A., Alloway, B. and Shane, P. 2001. A 28000 -66 00 cal yr record of local and distal volcanism preserved in a paleolake, Auckland, New Zealand. New Zealand Journal of Geology and Geophysics 44, 323-3 36. Schlinger, C.M., Smith, R.M. and Veblen, D.R. 1986. Geologic origin of magnetic volcanic glasses in the KBS tuff. Geology 14, 959 -96 2. Shane, P. 1998. Correlation of rhyolitic pyroclastic eruptive units from the Taupo volcanic zone by Fe-Ti oxide compositional data. Bulletin of Volcanology 60, 224-23 8. Shane, P. 2005. Towards a comprehensive distal andesitic tephrostratigraphic framework for New Zealand based on eruptions from Egmont volcano. Journal of Quaternary Science 20, 45-5 7. Shane, P. and Hoverd, J. 2002. Distal record of multi-sourced te phra in Onepoto Basin, Auckland, New Zealand: implications for volcanic chronology, frequency and hazards. Bulletin of Volcanology 64, 441- 454. Shane, P., Lian, O.B., Augustinus, P., Chisari, R. and Heijnis, H. 2002. Tephrostratigraphy and geochronology of a ca. 120 ka terrestrial record at Lake Poukawa, North Island, New Zealand. G lobal and Planetary Change 33, 221-24 2. Smith, V.C., Shane, P. and Nairn, I.A. 2005. Trends in rhyolite geochemistry, mineralogy, and magma storage during the last 50 kyr at Okataina and Ta upo volcanic centres, Taupo Volcanic Zone, New Zealand. Journal of Volcanology and Geothermal Research 148, 372 -40 6. Thorarinsson, S. 1944. Tefrokronologiska studier på Island. Geografiska Annaler 26, 1-2 17. Turner, M.B., Cronin, S.J., Bebbington, M.S. and Platz, T. 2007. Developing probabilistic eruption forecasts for dormant volcanoes: a case st udy from Mt Taranaki, New Zealand. Bulletin of Volcanology (DOI: 10.1007 /s0044 5-0 07- 015 1-4 ). Venezky, D.Y. and Rutherford, M.J. 1997. Preeruption conditions and timing of dacite-andesite magma mixing in the 2.2 ka eruption at Mount Rainier. Journal of Geophysical Research 102, 20069 -20 086. Wolf, K.J. and Eichelberger, J.C. 1997. Syneruptive mixing, degassing, and crystallization at Redoubt Volcano, eruption of December, 1989 to May 1990. Journal of Volcanology and Geothermal Research 75, 19-37. Chapter 5 Non-explosive, Dome-forming Er uptions 121 Chapter 5 Non-explosive, Dome-forming Eruptions 5 Chapter 5 Chapter 5 demonstrates that the Tahurangi eruption in AD 1755 was followed by a small, extrusive dome-forming event. In order to evaluate the physico-chemical conditions of dome formation at Taranaki, the structure and composition of the summit dome was investigated in detail. Magma stor age depths, extrusion and ascent rates, and eruption duration were interpreted for the first time ever at Taranaki. In addition, the age of the latest eruption at Mt. Tara naki, the Pyramid eruption, is discussed. 5.1 Introduction Lava dome-forming eruptions can last from a few days up to several decades. In most cases, the impacts of these typically small vol canic eruptions are forgotten within years or decades after the event. This leads to complacency within the surrounding population and a consequent underestimation of volcanic hazards. This is particularly the case where population pressures are high and the level of government regulation low. A lava dome located at high altitudes within a summit crater and in parts emplaced onto the outer steep flanks is in a metastable state. If the dome is growing, parts of it can collapse forming devastating Block- and-Ash Flows (BAFs). This hazard has been known for centuries and is included in haza rd risk analysis. However, th e hazard potent ial of a cool lava dome is mostly not regarded because the volcano is inherently considered dormant or even extinct. But a cool metastable lava dome could in fact collapse at any time, decades to centuries after its emplacement. In addition, if volcanoes are going to erupt Chapter 5 Non-explosive, Dome-forming Er uptions 122 most people expect large eruptions and major impacts on the environment and the community. But eruptions may not be that devastating and could be short-termed. Hence, hazard analysis should also take th ese events into consideration and eruption scenarios and conditions for these types of eruption need also to be investigated. Studies on the lava dome forming the current summit of Mt. Taranaki were carried out to reconstruct its original geometry, grow th dynamics and also zones of weakness within it. This in part enables a view into the possibly very common, but non-dramatic dome effusion events possible at Mt. Tara naki. Alongside studies of the typically hazardous sub-Plinian events, it is importa nt to examine the conduit and eruption conditions that lead to non-e xplosive dome-forming events. From the properties of the summit dome eruptive parameters such as magma storage depth(s), magma ascent rates, and eruption duration can be interpreted. Field observati ons and geochemical studies will show that the remnant summit dome represents an individual event and postdates the Tahurangi eruption of AD 1755. The youngest eruption of Mt. Taranaki is here named the Pyramid eruptive event which produced the summit lava dome (i.e. Pyramid Dome). 5.1.1 Lava Domes Lava domes are typically hemispherical m ounds of volcanic rock representing one or more lava units extruded above a vent and t oo viscous to flow far (Fink and Anderson, 2000). The extrusion of lava domes can occur above the central vent of a volcano (e.g. Mt. Taranaki, New Zealand; Gunung Merapi, Indonesia; Mt. St. Helens, USA), on the lower slopes of volcanoes either in isolation or in groups (e.g. Mt. Taranaki, New Zealand; South Sisters, Oregon, USA), and along fault lines or caldera ring faults (Mono Lake - Inyo Craters, California, US A; Cerro Chascon-R untu Jarita Dome Complex, Bolivia; Pago, New Guinea). Lava do mes were initially subdivided into four categories based on their morphology (Blake , 1990): up-heaved plugs; Pelean domes; low lava domes; and coulées. Additional cons ideration of dome surface textures and eruptive style were used to classify them into spiny or Pelean domes, lobate domes, platy domes and axisymmetric domes (F ink and Anderson, 2000). Spiny domes are steep-sided with smooth upper surfaces penetr ated by tall near-vertical spines (Fig. 5-1a). These domes often grow horizontally an d are largely confined by the geometry of Chapter 5 Non-explosive, Dome-forming Er uptions 123 their vents to display near-circular outlines (e.g. Mont Pelée, Martinique; Soufrière Hills, Montserrat, 1995-1 9 9 6 ). Lava domes w ith lower side-slopes that consist of individual lava lobes with an irregular plan outline are characterised as lobate domes (Fig. 5-1b). Their surfaces are partially smooth or covered by small blocks (e.g. Pinatubo, Philippines; Unzen volcano, Japa n, 1991-19 9 3 ). Platy domes (Fig. 5-1c) are represented by a low cross-sec tional profile, a bloc ky carapace, with sh allower fractures and more abundant surface ridges (e .g. Gunung Merapi, Indonesia, 1994-199 5 ; Obsidian Dome, Inyo Craters, California). A further type, axisymmetric domes, are characterised by a relatively low relief with near-level upper surface, an irregular plan outline and small blocks on their surfaces (e.g. Dacite Flow and Little Glass Mountain, Californian, USA). These morphologies can be related to eruptive conditions, where a series can be recognised following the trend of increasing extrusion/cooling rates and yield strengths from: axisymmetric, platy, l obate to spiny forms (Griffiths and Fink, 1997; Fink and Griffiths, 1998). Common to all la va domes is that they were formed by highly viscous lava. Within a re stricted range, however, they vary in their viscosity and rheology depending on their composition, phe nocryst, microphenocryst and microlite content, along with their di ssolved water concentrations. Figure 5-1 : Lava dome types: a) spiny (Rock Mesa ENE, Oregon), b) spiny-lobate (Mt. St. Helens lava dome, Washington, July 2004), c) lobate-platy (Big Obsidian Flow, Newberry, Oregon). Chapter 5 Non-explosive, Dome-forming Er uptions 124 5.1.1.1 Lava Dome Distribu tion in Taranaki Four lava domes occur on the lower slopes of Mt. Taranaki: The Dome to the NNW, Skinner Hill to the NW, and the Beehives (two lava domes) on the southern flank (Grant-Taylor, 1964; Neall, 1971; see Fig. 1-3) . The lava domes on the lower flanks are probably located on radial fissures of Mt. Taranaki (Neall, 1971). Two other domes occur farther from the stratocone: Pukeiti (between Kaitake and Pouakai) and German Hill (13.5 km to the NNE of Mt. Taranaki). Pukeiti and The Dome are directly situated along the overall volcanic lineament defined by Kaitake, Pouakai and Mt. Taranaki. Also, the Sugar Loaf Islands and German Hill are aligned parallel to this lineament. The present summit of Mt. Taranaki is part of a la va dome structure, partially filling a crater form. This has been attributed to the latest eruption of the volcano, thought to be a correlative of the Tahurangi tephra fall dated at AD 1755 (Stewart et al., 1996). 5.2 Field Observations 5.2.1 Tahurangi Eruptive Deposits Previous workers (Druce, 1966; Neall, 1972; Ne all et al., 1986; Plat z, 2001; Cronin et al., 2003) interpreted the AD 1755 Tahurangi tephra, as either a fall deposit or ash cloud surge deposits related to the emplacement of the summit dome. Although Neall et al. (1986) and Cronin et al. (2003) recognised that individual Block-and-Ash Flow (BAF) units on the northwestern sector of Mt. Ta ranaki are genetically related to the established tephrostratigraphy (cf. Druce, 1966), they did not consider the present summit dome to be a separate event. Field observations (Platz, 2001; this study) id entified two BAF deposits on the NW sector as being associated with the Tahurangi eruption (Fig. 5-2). The older Tahurangi BAF A is characterised by distinctive featur es compared to both Tahurangi BAF B and other BAF deposits of the Maero Eruptive Peri od. It extends over the entire NW sector, primarily within the Pyramid, Maero and Hangatahua catchments and is exposed from proximal (3.5 km) to distal reaches (12 km ; planimetric distance). The Tahurangi BAF A deposit is very poorly sorted and varies in thickness from >12 m at proximal locations to about 0.4 m at more distal locations . The angular to subangular clasts are predominantly grey to dark grey, dense andesite lithics with minor reddish, orange or Chapter 5 Non-explosive, Dome-forming Er uptions 125 brown-coloured accessory clasts. The most distinguishing sedimentological feature of the deposit is its subdivision into a clast-supported, virtua lly matrix-free, lower portion and a matrix- to clast-supported upper porti on with matrix contents of about 30-35% (see Fig. 3-4c and e). The internal boundary is transitional and the proportions of the two parts vary considerably. The upper massive portion also often displays coarse-tail reverse grading. Observed clast diameters vary considerably within and between described sections and reach boulder size up to 2 m in diameter. Only at two distal locations (outcrops 21 and 29; Fig. 5-3) the maximum clast size are cobbles (i.e. <256 mm) compared to boulder at proximal to me dial sections. The clasts show magnetic alignment1 at most locations, indicating a deposit ion temperature above 350 °C (Hoblitt and Kellog, 1979). The distinct twofold zoning of the deposit is apparent even at its most distal occurrence c.12 km from source. Figure 5-2 : NW sector of Mt. Taranaki showing the main deposition area of Maero BAF deposits. The extent of the Tahurangi BAF A is outlined (light grey ); unit B is only observed in the Maero Stream and is omitted for clarity. The rock-avalanche deposit (R AD) is outlined as observed on aerial photographs from 1959 (mid-grey) with an additional area based on field observations (dark grey). Contour interval is 20 m. The upper right inset shows the general slope inclinations. The squares are the locations of the sections reproduced in Figure 5-3. The basal contact of Tahurangi BAF A is sh arp and erosive in places. Where it is not erosive, it consists of a few centimetres of fine to medium grey ash overlying a black, humic paleosol. Charcoaled twigs or in situ charred tussock have been found in some 1 The natural remanent magnetisation (NRM) was measured using a portable fluxgate magnetometer (Platz, 2001 ). Chapter 5 Non-explosive, Dome-forming Er uptions 126 locations. Two substantially different degrees of development of directly underlying paleosols (0.5 cm humus layer vs. up to 30 cm of weathering) were observed which represent soil development on surfaces of different ages. A radiocarbon age (NZ5593) of charred tussock in the basal portion of the deposit was too young for the method at <250 yrs B.P. (V.E. Neall, unpubl. data). Three proximal sections (Ma 17-19; Fig. 5-3) located at slightly elevated positions on top of a buried lava ridge contain a poorly sorted, 0.5-0.7 m thick ash and lapilli unit with sharp lower erosive contacts that contain rip-up clasts and charred tussock stems. A thin (<0.5-1 cm) basal po cketing grey fine ash is also observed. Mapping the distribution of the deposit, together with ex amination of lithologies and distribution on paleomorphic surfaces, indicates that these fine-grained units probably represent an “overbank” or surge-like deposit, resulti ng from deflection of the main body of the Tahurangi BAF A by a major underlying lava ridge. The axial flow travelled from WNW to NW due to this deflection. At three locations (Ma 16, 44, and 46) the Tahurangi BAF A is overlain by another BAF unit. This unit has a more restricted distribution both in terms of runout (<6.5 km from source) and lateral extent, being only e xposed in the Maero Stream (Fig. 5-3). The unit varies in thickness from 0.6 to 1.1 m and c onsists of a grey to greyish brown ashy predominantly matrix-supported lapilli to bl ock breccia. In compar ison to the Tahurangi BAF A, it has consistently smaller maximum clast sizes (< 20 cm). The majority of clasts comprise grey angular, dense andesi tes (80-95%) along with reddish and brown coloured variants (5-20%). The aligned magne tic orientation of eight clasts at location 10 suggests a depositional temperature above 350 °C. The absence of soil development between the BAF A and B deposits as well as their similar lithologies supports their being derived from the same source. The consistently smaller maximum clast diameter in BAF B could also be caused by a different exposed cross-sectional area than for BAF A, i.e. centre vs. marginal flow facies or thermal fracturing of more coherent dome material where clast size is controlled by physical properties of the dome rocks. Both BAF deposits are younger than 250 yrs B.P. and represent the youngest BAFs known on the mountain. This corresponds to the youngest known ashfall unit on the volcano (potentially in part a co-ignimbrite ash), the AD 1755 Tahurangi Ash. Chapter 5 Non-explosive, Dome-forming Er uptions 127 Figure 5-3: Correlation of Tahurangi BAF A and B units and the rock-avalanche deposit across the Pyramid-Maero-Hangatahua area. The exposures are so rted by stream and planimetric distance from source (filled squares in Fig. 5-2). See figure legend for further details (RAD – rock-avalanche deposit). Magnetism was measured by a portable fluxgate magnetometer. For clarity, older exposed units were omitted. Outcrop numbers and profiles refer to Platz (200 1) except S04-13 3. Chapter 5 Non-explosive, Dome-forming Er uptions 128 5.2.2 The Present Summit Lava Dome The summit lava dome of Mt. Taranaki is located within a 420 m diameter partial crater with intact walls preserved from the NE th rough E to SW. The summit of the lava dome at 2518 m constitutes the highest point of Mt. Taranaki. When viewed from the southeast, the lava dome appears hemispheric with an extrapolated original diameter of 320 m (Fig. 5-4a). The crater floor is usually obscured by semi-permanent ice and snow but is exposed at the south en trance of the crater at the Okahu Gorge. Here the brown vesicular dome carapace directly overlies older hydrothermally altered rocks of the crater rim. The dome fills the crater to varying depths governed by how it impinges on the crater walls. The dome height is therefore estimated to be c.105 m above the crater floor level (assuming a flat crat er floor rather than a funnel shape). The south side of the dome at the Okahu Gorge is m odified by a small, wedge shap ed collapse zone, about 25 m wide at the top and 50 m wi de at the bottom, with the collapsed material deposited within the crater. The western margin of the dome forms the wall of a horseshoe-shaped amphitheatre open to the west (Fig. 5-4b-c). This implies >50% of the original dome is missing. The collapse scarp exposes the upper 45 m of the dome, with wall inclinations ranging from 55°-83° in a parabolic form (Fig. 5-4c). The scarp reveals a central 80 m-wide zone of hydrothermal alteration comprising white- to yellowish-brown rock; the lithologies at the northern and southern sides are grey to dark grey with reddish stained surfaces (Fig. 5-4b). A succession of three small coulée flow s extend from the northern side of the amphitheatre and an unstable bouldery field of talus from the scarp drapes the surface below the western face. On the northern side of the dome, two blocky lava flows each 4- 6 m thick occur between remnants of the form er northern crater rim with a blocky flow field extending down to 2230 m altitude. The carapace of the dome is characterised by brown, vesicular scoria that is mostly <1.5 m thick. The thin, vesicular lava dome rind is often penetrated by grey to dark grey, dense micro-vesicular lava which was extrud ed laterally and sub-vertically. The summit itself is located on a sub-vertical extrusion which shows weak columnar jointing (Fig. 5-4d). Other pinnacles are remnants of de nse interior dome rock exposed by partial collapse, e.g. in the southern part of the dome. The eastern side of the dome is smooth and is characterised by a thicker carapace (c.2 m). Chapter 5 Non-explosive, Dome-forming Er uptions 129 Figure 5-4: The remnant present summit dome. a) hemisphe rical shape of the dome as viewed from the SE; arrow points to a person for scale. Photographed by S.J. Cronin. b) dome amphitheatre as viewed from the W; the arrow points to the summit marker (251 8 m); the dashed lines mark the hydrothermally altered central dome portion. c) northern scar of th e amphitheatre showing listric faults. d) the summit marker is a sub-vertical extrusion penetrating the carapace; note weak columnar jointing; summit marker is highest point of the dome. e) orthophotograph of the summit region of Mt. Taranaki; outline shows mapped deposits associated with the lava dome; note the blocky lava flow to the N; crosses mark sample locations. f) the ‘Three Sisters’ (background) mark the NW border of the intra-crater collapse zone with resulting deposit still preserved in the crater (foreground). Chapter 5 Non-explosive, Dome-forming Er uptions 130 5.2.3 Rock-Avalanche Deposit A widespread bouldery deposit confined to the upper Pyramid – Maero Stream catchments on the NW flank of Mt. Taranaki (Fig. 5-2) has been photographed by the surveyor H.M. Skeet between 1898-1901 show ing a still fresh bouldery surface (Fig. 5-5). On present-day aerial photographs the su ppression of vegetation is clearly visible with the depression of the tussock zone mark ing the extent of this deposit (see Fig. 3- 4a). From 1959 aerial photographs an extent to 4.8 km from the summit is indicated, although the maximum extent is 5.3 km, ba sed on exposed and correlated units in Pyramid Stream (Fig. 5-3). The maximum depositional area is estimated at c.2×10 6 m2 . Figure 5-5: Black and white photograph taken between 1898 and 1901 showing the fresh bouldery rock-avalanche deposit (centre to right). The photograph is taken from the Round-The-Mountain- Track just west of Maero Stream from the top of a buried lava ridge. The view is NNW towards Pouakai. Photographed by the surveyor H.M. Skeet. The poorly sorted, predominantly matrix-s upported deposit is up to 2 m thick, consisting of a brownish grey, sandy matrix and angular to subrounded clasts with the largest (up to 4 m in diameter) concentrated at or near the surface often in boulder trains (Fig. 5-3). Maximum grain si zes range between pebbles and boulders, often with reverse coarse-tail grading through the unit. Weak planar fabr ic is observed in longitudinal exposures, including clast l ong-axes being aligned in a down-flow direction. The unit has a polylithologic clast assemblage, comprising mainly dense grey, Chapter 5 Non-explosive, Dome-forming Er uptions 131 greyish brown, pale brown and reddish ande site clasts. The lowermost 20-30 cm is lithologically similar, but is separated from the main body of the deposit by a pocketing pale brown, 0.5 cm thick fine ash layer. The deposits overlie either BAF deposits from the Tahurangi eruption ( AD 1755), separated by an 4-5 cm thick medial ash paleosol, or reworked volcaniclastic material (Fig. 5-3). The presence of a bouldery deposit and dome remnants in the summit crater indicate that the former intact dome collapsed with the bouldery unit being the resultant deposit. Based on the NRM of clasts from the depos it, which show a random orientation, it can be concluded that the deposition temperature was below 350 °C (Platz, 2001). This conclusion also implies that at least a portion of the dome that collapsed also had a temperature below 350 °C at the time of collapse. The precise date of the lava dome collapse is not known but it may have occurred in the 1880s. This is supported by Davis and Carrington’s summit crater map of AD 1885 which shows a similar crater configuration to the present day (N eall, 1973). The deposit does not result from a BAF, and is more appr opriately termed a rock -avalanche deposit. The deposit has a more polylithologic and subrounded clast assemblage than the typical predominantly monolithologic angular to subangular BAF units exposed in the area. 5.3 Sample Sites and Methods For a detailed study of the dome composition, 15 samples at various sites – dome top, margin and centre - were chosen (Table 5-1) . The following methods were used in the investigation: polarised micr oscopy (mineralogy and textural studies); XRF, Laser ICP- MS, EMPA (geochemistry); He-pycnometry, porosimeter, binocular microscopy (physical rock properties); GPS and Ar cGIS (mapping and volume calculations). Detailed descriptions of the methodologi es applied are listed in Chapter 2. Chapter 5 Non-explosive, Dome-forming Er uptions 132 Table 5-1. Sample location and description of studied summit dome rocks. 5.4 Results 5.4.1 Dome Volume Calculations The remains of the dome have been mapped using GPS and aerial photographs (Fig. 5-6a). Its original geometry and that of the crater was ex trapolated from the intact portions. The pre-collapse crater rim is inferr ed to have had near-circular dimensions of 420×40 5 m. The dome remnants occur within the crater as well as on the upper outer NW-flank. The portion of the dome within the crater is hemispherical with a near semi- circular base, whereas the portion on the flank is elongated, fo rming a strong semi- elliptical base. To fit the original 3-dimensional elliptical paraboloids were combined with an assumed even underlying surface. The elliptical paraboloids were combined to simulate the crater- and fla nk-portions of the original dome (Fig. 5-6b, c). The linkage of both halves coincides with the break in slope/crater rim. In addition, the centre point Chapter 5 Non-explosive, Dome-forming Er uptions 133 of the combined paraboloid (i.e. centre of the linkage line) roughly coincides with the centre point of the inferred crater geometry. The modelled dome has a maximum volume of 5.9×10 6 m3 , whereas from a 20 m resolution DEM 2 , the remnants of the dome comprise only 1.5×10 6 m3 (Fig. 5-6b, c). Bulk vesicularity of eight dome rock samples ranged between 12.4 – 22.8%. Using the mean bulk vesicularity (18.3%), a dense rock equivalent (DRE) volume for the modelled dome is 4.9×10 6 m3 . It should be noted that the depositional volume of the rock avalanche is approximately 2×10 6 m3 , if an average thickness of 1 m is assumed (deposition thickness varies between 0.5-2 m). This disparity with the missing dome volume (c.4.3×10 6 m3 ) could be due to a portion of the collapse having entered the Okahu Gorge or it may relate to an overestimation of the original dome volume. Figure 5-6: Reconstruction of the Pyramid Dome geometry. a) dashed white line illustrates the former ideal crater wall position; the solid line marks the inferred dome outline; the black dot is the assumed vent location at the break in slope. b) view of the remnant summit dome from the W. c) top view of the combined paraboloid with a composite elliptical base; dimensions of elliptical radii are given. d) side view of the inferred dome geometry; the dome remnants are in dark grey; the inferred underlying slope on the upper flank is estimated to be c.20°. 2 The remnant dome volume was calculated by J.N. Procter using ArcGIS software. Chapter 5 Non-explosive, Dome-forming Er uptions 134 Direct comparison of the Taranaki dome to published studies of dome extrusions is difficult since most domes rest on a flat or slightly inclined surface often with a near circular base. Moriya (1978) defined a relatio nship between height (h), diameter (d), and radius (r) of lava domes where the maximum dome height is calculated at: h = 0.32 d = 0.32 (2r) [Eq. 5-1] Equation 1 can also be expressed as h/r = 0.64 and the tangent being [Eq. 5-2] tan 0.64 = 32.6° [Eq. 5-3] where the corresponding angle is close to the angle of repose for loose rubble (Swanson and Holcomb, 1990). Although not all lava domes can be compared to a pile of loose material, this ratio can be used to asse ss whether a dome is metastable (Swanson and Holcomb, 1990). The dimensions of the remnant dome of Mt. Taranaki (here the dimensions of the hemispherical SE part of the dome: h=105 m, d=320 m), only slightly exceed the maximum h/d ratio with h/d = 0.33 (= 33.3 ° ) . For comparison, the dimension of other Taranaki flank do mes are listed in Table 5-2. Table 5-2. Comparison of the modelled summit dome to other Taranaki flank lava domes. 5.4.2 Mineralogy and Mineral Chemistry The mineral assemblage of the lava dome comprises plagioclase (19-30% ) , hornblende (5-18%), clinopyroxene (4-14%) , Fe-Ti oxides (c.1-4%), biotite (up to 1%), and traces of orthopyroxene and apatite. The major phase s (plg, hbl, cpx) range in grain size from macrocrysts to microlites. The rocks are porphyr itic, holo- to hypocrystalline with partly fluidal textures, and crystal contents rangi ng between 36 and 49% (vesicle free). The Chapter 5 Non-explosive, Dome-forming Er uptions 135 most abundant mineral phase, plagioclase (up to 31%), is euhedral to subhedral with both normal and reversed zoning, and partly si eved-textured cores. Hornblende (5-18%) is commonly observed as brown to reddish-b rown, euhedral to subhedral pheno- and microphenocrysts and frequently contains Fe-T i oxide, and plagioclase, as well as minor melt inclusions. Weak zoning of greenish- brown cores outward to brown rims are sometimes observed within single phenocrysts. Three types of hornblende can be distinguished, based on their degree of decomposition and opacitisation (Fig. 5-7). The most comm on (type 1) hornblende shows black crystal rims composed of minute Fe-Ti oxides, plag ioclase and pyroxene. Type 2 only occurs in sample SD1 and shows hornblende decomposition rims in an earlier stage, where individual Fe-Ti oxide crystals can still be identified and are not entirely fringing the crystal. Type 3 hornblende is partially or fu lly replaced by minute grains of plagioclase, Fe-Ti oxide, and pyroxene and is present as a minor constituent in all dome rocks. In one sample (SD6), type 1 hornblende show s a rim of pale brown glass approx. 9 µm thick, surrounding the reaction rim. Type 1 rim thicknesses are approx. 10 µm. A near Gaussian distribution for the mean reac tion rim thickness was noted (Fig. 5-8). Figure 5-7: Hornblende types. a) type 1 with continuous reaction rim. b) type 2 with discontinuous reaction rim; present only in sample SD1; note individual Fe-Ti oxide crystals are visible. c) type 3 partial to fully replaced hornblende crystals. d) type 1 with observed brown glass fringing the reaction rim; only observed in sample SD6. Scale bar is 100 µm in a) otherwise 25 µm. Chapter 5 Non-explosive, Dome-forming Er uptions 136 Geochemically, the hornblende phenocryst s and microphenocrysts are pargasite and titanian pargasite (Leake et al., 1997a, b, 2003), although core and rim compositions differ in all major element oxides. Octahedral Al VI , the Mg# 3 and the abundance of Na and K in the A-site is usua lly higher in cores than in rims. Microphenocrysts, on the other hand, have higher Mg# and tetrahedral Al IV and are substantially lower in Na and K than phenocrysts (Fig. 5-9). The increase of Mg# in microphenocrysts is likely caused by progressive oxidation of the magma wher e ferrous iron is converted to ferric iron. Figure 5-8: Histogram of type 1 hornblende reaction rim thicknesses averaged for crystals and entire samples. Clinopyroxene in Taranaki dome rocks usually occurs as green to pale green, euhedral to subhedral crystals or as coherent glomerocrysts with attached Fe -Ti oxides of varying sizes from macro- to microphenocrysts. Glass, plagioclase and Fe-Ti oxides are common inclusions. Clinopyroxenes are pred ominantly augitic in composition and some are diopsides (Morimoto, 1989). Gene rally, clinopyroxene cores show higher SiO 2 , MgO and CaO, and lower TiO 2 , Al 2 O 3 and FeO abundances than their rims. Fe-Ti oxides are either present as phenocryst s, groundmass constituents, or as inclusions in clinopyroxene and hornblende. Three di stinct groups are recognised according to their appearance as phenocrysts or inclusions (Fig. 5-10). Fe-Ti oxides in hornblende show a large range in Fe 3+ # from 76.1-90.0 with correspondin gly increasing Ti/Al ratios up to 1.7. Inclusions in clinopyroxene , however, show a narrow range in Fe 3+ # predominantly between 91.1 and 94.4 in a na rrow Ti/Al interval of 1.3-1.9. One outlier is noted with a Ti/Al ratio of 7.1. Phenocrysts, on the other hand, display a large variation in Ti/Al ratios from 1.2 to 5.4 over only a minor range in Fe 3+ # (90.1 to 96.6). 3 Mg# = 100[ Mg 2+ /(Mg 2 +Fe 2+ )]; minimum Fe 3+ estimated after Schumacher (1997). Chapter 5 Non-explosive, Dome-forming Er uptions 137 It is noted that phenocryst rim and core compositions generally remain nearly constant but this can be masked by cross-sectional effects. Figure 5-9: Hornblende compositions of the summit lava dome, Mt. Taranaki. a) Na+K (A-site) vs. Al IV . b) Mg# vs. Si; (c.p.f.– cation per formula unit). For co mparison are shown recalculated pargasitic hornblende compositions of Unzen volcano, Japan (Sato et al., 1999 ; Browne et al., 2006 ; Nakada and Motomura, 1999 ) [Unzen matrix refers to groundmass crystals], Soufrière Hills Volcano, Montserrat (Barclay et al., 1998 ; Rutherford and Devine, 2003 ), Mt. St. Helens, US A (Rutherford and Hill, 1993 ), Colima, Mexico (Luhr, 2002) and Cerro la Pilita, Mexico (Barclay and Carmichael, 2004). Biotite is a minor pheno- and microphenocryst often showing reaction rims or rounded margins. Based on Al IV and Mg# contents biotite lies close to the annite-phlogopite series and has 62-81% phlogopite component (Rieder et al., 1998). Orthopyroxene is a subordinate groundmass constituent and rarely observed as a microphenocryst. The general appearance is of weak brownish to tr anslucent colour and is slightly pleochroic with rounded margins. Orthopyroxene cores are sometimes composed of olivine. Apatite is commonly observed as an inclusion in clinopyroxene, hornblende and Fe-Ti oxides but is also a trace constituent in the groundmass. Its co mposition is relatively uniform with Cl and SO 3 contents ranging between 0.8-3 wt.% a nd 0.3-1.7 wt.%, Chapter 5 Non-explosive, Dome-forming Er uptions 138 respectively. Since F was not measurable, apatite crystals cannot be clearly classified as chlor- or perhaps fluorapatite. The dome groundmass is mostly hypocrystalline, consisting predominantly of plagioclase and minor clinopyroxene microlit es with small proportions of interstitial colourless glass. Large xenocrystic clusters up to 1.5 cm in diameter are also present in dome samples and predominantly comprise plagioclase, clinopyroxene, Fe-Ti oxide ± hornblende and interstitial pale brown glass. Figure 5-10: Compositions of Fe-Ti oxide phenocrysts and inclusions in clinopyroxene and hornblende. a) Al vs. Ti. b) Ti/Al vs. Fe 3+ # (c.p.f. – cations per formula unit). Cations are calculated on the basis of 32 oxygens. Major element compositions of melt in clusions trapped in hornblende and clinopyroxene were determined. Hornblende melt inclusions (n=7) are distinct from clinopyroxene melt inclusions (n=21; Fig. 5- 11). Melt inclusions in clinopyroxene show higher silica contents from 64.5-71.6 wt.% compared to 59.7-66.3 wt.% SiO 2 in hornblende melt inclusions. Chapter 5 Non-explosive, Dome-forming Er uptions 139 Figure 5-11: Glass compositions of inclusions in clinop yroxene and hornblende. Silica is used as differentiation index. Note that inclusions in different host minerals form separate groups. 5.4.3 Bulk Rock Composition Volcanic rocks erupted in the past 10 ka at Mt. Taranaki are high- K, low-Si and high-Si andesites from the main summit cone, with minor high-K basalts erupted from the parasitic cone of Fanthams Peak. The gene ral geochemical trend over the past 130,000 yrs of Mt. Taranaki’s history shows that the rocks became progressively more potassic and siliceous (Zernack et al., 2006), but wi thin any given eruptive period variations across the whole spectrum of compositions may occur, as will be shown for the youngest products. Major, trace element and isotopic compositions of the broad Mt. Taranaki rock suite are described in more detail by Price et al. ( 1992), Stewart et al. (1996) and Price et al. (1999). The present summit lava dome is homogeneous in its major element chemistry (Fig. 5-12) with only minor variations in SiO 2 (56.4-57.5 wt.%), CaO (6.9-7.1 wt.%), Fe 2 O 3 (6.6-7.1 wt.%), and Al 2 O 3 (17.4-17.8 wt.%). The dome rocks can be classified as a high-K, low-Si andesite (Gill, 1981; Le Ma itre et al., 1989). Trace element abundances are also uniform across the sample suite with minor variations for Cr (<1-28 ppm) and Sr (591-62 7 ppm). A single sample stands out from the suite because of its higher Chapter 5 Non-explosive, Dome-forming Er uptions 140 values of Mg#, Cr, Cu and Zr values, and lowe st Sr content (lower coulée flow, sample SD13). Barium abundances appear bimodal with f our samples containing 755-789 ppm Ba and six samples in the range of 982- 101 9 ppm. This variation, howe ver, was the result of an analytical error in the XRF data; the four samples were analysed as one batch which gave consistently similar but lower abundances in Ba. Subsequent Laser ICP-MS analysis confirmed that the results were anomalous. In addition, if the Ba anomaly was real, K 2 O and/or Rb should show similar covarian ce with Ba due to their similar ionic radii and affinity for certain mineral phases. They do not. Figure 5-12: Bulk rock compositions of the summit lava dome and the Tahurangi BAF A and B deposits. Dome compositions are distinct to Tahurangi rocks as illustrated for Al 2 O 3 (a), Mg# (b, d) Fe 2 O 3 (e) and Zr (f). Chapter 5 Non-explosive, Dome-forming Er uptions 141 The chondrite normalised REE pattern (Fig. 5-13b) is smooth with a negative slope for elements La to Er (La N /Er N = 6.04-7.38) with a sub- horizontal progression from Er to Y. Normalised La N /Y N ratios vary between 6.15 and 7.21 i ndicative of relative enrichment of LREE over HREE. The pa tterns of trace elements normalised to N-type MORB (mid ocean ridge basalt; Fig. 5-13a) show great er abundances for the large-ion-lithophile elements (LILE – Sr-U) compared to more compatible and high field strength elements (HFSE – Nb, Zr, Hf, Ti). It is observed that high field-stre ngth (HFS) elements Nb, Zr and Hf have similar values; Ti is offset to lower abundances. The elements Ti to Yb are depleted in comparison to MORB (i.e. <1). On the relatively smooth Ba-Yb line Nb forms a negative trough. It is noted that Py ramid Dome and Tahurangi BAF rocks show higher Rb abundances than older Mt. Tarana ki rocks (see Fig. 1-5). The comparison of summit dome rocks and Tahurangi BAF A+B rocks show clear differences in Al 2 O 3 , Mg#, Zr (Fig. 5-12) and Cr, although normalis ed trace element patterns are virtually identical (Fig. 5-13). Figure 5-13: Trace element patterns of the Pyramid Dome and Tahurangi BAF deposits normalised to N-MORB (a) and chondrite (b). Pyramid Dome rocks and Tahurangi BAF deposits show nearly identical trace element patterns. Fo r the light rare earth elements slightly higher abundances in Tahurangi BAF deposits are noted. Normalisation after Sun and McDonough (1989). Chapter 5 Non-explosive, Dome-forming Er uptions 142 5.4.4 Microstructure, Density and Permeabilitiy of Dome Rocks Textural studies of dome rocks revealed micro-cracks that occur around individual crystals and a succession of sub-parallel cr acks within the groundmass also occur (Fig. 5-14). The bulk density of eight dome rock samples ranged from 2.19 to 2.49 gcm -3 with bulk vesicularities of 12.4 to 22.8%. The mean so lid density of five powdered samples is 2.84 gcm -3 . The permeability ranges over one order of magnitude between 6.8×10 -1 3 m2 and 5.8×10 -1 4 m2 . This permeability is similar to that of dome rocks inferred from decompression experiments (Melnik and Sparks, 2002; Mueller et al., 2005). Figure 5-14: Texture of rock sample SD6. a) photograph shows sub-vertical, near parallel crack patterns and cavities. b) modified image of a) highlighting cracks and cavities in black. 5.5 Discussion 5.5.1 Lava Dome Emplacement and Growth In order to reconstruct emplacement mechan isms of the Pyramid Dome, the underlying surface needs to be modelled. From the inferred dome dimensions, a SE-NW cross- section is created to calculate the underlying slope. A planar inner-crater floor is assumed. The centre point of the body represents the inferred vent location and is positioned at the break in slope. The underlyi ng slope of the upper flank of the volcano is calculated at c.20° (Fig. 5-6c), from the distribution of the dome remnants around the present-day amphitheatre surface (c.26°). The presence of a coulée lava flow to the north of the amphitheatre and two thick lava flows on the northern dome side, along with th e hemispherical shape of the SE side of the dome, suggest combined exogenous (e xtrusive) and endogenous (intrusive) dome Chapter 5 Non-explosive, Dome-forming Er uptions 143 growth. Exogenous and endogenous phases are thought to be related to changes in magma extrusion and/or magma ascent rates, such as those observed at Mt. St. Helens (Anderson et al., 1995). At Mt . Taranaki, both phases of growth occurred, either simultaneously or alternately. This is primarily caused by the vent location at or near the break in slope where the downslope side fa vours the formation of stacked lava lobes, whereas the “upslope” or cr ater side promotes growth by inflation. Endogenous dome growth is promoted by the development of a thick cooled dome cr ust that hinders the breakout of lava, and endogenous dome growth probably occurred at a very slow rate. Exogenous growth by fast downslope extrusio n of lava lobes and endogenous growth with a slowly inflating inner crater dome side therefore probably proceeded simultaneously. Even though a reconstruction of all dome gr owth phases and their relative duration is not possible from the exposed dome remnants, it is likely that dome emplacement and growth took place during a single sustained event (Fig. 5-15). Single lava lobes extruded onto the upper W-NW flanks a dvanced at least 300 m (true distance) downslope from the vent. With continuous lava production lava lobes were stacked on top of each other. At the same time, sma ller volumes of lava were extruded onto the inferred crater floor to form a near hemispherical shape. The scars of the amphitheatre reveal relatively smooth upper rock faces wh ich may be indicative for the predominant endogenous growth in this part of the dome (Fig. 5-4b). Penetration of the dome carapace is observed on the NE side, about 120 m from the inferred vent location, where lava extruded forming two block lava flow s. The smooth rock faces only c.30 m from the discharge point, however, do not show a ny textural changes which may indicate exogenous growth. Based on field observation it is likely that the two NE lava flows as well as the small spine forming the highest point of the remnant dome were extruded during the final growth stage (Fig. 5-15e). The blocky flow field to the north is an indirect indication for high extr usion rates of the two lava flows where rapid fracturing of the flow surfaces generated abundant small blocks. A correlation between block size distribution and extrusion rate was observed at Mt. St. Helens and Unzen volcano where low extrusion rates cause less intense fl ow surface fragmentation promoting the development of large slabs. By contrast, high extrusion rates cause the formation of more abundant smaller blocks (Anderson et al., 1998). However, the steep north slopes of c.32° may have caused additional fr acturing due to mechanical and thermal processes. Chapter 5 Non-explosive, Dome-forming Er uptions 144 Figure 5-15: Lava dome growth patterns are illustrated in a- e. Exogenous and endogenous dome growth occurred simultaneously. f) demonstration of the inferred exogenously (dark grey) and endogenously (light grey) formed surfaces as observed on the dome remnants. The development of a vesicular dome carapace (SE-E dome portion) leads to two potential conclusions. Firstl y, the vesiculation of the lava surface at atmospheric pressure proves that the magma ascent was at a sufficient high rate to retain magmatic gases which later exsolved from the carapace. This implies that even the endogenous growth may have occurred relatively rapidly. Secondly, in ferred high magma ascent and extrusion rates point to a low viscosity magm a. Studies of lava dome surface textures during dome formation at Mt. St. Helens (1980-1 9 8 6 ) conclude that at high magma extrusion rates, the lava does not degas comp letely which later promotes vesiculation of the surfaces and the formation of a carapace (Anderson et al., 1995). The hemispherical dome shape was probably not only caused by endogenous growth but was also influenced by the confining crater wall. Here the initially extruded lava impinged on the crater wall (see title phot ograph), forcing itself upwards and hence contributing to endogenous growth. The assumption of a relatively low viscosity of the magma is born out by the presence of two lava flows near the top of the dome. In summary, the homogenous interior struct ure and chemical composition implies that the Pyramid Dome was formed during a singl e and short extrusive event. Associated Chapter 5 Non-explosive, Dome-forming Er uptions 145 lava flows and lobes superimposed on a dominantly hemispherical shape indicate simultaneous exogenous and endogenous dome growth, although the latter seems to have been dominant. The formation of a vesicular dome rind (carapace) along with inferred relatively high growth rates point to a low viscosity (and hot) magma extruded at relatively high magma ascent and extrusion rates. 5.5.2 Lava Dome Collapse Evidence from the natural remanent magnetisat ion of clasts within the youngest rock- avalanche deposit suggests deposition temperatures <350 °C. This implies that since the summit dome is the obvious source of the landslide, it must have cooled to below 350 °C before the collapse occurred. This imp lies a considerable time lapse between emplacement and collapse. Additionally, some time is also implied by the presence of an extensive hydrothermally altered central zone in the central dome portions. The presence of fractures and faults transecting the entire dome is a further indication for its brittle state and hence relatively low temperature at the time of collapse (see Fig. 5-4c and below). If they were created while th e dome had a ductile interior then those fractures would likely be annealed or po ssibly deformed by subsequent flow. The presence of listric faults also implies formation in a brittle near-solid state, which is most likely in a post-eruption cooled conditi on (ie. a ductile dome interior can be excluded). The northern part of the collapsed amphitheatre along with the curvature of the amphitheatre itself indicates that the dome failure and disintegration occurred along these predefined fault planes (see Fig. 5-4c). Micro-cracks described around crystals in dome-rock samples are interpreted to represent rapid contraction of groundmass and crystals during cooling. Sub-parallel crack patterns are likely to be the result of stretching, indicating duc tile failures rather than brittle fractures (cf. Smith et al., 2001). Large gravitational forces (e.g. extrusion on steep slopes), viscosity in crease during cooling, and st rain intensification around rigid crystals are probably the main reasons for such crack development (also known as cavitation by Smith et al., 2001). Hence, th ese micro-cracks developed both syn- and post-eruptively. Physical weathe ring (including water and ice in teraction) may have also enhanced crack development. With the pres ence of abundant micro- cracks the measured permeability (and porosity) values shown a bove are probably much higher than the Chapter 5 Non-explosive, Dome-forming Er uptions 146 actual permeability at the time of emplacement. This supports the interpretation of the observed dome carapace. At the time of emplacement, gas escape was low, hence a vesicular dome rind could develop, which is probably also linked to high extrusion rates. The arcuate amphitheatre on the western summit dome face could have been formed syn- or post-eruptive. A similar arcuate su rface structure is observed on The Dome. Both lava domes are rest ing on steep slopes and these downslope opened arcuate structures may reflect subsidence, from the upper vent area by magma withdrawal. In the case of the summit dome, the arcuat e shape could also trace the underlying collapsed western crater structure4 . The reasons for dome failure may have included the development of a weak inner core structure through hydrothermal alteration above the vent zone. This instability was probably also exacerbated by cooling and post-emplacement subsidence fractures. The steep and unconsolidated substrate beneath the up to 100 m thick dome was likely an important pre-conditioning factor for dome instability. The failure event could have been triggered by an earthquake, heavy rain or even climatic disturbance (e.g. snow load). Distant tectonic earthquakes (up to 100 km) have been shown to destabilise growing lava domes (Walter et al., 2007). The occurrence of frequent earthquakes within a 25 km radius SW-NW of Mt. Ta ranaki could have caused sudden fracture initiation and propagation within the western part of the dome. Based on the location and curvature of the am phitheatre scar and the erosion gullies on the slope, four collapsed sectors are identified (Fig. 5-16) in an anticlockwise direction: I-northwest, where only the upper portion wa s removed; II-WNW with a collapse over the total dome depth; III-S S W with a smalle r collapse through the entire depth; and IV- south with only a superficial collapse with collapsed material remaining in the crater (see Fig. 5-4f). These sectors may have failed independently. Multiple deposition episodes are indicated by a thin pocketing fine ash layer within the rock-avalanche deposit observed at only two proximal locations (outcrops 47 and 48; Fig. 5-3). This demonstrates a small time break occurred betw een avalanches that allowed an elutriated cloud of ash from the initial collapse to settle. The thin silt layer is absent in distal deposits, which may indicate that th e initial collapse was only minor. 4 Although the western crater rim is breached and at least in parts collapsed to depths of >100 m (see Okahu Gorge) there is no other field evidence which may indicate a pre-dome, arcuate collapse structure of the western crater area. Chapter 5 Non-explosive, Dome-forming Er uptions 147 Figure 5-16: Reconstruction of the summit dome failure. a) erosion scars on upper flank (white arrows) as well as the curvatures of the amphitheatre, and the scars to the SW and S define the geometry of individual collapse sectors. b) dome geometry with individual sectors I-IV and their flow directions. c) cross-section of the dome showing the dome remnants (grey), and the disintegration of dome rocks along listric faults. Although it is impossible to fully reconstruct a time frame for the dome collapse process, the sequence of dome collapse even ts can be hypothesised. Partial sector collapse (from sector I) likely occurred before the major collapse from sector II because the erosion scars of collapse II are not obliter ated or modified by collapse I (Fig. 5-16a). This is also consistent with the assumption from deposit evidence above that stage 1 deposits are derived from a lower volume, part ial sector collapse. The collapse of zones III and IV are probably concurrent with the ma jor collapse event. The collapse of sector III is closely related to th e partial collapse of the older underlying, hydrothermally altered crater lava flow deposits at the entrance of the Okahu Gorge (Fig. 5-4e). The unstable bouldery talus field within the amphitheatre postdates the major collapse (II). Similar arcuate amphitheatres developed as a result of dome collapses have also been observed, for example, at Dikii Greben’ vo lcano, Kamchatka (Ponomareva et al., 2006). Here arcuate, concave uphill ridges are also observed which are against flow direction. The dome collapse is interpreted to have occurred after the main dome-building phase (Ponomareva et al., 2006). Chapter 5 Non-explosive, Dome-forming Er uptions 148 The polylithologic character of the resulting rock-avalanche deposit is primarily due to firstly, the variety of rock domains with in the dome that have undergone different degrees of vesiculation, oxidation, cooling and hydrothe rmal alteration. Some dome substrate material and also surficial accidental clasts may have also been incorporated into the avalanche (Fig. 5-16c). In contrast, many of the BAF deposits are monolithologic (e.g. Tahurangi BAFs A and B) and emplaced at temperatures above the Curie Point. The monolithologic nature of these units testifies to collapse during dome growth, soon after lava effusion that allowe d little time for hydrothermal alteration and differential degrees of oxidation. 5.5.3 Estimation of Eruption Parameters 5.5.3.1 Magma Source Areas The brittle-ductile transition zone (BDTZ) r ecognised at 10 km beneath Mt. Taranaki is interpreted to represent magmatic intrusions (Sherbur n and White, 2005). This implies magma storage at mid-crustal levels. The BDTZ represents a rheological boundary and often coincides with the level of neutral buoyancy explai ned by density equilibration between magma and its surrounding country rock (e.g. Holm, 1995; Marquez et al., 2001). Possible magma storage depth(s) can also be inferred from hornblende crystallisation pressures by applying the aluminium-in-hor nblende geobarometer. This geobarometer for calc-alkaline plutonic complexes was fi rst empirically developed by Hammarstrom and Zen (1986) and progressively experiment ally refined by Hollister et al. (1987), Johnson and Rutherford (1989), Blundy and Ho lland (1990), Thomas and Ernst (1990) , Schmidt (1992) , Anderson and Smith (1995). The barometer is experimentally calibrated to the mineral assemblage quartz + K-feldspar + plagioclase + biotite + Fe-Ti oxide + sphene over a pressure range of 2-8 kbar with an estimated error of ± 1 kbar. Because the required conditions for this ge obarometer are not met for Mt. Taranaki rocks (quartz, K-feldspar, and sphene are microscopically absent; the exception is sphene in some xenoliths; Gründer, 2006), th e total pressure estimates yielded are maxima for K-feldspar-absent samples. The role of sphene still remains unknown; an overestimation in calculated pressure may therefore be up to 1.5 kbar (c.f. Hollister et al., 1987; Johnson and Rutherford, 1989; Anderson and Smith, 1995). Chapter 5 Non-explosive, Dome-forming Er uptions 149 The data show a total calculated maximum pressure range for hornblende between 4.5 and 6.7 kbar based on minimum ferric estimates in hornblende after Schumacher (1997) [hornblende calculations based on 23 oxygens yield lo wer pressures by 0.42 kbar]. Assuming an overestimation of 1.5 kbar, hornblende crysta llisation pressures of 3-5 kbar 5 would correspond to depths of 9.5-17 km 5 . Two maxima are observed for hornblende core formation, the first between 4 and 4.5 kbar and a second between 3 and 3.5 kbar (Fig. 5-17). Hornblende growth co ntinues to pressures down to 3 kbar as calculated from rim compositions. Highest crystallisation pressures are recorded for some rim compositions and are the result of reverse zoning in some hornblende phenocrysts, probably caused by an increase in pH 2 O (Luhr and Carmichael, 1980). Figure 5-17: Aluminium-in-hornblende geobarometer shown as histogram for hornblende phenocrysts (core and rim) and microphenocrysts. Calculated after Johnson and Rutherford (1989) and corrected by -1.5 kbar. For temperature estimates of hornblende formation, the amphibole-plagioclase thermometer of Holland and Blundy (1994) is used, which can be applied over a temperature range of 400-1000 °C at 1, 5, 10 and 15 kbar. The uncertainty of this thermometer is ±30-40 °C, however, it may be higher if the iron-ri ch hornblendes are oxidised and the ferrous/ferric ratio in hor nblende is not well defined. Temperature estimates for 19 hornblende-plagioclase pair s range between 867-928 °C at an assumed pressure of 5 kbar (c.17 km). Eight pairs are restricted between 900-920 °C. It is 5 Calculations of the lithostatic pressure include a 2 000 m high edifice, i.e. 1.1 kbar correspond to 2 km depth below sea level (see Fig. 5-18). Chapter 5 Non-explosive, Dome-forming Er uptions 150 assumed that hornblende and plagioclase are in equilibrium since no rounded edges or resorption of plagioclase inclusions were observed. The data of this study are consistent with the literature (see above), and suggest that hornblende in the dome rocks started crystall ising at approximately 17 km depth (see also Fig. 5-18). The two hornblende populati ons likely reflect two crystallisation pressures, and hence, two storage levels at c.9.5 and 14.5 km depth, at temperatures of about 900 °C. The crystallisation of hornblende at mid-crustal levels is generally in agreement with the proposed model of the Mt. Taranaki magma evolution by Stewart et al. (1996) where ascending hydrous basaltic andesite melts re-enter the hornblende stability field at shallower depths. Figure 5-18: Comparison of calculated hornblende crystallisation pressures for various Mt. Taranaki rocks and xenoliths. Granodiorite xenoliths contain all required mineral phases for the Al- in-hornblende geobarometer (Johnson and Rutherford, 1989 ) and therefore were not corrected. Note that some hornblende crystals of hornblende gabbros and hornblende-pyroxene gabbros indicate crystallisation below (<1 kbar) the inferred hornble nde stability limit. Data of Gründer (200 5) and A.V. Zernack were recalculated us ing the procedure described above. There are two possibilities to explain the pr esence of two hornblende populations within an erupted magma: 1) the two hornblende populations belong to the same magma which stagnated at two different levels within the lower and middle crust, or 2) it represents mixing of two hornblende-phyric magma ba tches where the deeper-stored magma intrudes into the high-level source area. Alt hough scenario 1 appears to be the simplest Chapter 5 Non-explosive, Dome-forming Er uptions 151 explanation, its disadvantage is that the final ascent to the surface from the high storage level has to be initiated by internal processes, i.e. the threshold of neutral buoyancy must be overcome by internal processes such as crystal segregation and fractional crystallisation processes. The most common trigger of mobilising magma is, however, the input of heat by a more buoyant magma (e.g. Soufrière Hill s Volcano, Montserrat [Murphy et al., 1998], Mt. Pinatubo, Indonesi a [Pallister et al., 1992], Augustine volcano, Alaska [Roman et al., 2006]), and th is favours scenario 2. Scenario 2 requires homogenous mixing since neither macroscopi c nor microscopic textures imply the existence of two magmas. Despite this, sc enario 2 is favoured. Magma mixing of two magmas at depth can be petrographically indistinguishable because the mineral assemblage in hornblende-phyric magmas of ba salts, basaltic andesites and andesites is nearly identical (see Chapter 7). Homogenous mixing is even more likely when the melts of both batches contain sufficient water (indicated by hornblende crystallising), and hence have relatively low viscosities. In addition, the extrude d Pyramid magma is geochemically distinct from any other dome rocks extruded during the Maero Eruptive Period, including the immediatel y preceding eruption. Given the general trend of K 2 O enrichment in Mt. Taranaki rocks over time the latest magma is anticipated to have highest values for K 2 O (and also SiO 2 ). Instead it has relatively low K 2 O and SiO 2 contents, along with higher Al 2 O 3 wt.% and lower Mg# than other dome-forming magmas (see also Fig. 7-2 and Fig. 7-6) . This suggests mixing of at least two geochemically distinct magmas: 1) a low- level magma (c.14.5 km depth) basalt/basaltic andesite magma with high Al2 O 3 and Mg# and low SiO 2 and K 2 O abundances (similar to those observed at Fanthams Peak); a nd 2) a high-level more evolved (probably andesitic) magma. The mixing of these two ap pears to generate the unique properties of the Pyramid lava dome within the Maero series. The low dome volume may be consistent with a small volume, but mobile magma at depth. 5.5.3.2 Magma Storage at the Level of Neutral Buoyancy Magma storage depths are often controlled by their level of neutral buoyancy. In order to overcome this threshold and rise to the surface, the magma has to reduce its density. On the basis of hornblende-geobarometry and geophysical evidence, the mid-crustal storage region occurs at c.9.5 km depth at or near the BDTZ. Although smaller magma Chapter 5 Non-explosive, Dome-forming Er uptions 152 batches may be stored temporarily at even lower pressures within a 5 km-diameter dyke complex inferred below the volcano (Locke and Cassidy, 1997; Sherburn et al., 2006), a major magma source area at shallower de pths is unlikely since there were no geophysical indicators in the 1993 and 2001-2002 surveys (Cavill et al., 1997; Sherburn and White, 2005). The main reason for final magma ascent was likely the injection of hot and probably less evolved magma from depth, as discussed in the previous chapter. However, another process may have preceded this event. Frac tional crystallisation may also reduce magma density, although it requires more time than th e injection process. The crystallisation and separation of mafic mineral phases (i .e. hornblende and clinopyroxene) would significantly change the remaining melt density. The presence of abundant microscopic and macroscopic cumulates of hornblende and clinopyroxene ± plag ioclase show that this process was at least partly in operation. Extensive hornblende crystallisation is evid ent by geochemical changes of melt and Fe- Ti oxide phases. Three compositionally distin ct groups of Fe-Ti oxi des are recognised with increasing Fe 3+ #, as inclusions in hornblende and clinopyroxene, and as phenocrysts (Fig. 5-10). The same grouping is re flected in Al contents. Fe-Ti oxides are sensitive to changes in the magmatic system, especially fO 2 and temperature, which are geochemically recorded even on short eruptive time scales, similar to reaction rim formation of hornblende crystals during decompression (e.g. Rutherford and Hill, 1993; Venezky and Rutherford, 1999). The rapid drop of Al 2 O 3 in Fe-Ti oxides at near constant TiO 2 contents is primarily caused by hornblende crystallisation, since it is composed of 11.0-13.5 wt.% Al 2 O 3 . Clinopyroxene, co-crystallising at a later time, contains Fe-Ti oxide inclusions of an intermediate Al 2 O 3 -TiO 2 composition (Fig. 5-10). The suite of Fe-Ti oxide phenocrysts has a wide range in Ti/Al ratios, and the progressively lower Al2 O 3 contents signal the onset of plagioclase crystallisation and simultaneously the cessation of hornblende formation. Plagioclase crystallisation is sensitive to water pressures and could indicate an advanced stage of magma degassing, also supported by increased Fe 3+ #. The sharp change in Ti/Al ratios for groundmass Fe- Ti oxides probably indicates crys tallisation at reduced pressures and water contents, i.e. crystallisation during magma ascent. It appears that melt inclusions record a similar history; first the melt experi ences dominant hornblende crystallisation evolving from c.60 to 66 wt.% SiO 2 , then hornblende crystallisation and growth ceased and dominant clinopyroxene (and plagioclase) crystallisation governs the melt evolution, shifting melt Chapter 5 Non-explosive, Dome-forming Er uptions 153 silica contents to higher values up to 72 wt.% (Fig. 5-11) . Fractionation by gravitational settling of both mafic phases produces a restite layer at the bottom of the magma body; observed microscopic and macroscopic cumulates could therefore represent either co- genetic assemblages or xenoliths if in th e latter case the ascending magma penetrates other source areas. 5.5.3.3 Inferences for Hornblende Stability The occurrence of hornblende in many subduction-related volcanic ro cks is associated with water-rich magmas, and hence higher potential for explosivity. Understanding these magmatic conditions has been a focus in experimental petrology over the past two decades. Studies were directed towards well documented eruptions of hornblende- andesite (e.g. Soufrière Hills Volcano, Mont serrat; Barclay et al., 1998; Rutherford and Devine, 2003) and dacitic volcanoes such as Mt. St. Helens, USA (Rutherford and Devine, 1988; Rutherford and Hill, 1993), Mt . Unzen, Japan (Sato et al., 1999; Holtz et al., 2005), and Mt. Pinatubo, Ph ilippines (Prouteau and S caillet, 2003). However, experimental studies on natu ral hydrous, hornblende -bearing basalts, basaltic andesites and andesite are less common (Moore and Carmichael, 1998; Barc lay and Carmichael, 2004). Experiments based on natural rocks w ith pargasitic hornblende by Rutherford and Devine (1988), Foden and Green (1992), Rutherford and Hill (1993), Moore and Carmichael (1998), Sato et al. (1999) and Ba rclay and Carmichael (2004) are used for comparative purposes in this study. The stability of synthesized pargasite was first studied by Boyd (1959) and later by Holloway (1973) who determined a minimum stab ility of 1 kbar at 1000 °C. In contrast to the conclusions of Westrich and Hollo way (1981), Lynkins and Jenkins (1992) found that the stability of pargasite in equilibri um with olivine and orthopyroxene in the system Na 2 O-CaO-MgO-Al 2 O 3 -SiO 2 -H 2 O was stable at low pressures (0.4 kbar) at 750 °C. The crystallisation of groundmass pa rgasite during the 1991-19 9 5 Unzen eruption also occurred at a relatively low stability limit of 0.98 kbar at 900 °C and 0.48 kbar at 820 °C (Sato et al., 1999), wh ereas Mt. St. Helens pargasite was shown to be stable above c.1.1 kbar and 1.7 kbar at 830 °C and 910 °C, respectivel y (Rutherford and Devine, 1988). Sato et al. (1999) argue that th e reduction in pargasite stability in dacite Chapter 5 Non-explosive, Dome-forming Er uptions 154 systems at Unzen volcano relative to Mt. St . Helens are probably due to higher SiO 2 contents and higher K 2 O/Na 2 O ratios in the Unzen groundmass compositions. Andesites (Colima volcano, Mexico) and basa ltic andesites (Masco ta volcanic field, Mexico) were investigated by Moore an d Carmichael (1998) who determined hornblende stability limits of 0.5 kbar at 950 °C and 1 kb ar at 1000 °C, respectively. Barclay and Carmichael (2004) determined hornblende stability limits of 0.5 kbar at c.1010 °C, or 0.5 kbar at 950 °C in the presence of biotite, in the hydrous basalt of Cerro la Pilita, Mexico. Taranaki hornblende is char acterised by a considerable range in Mg# but with much narrower variations in AlIV compared with hornblende compositions from Unzen volcano (Fig. 5-9). Taranaki ho rnblende also has higher K 2 O contents with high Na and K cation occupancy of the A-site vacancy (F ig. 5-9). Only Colima volcano and Cerro la Pilita (Mexico) have hornblende with sim ilar compositions. In terms of hornblende as well as whole rock composition, Mt. Taranaki magmas are closest to Mascota basaltic andesites than to any other previously listed volcanic centres. It is therefore assumed that Taranaki hornblendes are stable to simila r pressures of as low as 1 kbar (i.e. 2 km depth; Fig. 5-18). This hornble nde stability limit is therefore compatible with temporary magma storage between c.9.5 and 2 km depth. 5.5.3.4 Magma Ascent Rate The formation of hornblende reaction rims was investigated in decompression experiments using Mt. St. Helens dacite compositions, and these clearly showed a correlation between rim thickness and incr easing time spent outside the hornblende stability field (Rutherford and Hill, 1993) . Hornblende crystals in isothermal decompression experiments developed 11 µm thick rims after 7 days at 1.6-0.02 kbar and 10 days at 2.2-0.02 kbar, and rim thickn esses of 15 µm after 8 and 11 days, respectively (Rutherford and Hill, 1993). Rece nt experiments on dacites (840 °C) of Redoubt volcano, Alaska (Browne and Gardner, 2006) indicate that hornblende could linger much longer outside its stability field, about twice the time given by Rutherford and Hill (1993). Observed average rim thickness per sample of Taranaki dome rocks is approximately 12 µm. If compared to the a bove experiments, this would correspond to a period of about 7 or 10 days outside the stability field (cf. Rutherford and Hill, 1993) or Chapter 5 Non-explosive, Dome-forming Er uptions 155 17 days (cf. Browne and Gardner, 2006). Th e conversion of hornblende rim thickness into time (i.e. time for reaction formation) and the subsequent estimate of magma ascent rate (assuming the hornblende pressure lim it is known) requires one further condition: that the hornblende stored within a dome must cool immediately to prevent further reaction rim formation. In other words, if the lava forming the summit dome contained hornblende crystals with 12 μm thick rims, it would have had to collapse immediately (forming a BAF) so that it cooled quickly enough to preserve the hornblende texture and to prevent further reaction rim growth. However, given that the samples analysed are from the intact portion of the lava dome, they might have cooled to ambient temperatures quickly. Hence, the observed ho rnblende rims could have been formed within the dome after extrusion, rather than during ascent. This is supported by the fact that only one sample (SD1; small spine pe netrating the dome rind – and hence rapidly cooled) shows a weak breakdown reaction (type 2, Fig. 5-7b). In Rutherford and Hills’ (1993) constant-rate decompression experime nts (1.6-0.02 kbar), hornblende first showed reaction rims after 4 days outside the stability field. Browne and Gardner (2006) suggest that a time of 7 days is required for reaction rim formation. In the latter case, single-step and multi-step experiments we re carried out, includi ng stalling times at various pressures in order to investigate various hornblende text ures, e.g. crystal size variations and crystal rounding (Browne and Gardner, 2006). However, Taranaki hornblende compositions and inferred magma temperatures are closer to Mt. St. Helens than to Redoubt volcano so the data from Ruth erford and Hill is considered to be more applicable here. Hence, if Ta ranaki hornblendes are stable up to 1 kba r and are able to spend up to 4 days outside the stabilit y limit before showing any signs of decompression-breakdown reactions, then the magma ascended in a 4 km long dyke or conduit at rates of 0.012 ms -1 . In this scenario the magma ascent rate calculation (1000 md-1 ) is a minimum. 5.5.4 Eruption Duration Effusive, lava dome-producing eruptions with or without explosive, phreatic, vulcanian to sub-Plinian phases, are often long-term , ongoing events that may last for several months to years, or even up to decades, as recently observed at many volcanoes, e.g. Mt. Merapi (Indonesia; 1992- 200 2 ) , Santa Maria (Guatema la; 1922-today), Soufrière Chapter 5 Non-explosive, Dome-forming Er uptions 156 Hills Volcano (Montserrat; 1995-today), Unzen volcano (Japan; 1991-19 9 5 ) , Mt. St. Helens (USA; 1980-198 6 , 2005-today), and Mt. Lamington (Papua New Guinea; 1951- 19 5 6 ). However, some lava dome-producing erup tions can instead be short, lasting for only a few weeks, or even for only a few days (Table 5-3). Table 5-3. Historic lava dome eruptions compiled from Newhall and Melson (1983 ) and the Smithsonian Institution Catalogue 1 . Laboratory experiments by Blak e (1990), Griffiths and Fink (1997), Fink and Griffiths (1998) and Lyman et al. (2004) show that lava dome morphologies can be related to eruptive conditions. If the init ial parameters such as morphology, extrusion rate, rock composition and underlying slope are known, then the yield strength of an active lava dome can be estimated. However, the study of historic lava domes is more complicated since only the geometry and composition is known. Since the yield strength of the Mt. Taranaki lava dome is unknown, the assumption must be made that the inferred dome geometry reflects its yield strength (B lake, 1990). Based on a homogenous dome composition and a short emplacement duration, the yield strength σo is estimated using the equation (Blake, 1990): R gH ρσ 2 0 )016.0323.0( ±= [Eq. 5-4] where H and R are dome height and radius, ρ density and g gravitational acceleration. The density needs to be estimated carefully since it has a considerable impact on the resulting yield strength. Solid densities as well as bulk densities of dome rocks are assumed to over- and underestimate the lava dens ity at the time of ex trusion, i.e. at 900 Chapter 5 Non-explosive, Dome-forming Er uptions 157 °C. The mean bulk density of 2321 kgm -3 , approximately equal to the density of the andesitic glass6 , is not realistic and should be clos er to the solid density. The lava density (at 900 °C and 1 bar) should be 10% less than the solid density to account for thermal expansion of glass and crystals and the presence of cavities and/or vesicles. A lava emplacement density of 2555 kgm -3 will therefore be used for further calculations. Since the studied dome is resting in part flat and in part inclined surface, and its surface area was described by three radii, a mean ra dius of 203 m and a he ight of 105 m were applied to estimate a yield strength of σ 0 = 4.4×10 5 ± 2.2×10 4 Pa. This estimate is in the same order of magnitude to that of other andesitic dome-forming lavas (Fink and Griffiths, 1998). Recent laboratory experiments of Lyman et al. (2004) show that effusion rates and hence, the eruption duration of historic domes can be estimated. To calculate the effusion rate Q and eruption duration t e, the following equations are used based on Lyman et al. (2004): 3 0 )/( 1 σρψ Δ= g t Q s B [Eq. 5-5] Q V te = [Eq. 5-6] where g is the gravitational acceleration, ∆ρ is the density difference between magma/lava and the environment, σ 0 is the isothermal yield strength, ψB is the dimensionless ratio of the timescale of solidification (t s) and advection (t a), and V is the volume of the lava dome. The values of parameters used in the equations are listed in Table 5-4. The minimum extrusion rate, Q min, for the modelled summit dome is estimated to be 6.0 m3 s-1 (5.2×10 5 m3 d-1 ) which gives a maximum eruption duration of 11.5 days. The calculated extrusion rate compares well with recent dome-forming eruptions of Mt. St. Helens, Washington (1980-19 8 3 : Q max =20.1 m 3 s-1 =1.74×10 6 m3 d-1 ; Anderson and Fink, 1990), Mt. Lamington, Papua New Guinea (Q max =22.2 m 3 s-1 =1.92×10 6 m3 d-1 ; Taylor, 1958) and Unzen volcano, Japan (Q max =15.1 m 3 s-1 =1.3×10 6 m3 d-1 ; BGVN 17:11). If a maximum time of 4 days is assumed (from the lack of decompression-induced reaction rims of hornblende; Rutherford and Hill, 1993) then a minimum extrusion rate of 17.2 6 The Tahurangi Ash glass composition is inferred to be similar to the dome glass composition. Its mean density is calculated to be 2355 kgm -3 at 900 °C, 1 bar and 1 wt.% H 2 O. Chapter 5 Non-explosive, Dome-forming Er uptions 158 m3 s-1 (1.5×10 6 m3 d-1 ) is estimated. Hence, the calculated extrusion rates using the two different methods are consistent. Another eruptive parameter, the vent diamet er, may be estimated on the basis of magma extrusion and ascent rates. Usi ng the minimum values of 0.012 ms -1 and 6.0 m 3 s-1 , a cylindrical vent diameter of 26 m immediatel y below the dome results (Table 5-4). The vent diameter is about one third the size of the c.80 m wide hydrothe rmally altered zone within the amphitheatre. It is impossible to estimate the vent-shape changes with depth, but if a narrower conduit diameter of 10 m was present, much higher magma ascent rates of 0.189 ms -1 result. Table 5-4. Physical parameters and results for equations 5-4 to 5-6. 5.5.5 Approximation of the Time of Eruption Although it has been shown that the present summit dome represents a single eruptive event, the date of the erup tion still remains unknown. It is thought that the collapse of the dome may have occurred prior to AD 1885 according to Davis and Carrington’s summit crater map (Neall, 1973). Druce (1964) argued that a major landslide occurred in the early 1880s: “Stability on this side of the mountain has not been regained since; successive landslides, initiated in the area between the Turtle and Bobs Ridge by torrential falls of rain have poured down the upper slopes building new fan surfaces”. Chapter 5 Non-explosive, Dome-forming Er uptions 159 According to this it can be assumed that the eruption date occurred between AD 175 5 and 1885. The timing of eruption can be further constrained by examining the cooling history of the dome. It is inferred that the dome collap se occurred when its temperature was below 350 °C, as determined by NRM of clasts within the rock av alanche deposit. To approximate the time required to cool the dome centre to 350 °C, a simple conductive cooling model can be applied. Jaeger (1968) proposed a non-dimensional time for heat transfer across a static body: 2a t t D κ= [Eq. 5-7] where κ is the thermal diffusivity, t is the tr ansfer time (i.e. the time elapsed since extrusion) and a is the radius of the sphere (i.e. the ra dius simulating the dome). The value of tD =0.2 corresponds to 50% co oling by ambient air (Sim mons et al., 2004) whereas tD >10 describes the heat transfer to be practically complete (Jaeger, 1968). Using the parameters listed in Table 5-5, approximately 26 years are needed for the dome to cool to half its initial temperature. As the dome emplacement temperature was probably around 900 °C, it would corre spond to c.450 °C, or 17.1 °Cyr -1 . If this rate is also applied to the remaining 100 °C (i.e. 450 °C-100 °C=350 °C), then a cooling time of c.32 years is necessary. An important factor for accelerated cooling is rainfall- quenching, especially at Mt. Taranaki. Here the mean a nnual rainfall at the North Egmont Visitors Centre (852 m) is 7.5 myr -1 , with a maximum of 10.6 myr -1 in the period 1991-20 0 3. Hence, rainfall would have had a significant impact on the cooling history of the dome. Simmons et al. (2004) estimated the quenching potential of rain ( ΔTR - change in rock temperature) by assu ming rain falls in a single event over an entire surface resulting in the equation hc icT T RR WWW R ρ ρΔ=Δ [Eq. 5-8] where ΔTw represents the temperature change of water between ambient and boiling temperature, ρ and c are the densities and specific heats for water and rock, i is the height of rain, and h is the th ickness of the dome. From this , the rainfall-quenching rate for Taranaki is estimated to be 11.5 °Cyr -1 . If both rates are combined a rate of 28.6 °Cyr -1 would imply that the summit dome would cool to 350 °C in approximately 19 years. It is noted that the simple conduc tive cooling model does not include thermal radiation, which would further increase the annual cooling rate. In addition, the applied Chapter 5 Non-explosive, Dome-forming Er uptions 160 dimension of a spherical dome results in maximum cooling times since the dome on the upper slope is less than 100 m thick. From the estimated maximum cooling time of c.19 years, it is here inferred that the dome was emplaced some time between AD 1755 and AD 1866. If complete cooling of the dome is considered and the same conditions are applied then about 74 years needed to elap se placing the date of extrusion before c.AD 1811. Table 5-5. Physical parameters used for conductive cooling and rainfall-quenching of the lava dome. Field studies showed that the rock-avala nche deposit overlies the Tahurangi BAF A deposit at two locations (outcrops 15 and 52; Fig. 5-3) and shows a weakly developed 4-5 cm thick soil. Revegetation and pedogene sis on bare rock-avalanche deposits take considerable time, especially at high altitudes. The lack of clay minerals along with the high porosity and permeability of those deposits further delay the recolonisation of those deposits. If there were better data on pedogenesis in montane regions after devastating debris flow or rock avalanche events, the eruption time of the Pyramid event may even be better constrained. The photograph of H.M. Skeet was taken more than 13 years after the rock-avalanche depositi on. The fresh bouldery surface is still recognisable at the time of the photograph with no indication of a vegetation cover (Fig. 5-5). This may indicate that revegetation a nd pedogenesis occurred within decades after deposition. As an hypothetical example, if revegetation an d pedogenesis on widespread rock-avalanche deposits under local conditio ns take at least 30 years then and considering the dome cooling time of 19 year s it can be estimated that the Pyramid eruption began c.11 years following the Tahuran gi eruption, i.e. the time window for the Pyramid eruption could then be narrowed down to occurring between c.AD 1766 and 1866. Chapter 5 Non-explosive, Dome-forming Er uptions 161 5.5.6 Historic References to Volcanic Activity in the 19 th Century Although Maori were living in Taranaki since the c.14 th century and potentially witnessed most eruptions of the Maero period, their oral traditions contain few references to some but not all of the eruptions of Mt. Taranaki (Faulkner-Blake, 1977). This is probably due to the predominantly extrusive, lava dome-producing, style of activity in that period with only one sub-Plinian event in AD 1655. The first ascent by Europeans to the summit of Mt. Tarana ki was by the German geologist Ernst Dieffenbach and the whaler James Herberley on the 23 December 1839 (Dieffenbach, 1843). Although no volcanic activity was obs erved by them on the summit of Mt. Taranaki, their brief description of th e summit region gives important clues. Dieffenbach (1843) entered the crater from th e north and estimated the crater area at approximately one square mile (c.1.61 km 2 ). This is clearly overestimated since given the ideal crater dimensions an area of only 0.52 km 2 results. However, the circumstances for Dieffenbach’s estimate can be more closely examined. Anyone who enters the summit of Mt. Taranaki today fr om the north (as Dieffenbach), would never estimate the crater area to be a square mile, ev en if viewed from the highest point of the present summit dome. An explanation for his overestimation could be that the crater area was just composed of a field of snow with some protruding rocks as he described (Table 5-6; Dieffenbach, 1843). From this pers pective (i.e. in the absence of the summit dome), the crater area can easily be overestimat ed if it is considered that the snow field gently dips to the west with the western crater rim missing. Further, if there was a summit dome at the time of Dieffenbachs as cent then this important morphological feature would have been described by him, fo r example, as a conical hill composed of ash, scoria and lava. Even if intensive snow cover made it more difficult to recognise the dome as positive relief, no “protruding rock s” should be recognisable (see above). In addition, reports stating that Mt. Tara naki was an active vol cano have not been previously known. The first citation is given by Hay (1832) , others are from 1848 and 1862 by Daubeney and Scrope, re spectively (Table 5-6). A lthough these references are not supported by other sources, th ey clearly raise the question as to wh ether an eruption of Mt. Taranaki during European settlement occurred unnoticed. Before European settlement started in th e early 1840s on the Tara naki peninsula (New Plymouth was founded in 1841), only whalers were present in this area. Since a very short time of dome emplacement is indicated, it may be possible that this small eruption was unnoticed in the early years of colonisation. Paintings, diaries and eyewitness Chapter 5 Non-explosive, Dome-forming Er uptions 162 reports (1841-1890) record useful observati ons (Table 5-6) which could refer to volcanic activity at Mt. Taranaki, but inform ation provided is not precise. Earthquakes felt in Taranaki are often accompanied by so und and explosions from Mts. Tongariro and Ngauruhoe and are frequently heard. Th ese common occurrences may have led to either any unusual sounds or shocks being igno red or that they were instantly related to other phenomena “outside” Taranaki. Only one observation from 1854 appears to describe volcanic activity at Mt. Taranaki (Table 5- 6). Although Dieffenbach’s observations of the crater area and various other eyewitness reports offer only weak evidence, a post- AD 1839 date for the Pyramid eruption cannot be excluded. This assumption is supported by the only known scien tific evidence. Lees and Neall (1993) reported Pinus pollen at two locations below Tahurangi Ash, indicative for an eruption that occurred during early colonisation of the Taranaki peninsula. An eruption date for the Tahurangi eruption close to AD 1860 has been suggested (Lees and Neall, 1993). If this finding is considered the Pyramid eruption occurred later. All that can be definitely stated about the Pyramid eruption are the following facts: 1) the collapse of the Pyramid Dome occurred prior to AD 1885; 2) the temperature of the rock-avalanche deposit was below 350 °C; 3) the development of faults within the dome implies they were created while the dome was entirely in a brittle state, when the dome interior was probably also below the temperature of 350 °C; 4) from the inferred dome dimensions, the dome needed at least 19 years to cool to about 350 °C, or about 74 y ears to cool completely; 5) the date of the dome-emplacing eruption was after the Tahurangi eruption and prior to c.AD 1866. A precise date of the Tahurangi eruption is needed, i.e. AD 1755 (Druce, 1966) vs. AD 1860 (Lees and Neall, 1993). Chapter 5 Non-explosive, Dome-forming Er uptions 163 Table 5-6. Selected eyewitness reports of the 18 th and 19 th century. Chapter 5 Non-explosive, Dome-forming Er uptions 164 Table 5-6. (continued). Chapter 5 Non-explosive, Dome-forming Er uptions 165 Table 5-6. (continued). Chapter 5 Non-explosive, Dome-forming Er uptions 166 5.6 Conclusions The dome remnants within the crater of Mt. Taranaki were extruded after the AD 1755 Tahurangi eruption, and hence, represent a di stinct eruptive event, the youngest known from the volcano, here termed the Pyramid erupt ion. This event resulted in the extrusion of a lava dome at the volcano’s summit, w ithout any apparent collapse to form Block- and-Ash Flows. Importantly from a hazard pe rspective, once the dome had cooled – at least two decades (c.19 yrs) after eruption, it collapsed to generate a large rock avalanche, extending over 5 km from source. Studies of the summit dome structure, mineralogy, chemical compositions, and geometrical shape show that its emplacement occurred rapidly, probably within days. Exo- and endogenous growth occurred due to part of the dome extending over a break in slope and onto the steep outer flank. The portion emplaced within the crater developed a lobate-platy dome shape. Tw o methods using independent parameters showed that minimum magma extrusion ra tes were in the order of 6 to 17 m 3 s-1 . Magma ascent rates of ≥ 0.012 ms -1 are estimated on the basis of hornblende breakdown reaction rims. A mid-crustal magma source area is iden tified near the BDTZ at c.9.5 km depth. It may also be possible that temporary magma stalling also occurred below 2 km depth, within the inferred hornblende stability limit. The cooling of the dome to ambient temperatures was complete in less than 74 years; while cooling to the inferred maximum dome temperature of 350 °C at the time of the collapse could have been within c.19 years. Chapter 5 Non-explosive, Dome-forming Er uptions 167 5.7 References Anderson, J.L. and Smith, D.R. 1995. The effects of temperature and fO 2 on the Al-in-hornblende barometer. American Mineralogist 80, 549- 55 9. Anderson, S.W. and Fink, J.H. 1990. The development and distributio n of surface textures at the Mount St. Helens Dome. In: Fink, J.H. (ed.) Lava flows and domes: emplacement mechanisms and hazard implications. IAVCEI Proceeding in Volcanology 2, 25-46. Anderson, S.W., Fink, J.H. and Rose, W.I. 1995. Mount St. Helens and Santiaguito lava domes: the effect of short-term eruption rate on surface texture and degassing processes. Journal of Volcanology and Geothermal Research 69, 105 -11 6. Anderson S.W., Stofan, E.R, Plaut, J.J. and Crown, D.A. 1998. Block size distributions on silicic lava flow surfaces; implications for emplacement conditions. Geological Society of America Bulletin 110, 1258 -12 67. Aylmers, J. E. 1862. Distant homes; or the Graham Family in New Zealand. Griffith and Farran. London. Barclay, J. and Carmichael, I.S.E. 2004. A hornblende basalt from western Mexico: Water-saturated phase relations constrain a pressure-temperature window of eruptibility. Journal of Petrology 45, 485- 506. Barclay, J., Rutherford, M.J ., Carroll, M.R., Murphy, M.D., Devine, J.D., Gardner, J. and Sparks, R.S.J. 1998. Experimental phase equilibria constraints on pre-eruptive storage conditions of the Soufriere Hills magma. Geophysical Research Letters 25, 3437 -34 40. B GVN 17:11. Smithsonian Institution, 1992. Unzen Volcano. Bulletin of the Global Volcanism Network vol. 17, no. 11. Blain, M. 2000. The Canterbury Association: a study of its members connections: a tribute on the 150 th anniversary of the Canterbury settlement. Occasiona l Paper 4, The Archives Committee, Anglican Diocese of Canterbury. Blake, S. 1990. Viscoplastic models of lava domes. In: Fink, J.H. (ed.) Lava flows and domes: emplacement mechanisms and hazard implications. I A VCEI Proceeding in Volcanology 2, 88-126. Blundy, J.D. and Holland, T.J.B. 1990. Calcic amphibole equilibria and a new amphibole-plagioclase geothermometer. Contributions to Mineralogy and Petrology 104, 208 -224. Boyd, F.R. 19 59. Hydrothermal investigations of amphiboles. In: Abelson, P.H. (ed.) Researches in Geochemistry. Wiley. New York. pp.377-396. Browne, B.L. and Gardner, J.E. 2006. The influence of magma ascent path on the texture, mineralogy, and formation of hornblende reaction rims. Earth and Planetary Science Letters 246, 161 -17 6. Chapter 5 Non-explosive, Dome-forming Er uptions 168 Browne, B.L., Eichelberger, J.C., Patino, L.C., Vogel, T.A., Dehn, J., Uto, K. and Hoshizumi, H. 2006. Generation of porphyritic and equigranular mafic enclaves during magma recharge events at Unzen volcano, Japan. Journal of Petrology 47, 301- 328. Cashman, K.V. and McConnell, S.M. 2005. Multiple levels of magma storage during the 1980 summer eruptions of Mount St. Helens, WA. Bulletin of Volcanology 68, 57-75. Cavill, A.W., Cassidy, J. and Brennan, B.J. 1997. Results from the new seismic monitoring network at Egmont Volcano, New Zealand: tectonic and hazard implications. New Zealand Journal of Geology and Geophysics 40, 69-76. Cole, J.W. and Nairn, I.A. 1975. Catalogue of the active volcano es of the world including solfatara fields. Part 22: New Zealand. International Associ ation of Volcanology and Chemistry of the Earth's Interior. Rome. Colenso, W. 1846- 184 8. Journal, Paper Vol. 2. Cotton, W.C. 1845 -18 46. Diary of W.C. Cotton. Crawford, J.C. 1880. Recollections of travel in New Z ealand and Australia. Trubner & Co. London. Cronin, S.J., Stewart, R.B., Neall, V.E., Platz, T. and Gaylord, D. 2003. The AD1040 to present Maero eruptive period of Egmont Volcan o, Taranaki, New Zealand. Abstract. Geological Society of New Zealand Miscellaneous Publication 116A, 43. Daubeny, C. 1848. A description of active and extinct volcanos, of earthquakes, and of thermal springs; with remarks on the causes of th ese phænomena, the character of th eir respective products, and their influence on the past and present condition of the globe. 2nd ed. Richard and John E. Taylor. London. Dieffenbach, E. 1843. Travels in New Zealand; with contributions to the Geography, Geology, Botany, and natural history of that country. Vol. I. John Murray. London. Druce, A.P. 1964. The vegetation. In: Scanlan, A.B. (ed.) Egmont National Park Handbook. Egmont National Park Board. New Plymouth. pp. 41-56. Druce, A.P. 1966. Tree-ring dating of recent volcanic ash and lapilli, Mt Egmont. New Zealand Journal of Botany 4, 3-41. Eiby, G.A. 1966. The modified Mercalli scale of eart hquake intensity and its use in New Zealand. New Zealand Journal of Geology and Geophysics 9, 122 -12 9. Eiby, G.A. 1968. A descriptive catalogue of New Zealand earthquakes. Part 1 Shocks felt before the end of 1845. New Zealand Journal of Geology and Geophysics 11, 16-40. Eiby, G.A. 1973. A descriptive catalogue of New Zealand ear thquakes. Part 2 Shocks felt from 1846 to 1854. New Zealand Journal of Geology and Geophysics 16, 857-9 07. Elsworth, D. 19 89. Thermal permeability enhancement of blocky rocks: plane and radial flow. International Journal of Rock Mechanics and Mining Sciences 26, 329 -3 39. Chapter 5 Non-explosive, Dome-forming Er uptions 169 Faulkner - Blake, J. 1977. Nga maunga Tarewarewa or the restless mountains. 7 audiotapes and brochure. Radio New Zealand, New Plymouth. Fink, J.H. and Griffiths, R.W. 1998. Morphology, eruption rates, and rheology of lava domes: insights from laboratory models. Journal of Geophysical Research - Solid Earth 103, 527-5 45. Fink, J.H. and Anderson, S.W. 2000. Lava domes and coulees. In: Sigurdsson, H. (ed.) Encyclopedia of Volcanoes. Academic press. London. pp. 307-31 9. Foden, J.D. and Green, D.H. 1992. Possible role of amphibole in the origin of andesite: some experimental and natural evidence. Contributions to Mineralogy and Petrology 109, 479- 493. Gill, J.B. 1981. Orogenic andesites and plate tectonics. Springer Verlag. Berlin. Grant - Taylor, T.L. 1964. Volcanic history of western Taranaki. New Zealand Journal of Geology and Geophysics 7, 78-8 6. Gregg, D.R. 1960. Volcanoes of Tongariro National Park. New Zealand Department of Scientific and Industrial Research Information series 28. Griffiths, R.W. and Fink, J.H. 1997. Solidifying Bingham extrusions: a model for the growth of silicic lava domes. Journal of Fluid Mechanics 347, 13-36. Gründer, K. 2006. Petrography and mineral chemistry of a xenolith suite from Mt. Taranaki, New Zealand: insights into the sub-volcanic lithospher e of an arc volcano. Unpublished Diploma thesis. Johannes Gutenberg Universität, Ma inz, Bundesrepublik Deutschland. Hammarstrom, J.M. and Zen, E.A. 1986. Aluminum in hornblende: an empirical igneous geobarometer. American Mineralogist 71, 1297- 131 3. Hawkesworth, J. 1775. An account of the voyag es undertaken by the order of His present Majesty for making discoveries in the Southern Hemisphere, and successively performed by Commodore Byron, Captain Wallis, Captain Carteret and Captain Cook in the Dolphin, the Swallow, and the Endeavour; to which is added: A voyage to the North Po le by Commodore Phipps. Vol. II. James Williams. Dublin. Hay, R.W. 1832. Notices of New Zealand. From origin al documents in the Colonial Office. In: The Royal Journal of Geographical Society of London . Vol. II. 133-1 35. Hoblitt, R.P. and Kellog, K.S. 1979. Emplacement temperature of unsorted and unstratified deposits of volcanic rock debris as determined by paleomagnetic techniques. Geological Society of America Bulletin 90, 633 -64 2. Hochstetter, F von 1864. Geologie von Neu-Seeland. Beiträge zur Geologie der Provinzen Auckland und Nelson. Reise der österreichischen Fregatte Novara um die Erde in den Jahren 1857, 1858 , 1859. Geologische Theil, Band 1, Abteilung 1 und 2. Hof- und Staatsdruckerei. Wien. Holland, T. and Blundy, J. 1994. Non-ideal interactions in cal cic amphiboles and their bearing on amphibole-plagioclase thermometry. Contributions to Mineralogy and Petrology 116, 433 -44 7. Chapter 5 Non-explosive, Dome-forming Er uptions 170 Hollister, L.S., Grissom, G.C., Peters, E.K., Stowell, H.H. and Sisson, V.B. 1987. Confirmation of the empirical correlation of Al in hornblende with pressure of solidification of calc-alkaline plutons. American Mineralogist 72, 231-2 39. Holloway, J.R. 1973. The system pargasite-H 2 O-CO 2 : a model for melting of a hydrous mineral with a mixed-volatile fluid, I. experimental results to 8 kbar. Geochimica et Cosmochimica Acta 37, 651- 6 66. Holm, D.K. 1995. Relation of deformation and multiple intrusion in the Death Valley extended region, California, with implications for magma entrapment mechanism. Journal of Geophysical Research- Solid Earth 100, 10495 -10 505. Holtz, F., Sato, H., Lewis, J., Behrens, H. and Nakada, S. 2005. Experimental petrology of the 1991- 1 995 Unzen dacite, Japan. Part I: Phase rela tions, phase composition and pre-eruptive conditions. Journal of Petrology 46, 319-3 37. Jaeger, J.C. 1968. Cooling and solidification of igneous rocks. In: Hess, H.H. and Poldervaart, A. (eds.) Basalts: The Poldervaart tr eatise on rocks of basaltic composition. Wiley. New York. pp. 503- 5 36. Johnson, M.C. and Rutherford, M.J. 1989. Experimental calibration of the aluminum-in-hornblende geobarometer with application to Long Va lley Caldera (California) volcanic rocks. Geology 17, 837- 8 41. Le Maitre, R.W., Bateman, P., Dudek, A., Keller, J., Lameyre, J., Le Bas, M.J., Sabine, P.A., Schmid, R., Sorensen, H., Streck eisen, A., Woolley, A.R. and Zanettin, B. 1989. A classification of igneous rocks and glossary of terms. Blackwell Scientific. Oxford. Leake, B.E., Woolley, A.R., Arps, C.E.S., Birch, W.D., Gilbert, M.C., Grice, J.D., Hawthorne, F.C., Kato, A., Kisch, H.J., Krivovichev, V.G., Lintho ut, K., Laird, J., Mandarino, J., Maresch, W.V., Nickel, E.H., Rock, N.M.S., Schumacher, J.C., Smith, D.C., Stephenson, N.C.N., Ungaretti, L., Whittaker, E.J.W. and Youzhi, G. 1997a. Nomenclature of amphiboles: report of the Subcommittee on amphiboles of the International Mineralogical Association Commission on new minerals and mineral names. Mineralogical Magazine 61, 295- 321. Leake, B.E., Woolley, A.R., Arps, C.E.S., Birch, W.D., Gilbert, M.C., Grice, J.D., Hawthorne, F.C., Kato, A., Kisch, H.J., Krivovich ev, V.G., Linthout, K., Laird, J., Mandarino, J.A., Maresch, W.V., Nickel, E.H., Rock, N.M.S., Schumacher, J.C., Smith, D.C., Stephenson, N.C.N., Ungaretti, L., Whittaker, E.J.W. and Guo, Y.Z. 1997b. Nomenclature of amphiboles: report of the Subcommittee on amphiboles of the International Mineralogical Association Commission on new minerals and mineral names. American Mineralogist 82, 1019- 103 7. Leake, B.E. Woolley, A.R., Birch, W.D., Burke, E.A.J., Ferraris, G. , Grice, J.D., Hawthorne, F.C., Kisch, H.J., Krivovichev, V.G., Schumacher, J.C., Stephenson, N.C.N. and Whittaker, E.J.W. 2003. Nomenclature of amphiboles: additions and revisions to the International Mineralogical Association’s 1997 recommendations. Canadian Mineralogist 41, 135 5- 13 62. Chapter 5 Non-explosive, Dome-forming Er uptions 171 Lees, C.M. and Neall, V.E. 1993. Vegetation response to volcanic eruptions on Egmont Volcano, New Zealand, during the last 1500 years. Journal of the Royal Society of New Zealand 23, 91-1 27. Locke, C.A. and Cassidy, J. 1997. Egmont Volcano, New Zealand: three-dimensional structure and its implications for evolution. Journal of Volcanology and Geothermal Research 76, 149-16 1. Luhr, J.F. 2002. Petrology and geochemistry of the 1991 and 1998- 199 9 lava flows from Volcán de Colima, México: implications for the end of the current eruptive cycle. Journal of Volcanology and Geothermal Research 117, 169 -19 4. Luhr, J.F. and Carmichael, I.S.E. 1980. The Colima Volcanic Complex, Mexico. I. Post-caldera andesites from Volcán Colima. Contributions to Mineralogy and Petrology 71, 343-3 72. Lykins, R.W. and Jenkins, D.M. 1992. Experimental determination of pargasite stability relations in the presence of orthopyroxene. Contributions to Mineralogy and Petrology 112, 405- 413. Lyman, A.W., Koenig, E. and Fink, J.H. 2004. Predicting yield strengths and effusion rates of lava domes from morphology and underlying topography. Journal of Volcanology and Geothermal Research 129, 125- 138. Marquez, A., Oyarzun, R., Ignacio, C. de and Doblas, M. 2001. Southward migration of volcanic activity in the central Mexican Volcanic Belt: asy mmetric extensions within a two-layer crustal stretching model. Journal of Volcanology and Geothermal Research 112, 175-1 87. Melnik, O. and Sparks, R.S.J. 2002. Dynamics of magma ascent and lava extrusion at Soufrière Hills Volcano, Montserrat. In: Druitt, T.H. and Kokelaar, B.P. (eds.) The eruption of Soufrière Hills Volcano, Montserrat, from 1995 to 1999. Geological Society, London, Memoirs 21, 153 -1 71. Messenger, W. 1853-1 858. Diary of W. Messenger. Moffat, K. 2000. The Puritan paradox: an annotated bibliography of Puritan and Anti-Puritan New Zealand fiction, 1860 -19 40. Part 1: The Puritan legacy. K ōtare: New Zealand Notes and Queries 3, 36-8 6. Moore, G. and Carmichael, I.S.E. 1998. The hydrous phase equilibria (to 3 kbar) of an andesite and basaltic andesite from western Mexico: constraints on water content and conditions of phenocryst growth. Contributions to Mineralogy and Petrology 130, 304 -31 9. Morimoto, N. 1989. Nomenclature of pyroxenes. Canadian Mineralogist 27, 143-1 56. Moriya, I. 1978. Morphology of lava domes. Bulletin of the Department of Geography - The University of Kanazawa 14, 55-69. Mueller, S., Melnik, O., Spieler, O., Scheu, B. and Dingwell, D.B. 2005. Permeability and degassing of dome lavas undergoing rapid decompression: an experimental determination. Bulletin of Volcanology 67, 526- 538. Chapter 5 Non-explosive, Dome-forming Er uptions 172 Murphy, M.D., Sparks, R.S.J., Barclay, J., Carroll, M.R., Lejeune, A.M., Brewer, T.S., Macdonald, R., Black, S. and Young, S. 1998. The role of magma mixing in triggering the current eruption at the Soufriere Hills volcano, Montserrat, West Indies. Geophysical Research Letters 25, 3433- 343 6. Nakada, S. and Motomura, Y. 1999. Petrology of the 1991-1 995 eruption at Unzen: effusion pulsation and groundmass crystallization. Journal of Volcanology and Geothermal Research 89, 173-1 96. Neall, V.E. 1971. Volcanic domes and lineations in Egmont National Park. New Zealand Journal of Geology and Geophysics 14, 71-81. Neall, V.E. 1972. Tephrochronology and tephrostratigraphy of western Taranaki (N108 -1 09) , New Zealand. New Zealand Journal of Geology and Geophysics 15, 507-5 57. Neall, V.E. 1973. Some aspects of western Taranaki geology and pedology. Unpublished PhD thesis. Victoria University of Wellington, Wellington, New Zealand. Neall, V.E., Stewart, R.B. and Smith, I.E.M. 1986. History and petrology of the Taranaki volcanoes. In: Smith, I.E.M. (ed.) Late Cenozoic volcanism. Royal Society of New Zealand Bulletin 23, 251-2 63. Newhall, C.G. and Melson, W.G. 1983. Explosive activity associat ed with the growth of volcanic domes. Journal of Volcanology and Geothermal Research 17, 111- 131. Newland, J. 1841- 187 3. Journal of John Newland, Vol. 2, New Plymouth. Pallister, J.S., Hoblitt, R.P. and Reyes, A.G. 1992. A basalt trigger for the 1991 eruptions of Pinatubo volcano? Nature 356, 426- 428. Platz, T. 2001. Mapping and characterisation of the vol caniclastic Maero formation deposits on the northwestern sector of Egmont volcano (Mt. Taranaki), New Zealand. Unpublished Diploma mapping thesis. Ernst-Moritz-Arndt Universitä t, Greifswald, Bundesrepublik Deutschland. Ponomareva, V.V., Melekestsev, I.V. and Dirksen, O.V. 2006. Sector collapses and large landslides on Late Pleistocene-Holocene vol canoes in Kamchatka, Russia. Journal of Volcanology and Geothermal Research 158, 117- 138. Pownall, R.W. 1885. Illustrated guide to the west coast of the North Island, N.Z., describing the various towns and adjacent land available for settlement. A.D. Willis. Wanganui. Price, R.C., McCulloch, M.T., Smith, I.E.M. and Stewart, R.B. 1992. Pb-Nd-Sr isotopic compositions and trace element characteristics of young volcanic rocks from Egmont Volcano and comparisons with basalts and andesites from the Taupo Volcanic Zone, New Zealand. Geochimica et Cosmochimica Acta 56, 941-9 53. Price, R.C., Stewart, R.B., Woodhead, J.D. and Smith, I.E.M. 1999. Petrogenesis of high-K arc magmas: evidence from Egmont Volc ano, North Island, New Zealand. Journal of Petrology 40, 167- 1 97. Prouteau, G. and Scaillet, B. 2003. Experimental constraints on the origin of the 1991 Pinatubo dacite. Journal of Petrology 44, 2203- 224 1. Chapter 5 Non-explosive, Dome-forming Er uptions 173 Rieder, M., Cavazzini, G., D’Yakono v, Y., Frank-Kamenetskii, V.A., Gottardi, G., Guggenheim, S., Koval’, P.V., Müller, G., Neiva, A.M.R., Radosl ovich, E.W., Robert, J.-L., Sassi, F.P., Takeda, H., Weiss, Z. and Wones, D.R. 1998. Nomenclature of the micas. Canadian Mineralogist 36, 905- 9 12. Roman, D.C., Cashman, K.V., Gardner, C.A., Wallace, P.J. and Donovan, J.J. 2006. Storage and interaction of compositionally heterogeneous magmas from the 1986 eruption of Augustine Volcano, Alaska. Bulletin of Volcanology 68, 240- 254. Rutherford, M.J. and Devine, J.D. 1988. The May 18, 1980 , eruption of Mount St. Helens 3. Stability and chemistry of amphibole in the magma chamber. Journal of Geophysical Research-Solid Earth 93, 1194 9-1 195 9. Rutherford, M.J. and Devine, J.D. 2003. Magmatic conditions and magma ascent as indicated by hornblende phase equilibria and reactions in the 1995-20 02 Soufrière Hills magma. Journal of Petrology 44, 1433 -14 53. Rutherford, M.J. and Hill, P.M. 1993. Magma ascent rates from amphibole breakdown - an experimental study applied to the 1980- 198 6 Mount St Helens eruptions. Journal of Geophysical Research-Solid Earth 98, 1966 7-1 968 5. Sato, H., Nakada, S., Fujii, T., Nakamura, M. and Suzuki-Kamata, K. 1999. Groundmass pargasite in the 1991-1 995 dacite of Unzen volcano: phase stability experiments and volcanological implications. Journal of Volcanology and Geothermal Research 89, 197- 212. Schmidt, M.W. 1992. Amphibole composition in tonalite as a function of pressure: an experimental calibration of the Al-in-hornblende barometer. Contributions to Mineralogy and Petrology 110, 304 - 3 10. Schumacher, J.C. 1997. The estimation of ferric iron in electron-microprobe analysis of amphiboles. Appendix 2. In: Leake, B.E. Woolley, A.R., Arps, C.E.S. , Birch, W.D., Gilbert, M.C., Grice, J.D., Hawthorne, F.C., Kato, A., Kisch, H.J., Krivovichev, V.G., Linthout, K., Laird, J., Mandarino, J.A., Maresch, W.V., Nickel, E.H., Rock, N.M.S., Schumacher, J.C., Smith, D.C., Stephenson, N.C.N., Ungaretti, L., Whittaker, E.J.W. and Youzhi, G. Nomenclature of amphiboles: report of the Subcommittee on amphiboles of the International Mineralogical Association, Commission on new minerals and mineral names. Canadian Mineralogist 35, 238- 246. Scrope, G.P. 1862. Volcanos. The character of their phenomena, their share in the structure and composition of the surface of the globe, and their relation to its inte rnal forces. With a descriptive catalogue of all known volcanos and volcanic formations. 2nd ed. Longman, Green, Longmans, and Roberts. London. Sherburn, S. and White, R.S. 2005. Crustal seismicity in Taranaki, New Zealand using accurate hypocentres from a dense network. Geophysical Journal International 162, 494-5 06. Sherburn, S., White, R.S. and Chadwick, M. 2006. Three-dimensional tomographic imaging of the Taranaki volcanoes, New Zealand. Geophysical Journal International 166, 957 -96 9. Chapter 5 Non-explosive, Dome-forming Er uptions 174 Simmons, J., Elsworth, D. and Voight, B. 2004. Instability of exogenous lava lobes during intense rainfall. Bulletin of Volcanology 66, 725- 734. Smith, J.V., Miyake, Y. and Oikawa, T. 2001. Interpretation of porosity in dacite lava domes as ductile- brittle failure textures. Journal of Volcanology and Geothermal Research 112, 25-35. Stewart, R.B., Price, R.C. and Smith, I.E.M. 1996. Evolution of high-K arc magma, Egmont volcano, Taranaki, New Zealand: evidence from mineral chemistry. Journal of Volcanology and Geothermal Research 74, 275-2 95. Sun, S. - S. and McDonough, W.F. 1989. Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes. Geological Society, London, Special Publications 42, 313- 345. Swanson, D.A. and Holcomb, R.T. 1990. Regularities in growth of the Mount St. Helens dacite dome, 1980 -19 86. In: Fink, J.H. (ed.) Lava flows and domes: emplacement mechanisms and hazard implications. IAVCEI Proceeding in Volcanology 2, 3-24. Taylor, R. 1833 -18 73. Journal of Rev. Richard Taylor. 14 vol. Typescript in Alexander Turnbull Library, Wellington. Taylor, G.A.M. 1958. The 1951 eruption of Mount Lamington, Papua. Australian Bureau of Mineral Resources , Geology and Geophysics Bulletin 38. The Taranaki Herald 1853. 5 January. Thomas, W.M. and Ernst, W.G. 1990. The aluminum content of hor nblende in calc-alkaline granitic rocks: a mineralogic barometer calibrated experimentally to 12 kbars. In: Spencer, R.J. and Chou, I.-M. (eds.) Fluid-mineral interactions: a tribute to H. P. Eugster. Geochemical Society Special Publication 2, 59-6 3. Venezky, D.Y. and Rutherford, M.J. 1999. Petrology and Fe-Ti oxide reequilibration of the 1991 Mount Unzen mixed magma. Journal of Volcanology and Geothermal Research 89, 213- 230. Walter, T.R., Wang, R., Zimmer, M., Grosser, H., Lühr, B. and Ratdomopurbo, A. 2007. Volcanic activity influenced by tectonic earthquakes: static and dynamic stress triggering at Mt. Merapi. Geophysical Research Letters 34, L053 04. Westrich, H.R. and Holloway, J.R. 1981. Experimental dehydration of pargasite and calculation of its entropy and Gibbs energy. American Journal of Science 281, 922- 934. Zernack, A., Stewart, R.B. and Price, R.C. 2006. Slab or crust - K 2 O enrichment at Egmont volcano, New Zealand. Geochimica et Cosmochimica Acta Supplement 70, 614. Chapter 6 Transition to Explosive Eruptions 175 Chapter 6 Transition to Explosive Eruptions 6 Chapter 6 Long-term dome growth by extrusion of viscous andesitic magma can be interrupted or terminated by explosive activity. To unde rstand the driving forces behind sudden changes from extrusive to explosive behavi our at andesitic volcanoes, a study of the AD 1655 Burrell Lapilli eruption episode at Mt Taranaki was carried out. Individual phases of the eruption have been reconstructed and a model is proposed that explains the diversity of erupted ejecta. Chapter 6 Transition to Explosive Eruptions 176 Chapter 6 represents the full version of a pub lished journal paper. Its format has been altered to match the overall thesis. Additional data collected in the context of the Burrell episode are presented in the appendices of this chapter. Title Transition from effusive to explosiv e phases in andesite eruptions – a case- study from the AD 1655 eruption of Mt. Ta ranaki, New Zealand. Authors Thomas Platz, Shane J. Cronin, Kath arine V. Cashman, Robert B. Stewart and Ian E.M. Smith. Status published in Journal of Volcanology and Geothermal Research 2 0 0 7 , vol. 161, 15-34. Principal investigator Thomas Platz Carried out - field description and sampling, - laboratory preparation of samples, - optical microscopy, - electron microprobe analysis. - scanning electron microscopy, - porosity and permeability measurements, - FTIR spectroscopy - manuscript preparation, writing and submission Co-investigators Shane J. Cronin, Katharine V. Cashman, Robert B. Stewart, and Ian E.M. Smith Aided the study by - fiel d assistance (Shane J. Cronin) - discussing results - ed iting and discussion of manuscript Chapter 6 Transition to Explosive Eruptions 177 6.1 Abstract The extrusion of viscous andesite lava forming domes can terminate in explosive activity. To understand the driving forces behind this behaviour, a study of an AD 1655 eruption episode at Mt Taranaki was carried out. We propose that simple changes in magmatic conditions of a single hydrous melt during ascent caused sudden changes in explosivity and gave rise to the highl y variable pumice with highly variable vesicularities and colour. Fractionation of hornblende + plagioclase + clinopyroxene + Fe-Ti oxide at the onse t of magma ascent, and step-wise crystallisation of plagioclase ± clinopyroxene in different pa rts of a single melt within the conduit was controlled by rates of initial rise, capping by an impermeab le lava dome, and differential rates of vesiculation and volatile exsolution. This resulted in a vertical stratification in the conduit, comprising a viscous, hyp o-crystalline lava cap, that overlay alternating zones of grey, brown and grey magma-foams. Hori zontal gradients in geochemistry in the conduit are also indicated by different clast textures. The eruption consisted of an initial extrusive phase followed by three pulses of sub-Plinian activity. Each phase or pulse, corresponded to individual layers within the conduit. Ejecta included Block-and-Ash Flow deposits, three pyroclas tic pumice-flow deposits of alternating grey, brown and grey pumice, as well as fallout deposits dominated by grey pumice. The brown magma foam contained more microlites, had a more-evolved matrix glass, and a higher temperature than the grey magma foams above and below. Its erup tion destabilised the sub-Plinian eruption column because it was more degassed. It fragmented less efficiently than the grey magma foams due to its lower viscosity, preventing pressure build-up in bubbles. Incomplete mixing at interfaces between brown and grey magma phases gave rise to banded pumices. Chapter 6 Transition to Explosive Eruptions 178 6.2 Introduction In many historic eruptions of andesitic and dacitic volcanoes an initially effusive lava emission phase was abruptly terminated by an explosive phase, often with highly hazardous consequences. Such shifts in er uptive style have been related to many complex sub-surface processes such as , decompression-induced crystallisation (Hammer et al., 2000), increase in magm a viscosity by increased crystallinity (predominantly microlites) and vesicularity, water-loss, temperature gradients (Stevenson et al., 1996; Manga et al., 1998), as well as conduit shear processes within vesicular magmas (Stasiuk et al., 1996; Masso l et al., 2001; Rust et al., 2003; Tuffen et al., 2003), and viscous dissipation at conduit walls (Polacci et al., 2001, 2005; Rosi et al., 2004). These sudden tran sitions in andesite eruptions are accompanied by the presence of ejecta with a wide range in vesicularity and permeability (Klug and Cashman, 1994; Gardner et al ., 1998; Rosi et al., 2004). Th is includes strongly banded pumice clasts, with grey to brown bands, interpreted as mingled compositions, or products of shear processes within conduits. Emplacement and destruction of lava domes is very common at subduction-related andesitic and dacitic volcanoes. Partial dome collapse is commonly caused by [1] over- steepening (Sparks et al., 2000), [2] gas ove r-pressurisation (Elsworth and Voight, 2001), and [3] rainfall (Matthews et al., 2002 ). However, dome collapse and partial unroofing of the vent does not always trigger a shift in eruptive style. This was true for most of the eruptions over the last c.1000 years at Mt. Ta ranaki, where several lava domes were emplaced, forming Block-and-As h Flows (BAFs) thr ough dome collapse. However, one of these dome-effusion peri ods was terminated by an explosive sub- Plinian eruption (Druce, 1966). These sub-P linian deposits, and all other known from sub-Plinian events at this volcano, cont ain notably high contents (10-30%) of monolithologic dense lithics, wh ich appear to be juvenile. Understanding the causes of such transitions in eruptive style is critical for constructing hazard models for andesitic volcanoes and hence developing realistic eruption scenarios and volcanic emergency plans. Here we propos e an alternative model to explain some types of transitions from effusive to explosive activity, primarily through changes of magmatic conditions during ascent. This model also helps explain the production of a variety of pumice textures as well as the or igin of some types of banded pumice. We base this study on the AD 1655 Burrell Lapilli eruption of Mt. Taranaki, an episode in which a small-scale dome effusion event was suddenly terminated by an explosive sub- Chapter 6 Transition to Explosive Eruptions 179 Plinian eruption that covered several 100’s of km 2 of New Zealand’s central North Island in fall deposits. 6.2.1 Geological Setting Mount Taranaki/Egmont volcano is located in the west of the North Island of New Zealand and is the youngest edifice in a 0.75-0 .13 Ma SE-trending volcanic alignment [Fig. 6-1; (Neall et al., 1986; Stewart et al., 1996)]. The volca nic history of Mt. Taranaki began c.130,000 years ago and has been characterised by several cone- building and collapse episodes (Alloway et al., 1995). The cu rrent edifice was constructed over the last c.10 ka (Neall et al., 19 86) with activity ch aracterised by 0.1-2 km 3 sub-Plinian eruptions on c.330 yr interv als, interspersed wi th more frequent effusive dome-and BAF-forming events and lava flows (Alloway et al., 1995). Figure 6-1: Mount Taranaki (lower right) has produced mainly lava dome eruptions in the past 800 years with Block-and-Ash Flow deposits making up the fan between the Maero and Pyramid Stream and in the Hangatahua River (BAF – Block-and-Ash Flow, ppf – pumice pyroclastic flow). Star (top left) indicates the most distal outcrop discussed in the text. To the NNW of Mt. Taranaki are the south flanks of Pouakai volcano. The inset shows the Taranaki peninsula with the Taranaki Volcanic Lineament (SLI-Sugar Loaf Islands, K-Kaitake, P-Pouakai, T-Taranaki). Major onshore and offshore faults: IF-Inglewood Fault, MF- Manaia Fault, NF-Norfolk Fault, OF-Oanui Fault. Contours are 300 m. Modified after Sherburn and White (200 5). Chapter 6 Transition to Explosive Eruptions 180 The latest eruptive sequence of Mt. Taranaki comprises at least nine separate episodes, collectively mapped as the Maero Formati on (Druce, 1966; Neall et al., 1986). Each episode was characterised by effusion of highly-viscous andesitic magma to form lava domes within the summit crater and through a breach in the crater onto the uppermost NW flank. Collapses and small explosions fr om these lava domes generated BAFs and surges, with the largest BA Fs travelling up to 13.5 km down the Hangatahua River. The Burrell Lapilli erupti on episode took place in AD 1655, according to 14 C dating of charcoal in a Maori oven that was buried by tephra fall (Wellman, 1962) and dendrochronology dating of trees damaged by the fall (Druce, 1966). In earlier descriptions of this eruption, pumice fallout deposits were mapped to the ESE (Druce, 1966). Three lobes, directed to the NE, SE , and ESE, were rec ognised based on grain size analyses (Topping, 1972). The pumice flows and BAFs associated with the Burrell Lapilli are newly described in this study. 6.3 Methodology For petrographic, mineral chemistry, and textural studies, pumice samples were impregnated with the low viscosity polymer, methyl methacrylate and polymerised by Cobalt-60 gamma radiation. Glass and mineral microprobe analyses were carried out on polished thin sections of impregnated pumices and dense clasts using an energy dispersive spectrometer (JEOL JXA-840) with an accelerating voltage of 15kV, a beam current of 600 pA and 100 seconds live time. Glass analyses were performed with a beam defocused to 20 µm diameter. For bulk rock geochemistry, pumice and lithic clast specimens were crushed and ground using a tungsten carbide mill. Major and trace elements were measured on prepared glass discs and powder pellets, respectiv ely, using X-ray fluorescence spectrometry (Siemens SRS 3000) and calibrated to a suit e of 36 international standards at the University of Auckland. Porosity and permeability measurements were made using cores (approx. 2.54 cm diameter and height) of clasts that were dr illed in two or three mutually perpendicular orientations (depending on sample si ze; Fig. 6-2). From the known volume (V cylinder) and mass of the cores, the bulk density ( ρbulk ) was determined. A He-pycnometer was used to measure the helium accessible volume (V He ) of the sample, resulting in an Chapter 6 Transition to Explosive Eruptions 181 estimate of skeletal density ( ρskeletal ). However, even rocks of dense appearance commonly possess isolated cavities and crack s, and vesiculated rocks usually have isolated pores (V iso) in addition to a connected pore network, and therefore ρskeletal is always lower than ρsolid. In order to determine ρsolid, representative samples of grey (n=6), brown (n=4) and black (n=2) clasts were powdered and their volume determined by He-pycnometry. The isolated pore volume, V iso, can then be calculated and the bulk vesicularity determined by φ= (V He + V iso)/V cylinder. Figure 6-2: Variations in bulk vesicularity and connected po rosity of single clasts for grey pumice (a), banded pumice (b), and black and brown pumice (c). Crosses represent the range in vesicularity per clast using minimum, mean and maximum values (see in set in a). Variations in bulk vesicularity refer to single cores cut in two ( φcore) and overall clast variations with multiple cores ( φclast). Note different scale in b). See text for details. Long drilled cores cut into tw o were analysed for reproducibility and homogeneity in a specific orientation. Pumice samples appear homogeneous, with differences in φ of 0.1 to 8.0% over the range of φ=34.2-83.6%. However, single la rger pumice clasts with up Chapter 6 Transition to Explosive Eruptions 182 to seven drilled cores per sample in multiple orientations showed differences in φ of up to 19.9% (Fig. 6-2). The gas permeability of pumice cores was measured in a capillary flow porometer with air as the working gas using calculations after Rust and Cashman (2004) and Wright et al. (2007). Scanning electron microscopy images of car bon-coated samples were collected at Massey University and the University of Or egon using accelerating voltages of 20 kV and 10 kV, respectively. 6.4 Results 6.4.1 Field Observations 6.4.1.1 Fall Deposits The pumice fall deposits are well exposed on the steeper parts of the mountain near walking tracks, whereas outside the National Park boundary on flatter terrain, exposures are rare or the unit is disturbed by soil tillage. The proximal to medial deposits (within 10 km ) are characterised by pale grey and pale brown (weathered) subrounded to subangular pumice and darker grey angular to subangular lithics (commonly dense grey ande site). Overall, be tween 10-15% lithic clasts are scattered throughout the fall unit, with fewer, la rger clasts near the base and finer, more-abundant clasts near the top. Re verse grading of the whole unit is also occasionally observed. Rare accessory clasts also occur, including iron-stained, weathered, and hydrothermally altered andes ites. While lithic cl ast abundances vary greatly between exposures, there is an overall decrease with distance from dispersal axis and source. 6.4.1.2 Pumice Pyroclastic Flow Deposits Pumice pyroclastic flows associated with the Burrell Lapilli eruption travelled predominantly down the southern flanks (F ig. 6-3a), but were also deposited on the upper southeastern and northeastern sectors (Fig. 6-4b). Thr ee flow units are preserved on the southern flank. Chapter 6 Transition to Explosive Eruptions 183 The basal unit (1) is up to 1.5 m thick and comp rises pale grey pumice of up to 70 cm in diameter in a medium- to coarse-ash ma trix. The subrounded to rounded pumice blocks and lapilli are uniform in colour or contain patches or bands of darker grey. They also commonly contain inclusions of grey, angular, dense lithic la pilli. Larger pumice blocks are concentrated at or near the upper surface of the unit. Lithic clasts up to 50 cm diameter are present throughout the unit, most commonly as grey and dark grey dense andesites. The dense grey andesites are often bread-crusted, jigsaw-jointed or fractured (Fig. 6-3c). Occasionally, thei r concentration increases upwards through the unit. Rare hydrothermally altered and other dense andesite lithologies are present. Figure 6-3: Field photographs of a) succession of three pumice pyroclastic flow deposits on the upper south flanks; sketch shows a general assembly of pumice types and grey dense lithics, b) grey pumice clasts of unit 3, c) eroded surface into unit 2 showing the scattered grey pumice [1] from airfall, brown pumice [2], banded pale grey to dark brown pumice [4], and the dense fractured andesite clasts [L]; d) lower contact of a distal BAF deposit, c.13.5 km from source (star in Fig. 6-1). See text for field description. The middle unit (2) is up to 2 m thick and comprises subangular to rounded, brown to dark-brown pumice up to 45 cm in diameter. Common lithics of grey and dark-grey dense andesites are concentrated toward the base of the unit (c.10% by volume), while glassy black, vesicular to semi-vesicular clasts are evenly distributed throughout it (c.5% by volume). Grey, dense, bread-crusted and jointed blocks are rare. Accidental lithics comprise hydrothermally altered or weathered andesite and iron-stained clasts. The lower and upper contacts are gradational and marked by concentrations of banded Chapter 6 Transition to Explosive Eruptions 184 pumice (Fig. 6-3c). Unit 2 is in part conso lidated and is the only unit to show partial welding textures. In places, it is also str ongly oxidised. The partially welded parts of the unit form small ridges on the volcano flanks above 1900 m, capping and preserving the friable unit 1. Deposits of unit 2 are more re stricted than the others and are found only within 1.2 km of the crater rim. Figure 6-4: Distribution of Burrell Lapilli deposits : a) isopachs in cm including the BAF deposit (black) to the NW for reference, black squares represent mapping locations for fall deposits only, P-Pouakai, contours 100 m; b) isopleths for pumice clasts and pumice pyroclastic flow deposits on the upper flanks (black), see inset in a) for location; c) isopleths for lithic clasts, same outline as in b) . Numbers in b) and c) are clast diameters in cm. The uppermost unit (3) is >1 m thick, but of ten partly eroded. N ear the summit, it contains pumice clasts up to 70 cm in diamet er, which are usually concentrated on the surface, along with dense grey bombs and blocks (Fig. 6-3b). Units 1 and 3 are generally similar in clast composition and texture and occur separately or together up to 3.3 km from the crater rim. Chapter 6 Transition to Explosive Eruptions 185 6.4.1.3 Block-and-Ash Flow Deposits A fan of Maero Formation BAF and related deposits occur on the NW sector of Mt. Taranaki. Since the last events in AD 1755 and c.1860, box-canyons of up to 80 m wide and deep have been eroded into the unconsolidated volcaniclastics (see Fig. 3-4b). The primary BAF deposits often contain charcoal ised twigs, branches or logs, hence the stratigraphic sequence is well constrained by radiocarbon dates (see Appendix A and Fig. 3-4k-l). One sub-set of the deposits can be related to the Burrell Lapilli event. The most distal exposed Burrell -related BAF deposit is found 13.5 km from source beside Hangatahua River (Fig. 6-4a). It consists of a brownish-grey, 1.75 m thick, medium ash matrix-supported diamicton with 45% lapilli to fine block-sized clasts of dense grey andesite, with rare weathered and hydrothermal ly altered andesite and occasional grey pumice. Pumice is not strongly represented in the -1.5 to -1 φ fractions, comprising c.6% by volume. The deposit drapes a 1 cm charred humus la yer containing charcoal twigs (Fig. 6-3d), and charcoal logs are imbricated throughout the unit. The natural remanent magnetism of 10 clasts measured with a portable fluxga te magnetometer shows coherent alignment in six directions for 8 clasts, indicating a de position temperature above or near the Curie Point. This deposit can be correlated upstream through a succession of units in the Maero Stream (see Appendix A and B) . Characteristics of these nearer-source deposits are similar, with only very ra re pumice clasts present. 6.4.2 Volume Estimates The minimum fall volume was estimated as 3.2×10 8 m3 , by applying V=13.08T 0 bt2 (based on Cole and Stephenson, 1972) to the mapped isopachs (Table 6-1) where T 0 is the extrapolated maximum thickness and b t the distance over which the thickness of the deposit halves (see also Pyle, 1989). ArcGIS was used to calculate the mini mum bulk volume of pumice flow and BAF deposits, at 1.3×10 6 m3 and 2.5×10 6 m3 , respectively (Table 6-1). On a volumetric basis, the Burrell Lapilli eruption is classified as a VEI 4 sub-Plinian eruption (cf. Newhall Chapter 6 Transition to Explosive Eruptions 186 and Self, 1982). Applying the model of Carey and Sparks (1986) the eruption column is estimated to have been up to 14 km. Lithic contents of the fall deposi t were counted in the -4 to 0 φ grain size fractions of 5 samples at varying distances from source. The volumetric lithic content shows an exponential decrease with distance, with maximum lithic contents between 28 and 39 wt.% (approx. 13-20 vol.%) at distances of 2.5 to 2.9 km. By 9.8 km, it had dropped to 19 wt.% (approx. 8 vol.%). These estimates include a combination of juvenile and accidental clasts, since distinguishing them was difficult optically. Using an overall lithic population of 14%, a lithic volume of 1.7×10 7 m3 (Table 6-1) is contained in the fall, corresponding closely to the theoreti cal maximum volume that a dome could occupy in the present crater dimensions (c.2×10 7 m3 ). Given that some of the dome is probably also contained in the BAF, the lithic s in the fall probably also derive from the upper conduit. Table 6-1. Minimum volume and mass calculations of erupted tephra. 6.4.3 Clast Types and Textures The different clast types of the Burrell Lapi lli eruption episode provide insights into the eruptive processes and the mechanisms of pumice formation. For simplicity, all clasts that developed individual vesicles within their groundmass are termed pumice, regardless of whether an interconnected bubble network is developed or not. Those clasts that only display cracks and cavitie s are collectively termed lithics. For Chapter 6 Transition to Explosive Eruptions 187 quantitative analyses, pumice samples were cat egorized into four groups based on their colour; [1] uniform pale gr ey, [2] brown, [3] black, and [4] banded to transitional pumice clasts ranging from pale grey to dark brown. Dense grey clasts present in BAF and fa ll deposits have a hypocrystalline groundmass with abundant plagioclase microlites and Fe-T i oxides, and very rare minor cracks (Fig. 6-5a). Pumice clasts [1], [2], and [4] ha ve a well-developed, interconnected vesicle network with areas of coalesced vesicles th at are often a few centimetres long (Fig. 6- 5d). Black pumice clasts [3] show two vesi cularity populations; semi-vesicular, with minor isolated vesicles in a glassy, microlite containing matrix (Fig. 6-5b), and vesicular, with more abundant isolated vesicles that display partial coalescence (Fig. 6- 5c). Figure 6-5: Thin-section photographs illustrating basic vesicularity differences of juvenile clasts. a) dense grey lithic, b) semi -vesicular black pumice, c) vesicular black pumice with isolated and coalesced vesicles, crosses mark plagioclase crystals; d) grey pumice with isolated large single vesicles as well as larger coalesced vesicle. Scale bar is 100 µm in a-c and 500 µm in d. See text for details. Type 1 pumice has a broad range in vesicle- size distributions (Fig. 6-6a) with large single and coalesced bubbles (Fig. 6-6c). Larger bubbles are elong ated but show no deformation, and only few microlites occur in the glass. By contrast, brown pumice [2] has a broad but bimodal range in vesicle sizes (Fig. 6-6b). Large, coalesced bubbles are Chapter 6 Transition to Explosive Eruptions 188 present, but simple large bubbles are absent . The bubble network is often deformed and bubbles walls are thicker with microlites in the glass. Banded pumices [4] contain type s [1] and [2] (Fig. 6-6d). Th e grey bands within them are often coarsely vesicular with interconnected vesicles. Black vesicu lar clasts [3] have comparatively small bubbles of uniform size, wh ich also appear in thin sections to be commonly isolated (Fig. 6-5c). The difference in pumice types is also expres sed in their solid densities (grey: 2.74 gcm-3 ; brown: 2.79 gcm -3 ). Figure 6-6: SEM images of grey (a) and brown (b) pumice (note a and b are binarised; black=vesicles, white=glass + crystals); c) show s a large coalesced vesicle; d) consists of three SEM images showing the tr ansition from grey to brown in banded pumice. Scale bar is 10 µm in c), otherwise 100 µm. Chapter 6 Transition to Explosive Eruptions 189 6.4.4 Mineralogy and Mineral Chemistry All pumice clasts contain common euhedral crystals of plagioclase, clinopyroxene, hornblende, Fe-Ti oxides, and mi nor biotite. Often crystals are broken fragments. In the glassy, highly vesiculated rocks, microlites of plagioclase, and to a lesser extent clinopyroxene, appear predominantly at the inte rsection of adjacent ve sicles or next to phenocrysts. In addition, hornbl ende and biotite crystals are fresh, showing no oxidation or breakdown reaction rims (Fig. 6-7a). Figure 6-7: Hornblende reaction textures in different clast types: a) fresh hornblende with no reaction rims in pumice, b) single Fe-Ti oxide crystals are attached to the hornblende rims in black semi-vesicular pumice, c) hornblende in dense grey lithic clast shows resorption textures and is partially replaced by clinopyroxene, plagioclase and Fe-Ti oxide crystals or is fully replaced (lower left ); note abundant plagioclase microlites in groundmass. Scale bar is 100 µm. Grey lithics, the most abunda nt lithic clast type within pumice flow and fall deposits, are dense, hypocrystalline, and comprise pl agioclase, clinopyroxe ne, hornblende, Fe-Ti oxides, and biotite. Hornblende and biotit e are partially oxidised, with Fe-Ti oxide reaction rims; in some cases the crystals are replaced by minute plagioclase, clinopyroxene and Fe-Ti oxides (Fig. 6-7c). Black dense and vesicular clasts, common in the pumice pyroclastic flow deposits, are glassy with the same mineral assemblage as grey, dense lithics, except with lower biotite contents. Hornblende appears fresh, with minor Fe-Ti oxide rim alteration (Fig. 6-7b). Chapter 6 Transition to Explosive Eruptions 190 6.4.5 Bulk Rock Geochemistry Whole rock analyses of pumi ce clasts and juvenile lithic s from the pumice pyroclastic flow, BAF and fall deposits are shown in Tabl e 6-2. Pumice clasts fr om units 1 to 3 and black dense and vesicular lithics show very little variation in major and minor element geochemistry. Silica conten ts range between 55.5 and 56.0 wt.% (Fig. 6-8). The common grey dense lithics in the deposits show a broader range in composition, with silica varying between 56.4 and 59.5 wt.%. Figure 6-8: Bulk rock geochemistry of pumice and gr ey andesite clasts in a multi element oxide vs. SiO 2 diagram. The calculated fractionation trend pumice – grey lithics is in good agreement for the majority of clasts (solid lin e) with some variation for the most evolved clast (dashed line). See Table 6-3 for details. Data for all major elements show linear rela tionships (Fig. 6-8) with the trends being best explained by fractionati on from pumice to grey lithics involving 45% hornblende, 34% plagioclase, 17% clinopy roxene, and 4% Fe-Ti oxides. This mineral assemblage represents common macroscopic and microscopic cumulates within Mt. Taranaki rocks. Sample P10 has been used as a starting co mposition based on its highest Mg# of 49.9. From this assumed magma composition, crysta ls have been subtracted from the magma composition using average mineral compositions for plagioclase and Fe-Ti oxides and average rim compositions for hornblende and clinopyroxene obtained by electron microprobe (Table 6-3). The modelled frac tionation is in very good agreement with observed rock compositions up to 16.5% of fractionation, however, there is a considerable deviation for Al2 O 3 , FeO and MgO contents be tween the model and the Chapter 6 Transition to Explosive Eruptions 191 Chapter 6 Transition to Explosive Eruptions 192 Chapter 6 Transition to Explosive Eruptions 193 Chapter 6 Transition to Explosive Eruptions 194 single rock at 59.5 wt.% SiO 2 (Table 6-3). An alternative explanation for these trends could also be mixing or mingling of magm as, although glass compositions (see below) do not support this. 6.4.6 Glass Composition Glass compositions were analysed for brown (unit 2), grey pumice clasts (unit 1 and 3), and black pumice present througho ut units 1-3 (Fig. 6-9; Ta ble 6-4). The grey pumices of unit 1 range in silica content from 64.6- 67.4 wt.% whereas the same pumice type of unit 3 only varies betw een 64.4 to 65.8 wt.% SiO 2 . Brown pumices (unit 2) have a higher silica content, ranging from 64.5- 68.2 wt.%. Black pumice glass compositions differ from units 1 and 3 pumice in their hi gher range in silica values (64.9-69.3 wt.%), but show a very similar trend in Al2 O 3 vs. SiO 2 as observed in brown pumice with a slightly steeper slope. Glass compositions of individual lapilli are homogeneous as illustrated in the inset of Figure 6-9. Figure 6-9: Groundmass glass compositions of pumice types presented in the Al 2 O 3 vs. SiO 2 diagram. Modelled glass composition changes due to plagioclase and clinopyroxene crystallisation (thick solid lin e) and is in good agreement with linear regression line (thin solid line). The small in set shows six data points of one lapillus (SD20) demonstrating relative glass homogeneity. The dimensions of the box are 1 wt. % for Al 2 O3 and SiO 2 . Chapter 6 Transition to Explosive Eruptions 195 Chapter 6 Transition to Explosive Eruptions 196 The general trend of all data (thin solid line in Fig. 6-9) can be modelled using plagioclase microlite and clinopyroxene rim analyses. The thick line in Figure 6-9 represents the change in glass compositions by crystallisation of 76% plagioclase and 24% clinopyroxene calculated by mass balance. It is noted that unit 1 pumice data lie on a flatter slope, although still situated within the entire data set. This may suggest a stronger involvement of clinopyroxene during late stage evol ution of glass chemistry. 6.4.7 Porosity and Permeability Porosity (n=221) and permeability (n=83) we re measured on cores of grey, brown and banded pumices and black vesicu lar clasts. There is a wide range in connected porosity and bulk vesicularity for all analysed cl ast types (Fig. 6-10). The dataset clearly illustrates a concentration of bulk vesiculari ties between c.57-77% , with a broader range in isolated pore volume up to 10%. Clasts with φ <54% have similar isolated pore volumes in a restricted range and these seem to decrease with decreasing φ. The gap in the dataset (36.0< φ <47.8%) may be due to low sample numbers but is more likely to be real (see discussion). The highest φ values (up to 83.6%) are observed for brown pumices. Figure 6-10: Connected porosity vs. bulk vesicularity of all pumice types. Brackets represent 95% confidence limit for the mean of each pumice population. Solid lines represent 0% and 10% and dashed line 5% isolated pore volume. Chapter 6 Transition to Explosive Eruptions 197 Figure 6-11: Connected and bulk vesicularity vs. permeability. a) data of this study with upper and lower data limits (black lines) of y=5×10 -19 x -4.5314 and y=6×10 -21 x -4.5314 , respectively. Note there are six specimens with three cores cut in three mutually perpendicular directions. Upper inset shows cores cut in two perpendicular directions. b) comparison of our data with published literature: Montserrat (Melnik and Sparks, 2002), Big and Little Glass Mountains (Rust and Cashman, 2004), Pichincha (Wright et al., 2007 ); grey lines are limits of Klug and Cashman (1996). Grey pumice clasts of unit 1 and 3 show si milar ranges in connected porosities from 53.1-73.5% and 44.1-72.7%, respecti vely, with isolated pore volumes, varying between 2.0-8.8% (unit 1) and 2.0-4.6% (u nit 3). There is an overall positive correlation of permeability and connected porosity (Fig. 6-11a), despite the fact that permeability ranges over three orders of magnitude. For example, two grey pumice clasts with φ= 6 4.8 and 64.9% differ considerably in permeability (1.0×10 -1 1 m2 and 3.0×10 -1 2 m2 ). However, data from three mutually perpendicular cores per sample show no positive Chapter 6 Transition to Explosive Eruptions 198 correlation of porosity with permeability (Fig. 6-11a). Above φ= 5 9.8%, all four investigated clast types show that the highest permeability occurs in the core samples having an intermediate bulk vesicularity value. Combining these data with the observations on pumice core samples in two mutually perpendicular directions (see inset in Fig. 6-11a), suggest that this is the result of an anisotropic bubble network which is present in all pumice types. In bande d pumice clasts this effect is particularly strong, where coarsely vesicular bands were probably able to effectively channel the pre- and syn-eruptive gas flows. Unusually, the permeability data for Burrell Lapilli pumice are on average one log unit higher than other silicic pumice (Fig. 6-11b). This could be caused by larger aperture sizes of connected bubbles and may be rela ted to lower viscosities in the Burrell magmas (e.g. Saar and Manga, 1999). If lowe r viscosities cause higher permeability of pumice clasts, then greater post-fragmentation bubble expans ion (and coalescence) is highly likely. 6.5 Discussion The ejecta of this eruption span a range of vesicularity, bulk silica content and colour, with systematic variations in hornblende phenocryst rim morphologies, microlite contents and glass compositions. These variations are related to the order of eruption in the following way: 1. Grey lithic clasts, whic h probably represent the earliest-extruded magma of the AD 1655 episode, are incorporated in ejecta fr om all phases of the eruption. These are hypo-crystalline rocks of low vesicularity and low permeability (the latter mostly represents cooling-induced mi cro-cracks). They have bulk ro ck silica contents that are 0.4 to 4 wt.% higher than co-erupted pumi ce clasts, contain hornblende and biotite phenocrysts with strongly altered/reso rbed rims and a groundmass dominated by plagioclase and clinopyroxene. 2. Pale grey pumice with porosities of betw een 53-74% and the hi ghest isolated pore volumes (up to 8.8%) represent the earliest e xplosive phase of the episode. This pumice contains a number of elongated large, single an d coalesced vesicles within an overall broad pore-size distribution. Chapter 6 Transition to Explosive Eruptions 199 3. Brown pumice, displaying the highest por osities (up to 84%) wa s next erupted. The large single vesicles are absent from these pumices, which, compared with the early pale-grey pumice, show a more restricted pore size-distribution, a more strongly deformed network, thicker walls between bubbles , and a higher microlite content. These differences are also observed between the different coloured zones within banded clasts. Both brown and the initial pale-grey pumice have identical ranges in glass composition, between 64-68 wt.% SiO 2 . 4. Black clasts, erupted as a minor concomita nt component during the middle stages of the explosive phase, are either low- or mode rately-vesicular, and are characterised by having isolated pores in a glassy microlite- rich matrix. They have a similar range in glass compositions to the brown and initial pale grey pumice, although extending slightly higher to 69 wt.% SiO 2 . 5. Banded pumice clasts predominantly comp rise grey and brown pumice networks and show macroscopically complex text ures of stretching, banding and folding. 6. The final explosive phase is represented by a pale grey pumice of similar appearance to that firstly erupted. Like the initial pumice, it has a low-moderate porosity range (<73%), but shows lower isolat ed pore volumes than the earliest pale-grey ejecta. This pumice also has the lowest silica glass (64-66 wt.% SiO 2 ), with the most restricted composition range, compared to earlier ejecta. To understand the phases of this eruption, its transitions in eruptive style and the range in physical and chemical properties of the ejecta, a range of magmatic processes within the upper conduit need to be considered. The factors causing different clast types are discussed in chronological order and divided into pre-climactic and syn-climactic processes, which form the basis for reconstructing the eruption dynamics. 6.5.1 Pre-climactic Conditions 6.5.1.1 Grey Lithics No large-scale magma reservoirs are eviden t in the upper crust at Taranaki (Sherburn and White, 2005), suggesting that the final ma gma storage level is at about 10-11 km depth. Magmas ascend through 6 km of Tertiary sediments into the 2.5 km high edifice. Chapter 6 Transition to Explosive Eruptions 200 Gravity measurements indicate that multiple dyke intrusions have created a c.5 km-wide dyke/stock complex below Mt. Ta ranaki (Locke and Cassidy, 1997). Dyke propagation provides a first-order cont rol of ascent rates of the initial rising magma. Rise rate controls a series of physic al and chemical processes, including degree of crystallisation, fractiona tion, degassing, vesiculation, viscosity and fragmentation that can be seen in the final ejecta. During rise, the t ypically hydrous Taranaki melts move outside the pressure/temperature and pH 2 O stability fields for hornblende and hence varying degrees of disequilibrium a nd hornblende breakdown occur that may be used to constrain magma rise processes. Decompression experiments (130-4 MPa) on Soufrière Hills andesites showed that hornblende can spend up to 6 days outside its stability field before showing any breakdow n reactions (Rutherford and Devine, 2003). The upper pressure boundary for hornblende stability at Taranaki corresponds to c.5.2 km below the summit (using densities of 2400 kgm -3 for the edifice and 2700 kgm -3 for the dyke/stock complex; Locke and Cassidy, 19 97). However, Sato et al. (1999) also showed that groundmass pargasitic hornblende in Unzen dacite is stable up to 70 MPa at c.880 °C, which is equivalent to >3 km de pth in a Taranaki conduit. These data imply that hornblende-breakdown reactio ns in Mt. Taranaki rocks li kely occurred in the upper conduit and/or within a dome. The grey, dense, dome-derived lithics have e volved from the same magma as co-erupted pumice. The difference in SiO 2 content can be modelled by fractionation of hornblende, plagioclase, clinopyroxene, and Fe-Ti oxides (see Fig. 6-8) . The major involvement of hornblende in the fractionation process is essential because it is the only major, low- silica mineral phase (a verage 41.7 wt.% SiO 2 ) to be able to move the bulk rock silica content by 4 wt.% during ascent. Although crystallisation and fractionation processes during ascent influences the ability of a magma to flow due to an increase in crystallinity and silica contents in the melt phase (Hammer et al., 1999), another prof ound impact on magma evolution is the degree of degassing. Gas escape through dyke wa lls is important (e.g. Tuffen et al., 2003), especially if the dyke intrudes in to porous media, although degassing along marginal shear zones is usually dominant (J aupart and Allegre, 1991; Stasiuk et al., 1996). Once the fractionated “grey” magma enters shallower levels, magma processes are governed by rheological changes due to increasing gas loss, particularly within the edifice, and subsequent, degassing- and decompression-induced groundmass Chapter 6 Transition to Explosive Eruptions 201 crystallisation (Bl undy and Cashman, 2001). Nucleation and growth of predominantly plagioclase microlites within the groundmass (Fig. 6-5) considerably increases melt and magma viscosities, thus reduc ing the ascent rate. As soon as a dome or cryptodome is emplaced on top of a conduit, it has such extremely low gas permeability that it forms an effective cap (Melnik and Sparks, 2002; Mueller et al., 2005). The remnant summit dome at Mt Taranaki (erupted approximately AD 1850-60 ) has permeability that ranges between 6.7×10 -1 3 and 5.8×10 -1 4 m2 at φ= 1 6.1-23.7% (Fig. 6-11b). However, these already low values are dominated by cooling-induced micr o-cracks (as seen macroscopically) and the permeability at th e time of emplacement would have been much lower. 6.5.1.2 Grey Pumice (Unit 1) Sealing and/or capping the conduit leads towa rds a closed-degassing system and also a partial stagnation of the magma in the upper conduit. Slow rates of vesicle nucleation and growth may still occur in the upper conduit (Fig. 6-12a, b, to level 1) over extended periods. This slowly forces the viscous overlying magma upwards, and drives dome growth (Fig. 6-12b). At the same time, crysta llisation continues, especially the growth of microlites. Should magma in the slowly vesiculating level 1 domain be explosively ejected during these slow vesiculation periods , the ejecta (pumice) should display a heterogeneous bubble network, containing a large range in vesicle sizes (cf. Klug et al., 2002). The upper-conduit processes of water lo ss/bubble growth and decompression encourages late-stage microlite crystallisat ion of mostly plagioclase and clinopyroxene. This, in turn, shifts liquid-compositions to higher silica contents (Fig. 6-9), and subsequently increases melt and magma viscos ities (Richet et al., 2000; Sparks et al., 2000). This leads to stretching of origina lly spherical bubbles as they expand, under steadily-applied shear stress. Melt films between bubbles al so thin and rupture as vesicles grow (cf. Manga et al., 1998; Klug et al., 2002). Any onset of vesicle coalescence within the upper conduit increases permeability of the magma foam. This eventually leads to an increased storage of additional volatiles supplied from underlying magma and a progressively higher gas pressure below the dome/pl ug (cf. Cashman et al., 2000). Chapter 6 Transition to Explosive Eruptions 202 Figure 6-12: Reconstruction of eruptive events during the Burrell Lapilli eruption. Changes in bulk rock silica contents are illustrated in Stage a. Bu bble nucleation levels 1-3 correspond with erupted units 1-3. See text for further details. Once uncapped, magma following from deeper in the conduit, rises rapidly, hence the time for crystallisation is short and the melt viscosity changes little. Under rapid depressurisation, vesiculation also proceeds rapidly, leading to pumice ejecta with narrow ranges in bulk silica content (i.e ., 0.5 wt.%; Fig. 6-8) , and a relatively homogenous size-distribution of vesicles. 6.5.1.3 Brown Pumice (Unit 2) As slow dome growth continues or stagnates and the overlying pressure slowly but steadily reduces, the nucleation level pr opagates downwards. Here, slower gas- exsolution and hence slower bubble-nucleatio n and growth rates promote microlite Chapter 6 Transition to Explosive Eruptions 203 crystallisation at higher magma temperatures as the degree of undercooling of the melt is less pronounced (Hammer and Rutherford, 2002; D’Oriano et al., 2005). Hence, the gradually formed (brown) magma foam (Fig. 6-12b) is characterised by thicker bubble walls containing more microlites, which concomitantly grew with the bubbles. The bubble-size distribution has generally a sma ller, narrower range. In addition, abundant microlites form, causing a broader range in glass composition compared to the (grey) magma foam above. The difference between brown and colourless matrix pumice glasses is likely caused by submicroscopic Fe 3+ -rich oxide “nanolites” (hematite and/or magnetite), which also increase magnetic su sceptibility in brown pumice (Schlinger et al., 1988; Paulick and Franz, 1997). Nucleation and growth of Fe 3+ -rich oxide “nanolites” predominantly in the brown ma gma is probably induced by slower volatile exsolution and hence, a higher oxidation state of the magma (Paulick and Franz, 1997). Post-depositional preci pitation of Fe-Ti oxides probably led to further glass colouring but it could not have affected the glass composition (see Fig. 6-9). It is inferred that, despite higher microlite contents, the brown magma foam retained a higher temperature and hence had a lower viscosity than the earlier formed grey magma foam. 6.5.2 Syn-climactic Conditions 6.5.2.1 Black, Banded, and Unit 3 Grey Pumice Clasts Initiation of the explosive pha se caused instant vesiculation and further bubble growth in the grey (unit 1) and brown magma foams (Fig. 6-12c ). Further, an immediate pressure drop, as well as magma ascent within the conduit caused rapid downward progression of the bubble nucleation to level 3. Here, the grey pumice of unit 3 was developed. Rapid volatile ex solution caused high degrees of undercooling of the melt, permitting syn-crystallisation of plagioclas e ± clinopyroxene, however, rapid ejection and cooling meant that the melt silica content was only driven up to 65.8 wt.%. The unit 3 pumice has slightly higher isolated pore vol ume, also suggesting a less mature bubble network than unit 1 pumice. As the three domains of magma rose within the conduit, the slightly di fferent viscosities contributed to further differences in clast textures. For example, the brown magma foam was easily plastically deformed, and bubbles co alesced more rapidly due to shearing of Chapter 6 Transition to Explosive Eruptions 204 the ascending magma (Fig. 6-12d). Higher connected porosities and permeabilities compared to the grey magma foams seems to be controlled by larger aperture size rather than greater vesicle size (cf. Saar and Manga, 1999). The occurren ce of black pumice clasts predominantly within unit 2 pyrocla stic flows and their glass composition being similar to brown pumice (Fig. 6-9) suggests a genetic link. The poorly vesicular texture of black pumice and its low abundance in th e pyroclastic flow deposit (c.5% by volume) suggests that it could represent a conduit-ma rgin variant of the brown pumice. They may have been derived under higher temperature conditions developed by friction during shearing of magma along the conduit walls (cf. Rosi et al., 2004). Black clasts show the highest difference in permeability per clast (Fig. 6-11a) with only medium connected porosities of 44.8% and 47.1%, but strongly anisotropic permeable clasts with differences of 1.2 log units. The larg e difference between semi-vesicular and vesicular pumice clasts ( φ=36.0-47.8%; Fig. 6-10) could be due to their genetic position in respect to the conduit margins, with an abrupt transition between a dense marginal facies and a porous interior facies. In addition, due to variable ascent veloci ties across and along the conduit, the lower- viscosity brown magma could partly intrude and mingle into the upper grey magma (unit 1) to produce flow bandi ng (Fig. 6-12d). The same proce ss occurred at the base of the brown magma. The banded pumices also clearly show an anisotropic bubble network (Fig. 6-11a). Contrary to a horizontal textural zonation of a conduit, where banded pumice forms in the outer margins and coarsely vesicular pumice occupies the core (as proposed by Polacci et al., 2001, Rosi et al., 2004, and D’Oriano et al., 2005), banded pumices of the Burrell Lapilli eruption are thought to be the result of vertical gradients within the conduit. Although D’Oriano et al. (2005) also argue for a vertical zonation of a conduit, they apply it only to th e degassed and variably crystalline plug of the upper conduit. 6.5.3 Eruption Dynamics Evolution of magmas in the dome and the upper conduit were governed by decompression-induced microlite crystallisa tion and growth in the melt fraction. This caused a large pressure gradient, due to excess fluid pressure in the melt as well as large viscosity differences by increasing crystallinity and melt silica content, and decreasing Chapter 6 Transition to Explosive Eruptions 205 dissolved water contents (S parks, 1997). Over-pressurisa tion from below the dome was probably an additional contributing factor for dome rupture and collapse (Fig. 6-12a). The dome collapse generated a large BAF, which due to the pre-existing crater morphology, was directed down the NW flank and travelled up to 13.5 km. The presence of primary grey pumice clasts within the BAF deposit further suggests over- pressurisation of the dome from below was a major factor. The explosive removal of lava dome material was accompanied by a vertically-directed sub- Plinian eruption of the fragmented grey pumice foam (unit 1) with dome clasts and associated conduit rocks incorporated and transported through the column. The strength of fragmentation was maintained until the fragmentation front reached a change in vesicularity. Variable degrees of vesicularity, visc osity and temperature down and across the stratified conduit were caused by the formation of three distinct magma foams. The partially vesiculated brown magma foam had the lowest melt viscosity and hence the lowest pressure build-up by vesiculation. Th is in turn led to lower exit velocities and lower eruption column heights. In static experiment Bu rgisser and Gardner (2004) showed that bubble coalescence is time- dependent and had already occurred by porosities of c.43%. A sufficiently coales ced bubble network prior to fragmentation may decrease fragmentation rates, and hen ce could produce the low, unstable eruption columns inferred for the eruption of the brown pumice (unit 2). However, this does not exclude post-fragmentation bubble expansion and bubble deformation during emplacement as a further contributing factor to changes in bubble shapes and therefore permeabilities (e.g. Thomas et al., 1994). Fluctuations in column height and explos ive energy are common during sub-Plinian and Plinian eruptions where the column partially, or completely, collapses to produce pumice pyroclastic flows. This was observed, for example, during the 22 nd July 1980 eruptions at Mt. St. Helens where three i ndividual sub-Plinian pulses with column heights to 14 km removed half of the pre-existing dome of 4.7×10 6 m3 (Christiansen and Peterson, 1981; Moore et al., 1981). At Mt. Taranaki, no brown pumice is found in preserved fall deposits, which indicates that the brown magma was not fragmented and ejected as efficiently as the grey magma foam above and below it. The sudden drop in exit velocity associat ed with the eruption of brown magma foam caused instabilities within the eruption column, implying a lower energy, “boiling-over” type eruption (cf. Ta ylor, 1958). Hence, grey pumice pyroclastic flows were generated before and after the brown unit (Fig. 6-12d). The minimum Chapter 6 Transition to Explosive Eruptions 206 estimate of erupted brown magma is approximately 6.9×10 4 m3 (DRE) corresponding to a cylindrical conduit length of c.880 m (r=5 m) , a sufficient segment within the conduit to affect magma ascent and fragmentation rates. Another potential means to alter exit velocity, fragmentation rate and hence column height is to widen the vent, but this can only be achieved at the very top of the vent and, it would also have affected the fragmentation of the grey, unit 3 magma. The boiling-over type of eruption at Mt. Taranaki is also consistent with higher emplacement temperatures of brown pyroclastic flows (unit 2) compared to units 1 and 3. Isopach and isopleth mapping suggests that th e eruption column was initially directed towards the SE, and then the dispersal axis progressively rotated northeastwards. Lithic isopleths and a normal followed by inverse grading of lithic clasts in the fall deposit may also suggest that the eruption occurred in two sub-Plinian pulses, or, perhaps more likely, was interrupted by a transient waning in the magma ascent rate. Pulsatory, sub- Plinian eruptive periods have been observed for example at the June/July eruptions at Mt. St. Helens (Christian sen and Peterson, 1981; SEAN 05:06; 05:07) at the 1991 pre- climactic eruptions at Mt. Pinatubo, Philippine s (Hoblitt et al., 1996) and at El Chichon, Mexico (SEAN 07:03). The total eruption column height at its highest energy phases was probably up to 14 km above the crater (based on a pplication of the isopach and isopleth data to models from Carey and Sparks, 1986) and transporte d fine ash more than 250 km east and northeastwards over the North Is land (Eden and Frogatt, 1996). 6.6 Conclusions It has been demonstrated through this AD 1655 Burrell eruption case -study that simple changes in magmatic conditions of a single hydrous andesitic magma during ascent can explain sudden changes in eruptive explos ivity and hazard potential. Our model of a stratified conduit with varying magma composition, vesicularity and viscosity can explain all observed pumice variations. Transitions from effusive to explosive phase s within these types of andesitic eruption appear to be mainly governed by internal/magmatic factors, including 1) dyke propagation and sealing, 2) degr ee of fractionation of magma, and 3) magma ascent rate and magma volume. The faster the magma ascends and the faster the dyke is sealed to Chapter 6 Transition to Explosive Eruptions 207 reduce gas loss, the more effectively a magm a foam is developed that can fragment explosively. Textural and colour variations of pumice clasts have been recognised at other volcanoes (Donoghue et al., 1995; Bourdier et al., 1997; Gardner et al., 1998; Clynne, 1999; Polacci et al., 2001; Rosi et al., 2004 and D’Oriano et al., 2005). Where these clast variations occur within individual stages of eruptive sequences (e.g. within layered fall deposits), they could also be explained by a similar model to that described here. Chapter 6 Transition to Explosive Eruptions 208 6.7 References Alloway, B., Neall, V.E. and Vucetich, C.G. 1995. Late Quaternary (post 28,000 year B.P.) tephrostratigraphy of northeast and central Taranaki, New Zealand. Journal of the Royal Society of New Zealand 25, 385-4 58. Anderson, A.T. 197 9. Water in some hypersthenic magmas. Journal of Geology 87, 509- 531. Blundy, J. and Cashman, K. 2001. Ascent-driven crystallisation of dacite magmas at Mount St Helens, 1980 -19 86. Contributions to Mineralogy and Petrology 140, 631-6 50. Blundy, J., Cashman, K. and Humphreys, M. 2006. Magma heating by decompression-driven crystallization beneath andesite volcanoes. Nature 443, 76-8 0. Bourdier, J.L., Pratomo, I., Thouret, J.C., Boudon, G. and Vincent, P.M. 1997. Observations, stratigraphy and eruptive processes of the 1990 eruption of Kelut volcano, Indonesia. Journal of Volcanology and Geothermal Research 79, 181- 203. Burgisser, A. and Gardner, J.E. 2004. Experimental constraints on degassing and permeability in volcanic conduit flow. Bulletin of Volcanology 67, 42-5 6. Carey, S. and Sparks, R.S.J. 1986. Quantitative models of the fallout and dispersal of tephra from volcanic eruption columns. Bulletin of Volcanology 48, 109 -12 5. Cashman, K.V. and McConnell, S.M. 2005. Multiple levels of magma storage during the 1980 summer eruptions of Mount St. Helens, WA. Bulletin of Volcanology 68, 57-75. Cashman, K.V., Sturtevant, B., Papale, P. and Navon, O. 2000. Magmatic fragmentation. In: Sigurdsson, H., Houghton, B., McNutt, S.R., Rhymer, H. and Stix, J. (eds.) Encyclopedia of Volcanoes. Academic Press. San Diego. pp. 419-4 30. Christiansen, R.L. and Peterson, D.W. 1981. Chronology of the 1980 eruptive activity. In: Lipman, P.W. and Mullineaux, D.R. (eds.) The 1980 eruptions of Mount St. Helens, Washington. U.S. Geological Survey Professional Paper 1250, 17-30. Clynne, M.A. 1999. A complex magma mixing origin for rocks erupted in 1915, Lassen Peak, California. Journal of Petrology 40, 105-1 32. Cole, J.W. and Stephenson, T.M. 1972. Calculation of the volume of a tephra deposit. Appendix. In: Cole, J.W. Distribution of high-alumina basalts in the Taupo Volcanic Zone. Geology Department Victoria University of Wellington Publication 1, 13-1 5. DeGroat - Nelson, P.J., Cameron, B.I., Fink, J.H. and Holloway, J.R. 2001. Hydrogen isotope analysis of rehydrated silicic lavas: implications for eruption mechanisms. Earth and Planetary Science Letters 185, 331 -34 1. Chapter 6 Transition to Explosive Eruptions 209 Donoghue, S.L., Gamble , G.A., Palmer, A.S. and Stewart, R.B. 1995. Magma mingling in an andesite pyroclastic flow of the Pourahu Memb er, Ruapehu volcano, New Zealand. Journal of Volcanology and Geothermal Research 68, 177 -19 1. D'Oriano, C., Poggianti, E., Bertagnini , A., Cioni, R., Landi, P., Polacci, M. and Rosi, M. 2005. Changes in eruptive style during the A.D. 1538 Monte Nuovo eruption (Phlegrean Fields, Italy): the role of syn-eruptive crystallization. Bulletin of Volcanology 67, 601-6 21. Druce, A.P. 1966. Tree-ring dating of recent volcanic ash and lapilli, Mt Egmont. New Zealand Journal of Botany 4, 3-41. Eden, D.N. and Froggatt, P.C. 1996. A 6500-year-old history of tephra deposition recorded in the sediments of Lake Tutira, eastern North Island, New Zealand. Quaternary International 34/36 , 55-64. Elsworth, D. and Voight, B. 2001. The mechanics of harmonic gas pressurization and failure of lava domes. Geophysical Journal International 145, 187- 198. Friedman, I. and Long, W. 1976. Hydration rate of obsidian. S cience 191, 347-3 52. Gardner, C.A., Cashman, K.V. and Neal, C.A. 1998. Tephra-fall deposits from the 1992 eruption of Crater Peak, Alaska: implications of clast textures for eruptive processes. Bulletin of Volcanology 59, 537- 555. Grove, T.L., Elkins - Tanton, L.T., Parman, S.W., Chatterjee, N., Müntener, O. and Gaetani, G.A. 2003. Fractional crystallisation and mantle-melting controls on calc-alkalin e differentiation trends. Contributions to Mineralogy and Petrology 145, 515 -53 3. Hammer, J.E. and Rutherford, M. 2002. An experimental study of the kinetics of decompression- induced crystallization in silicic melt. Journal of Geophysical Research - Solid Earth 107, 1-24. Hammer, J.E., Cashman, K.V., Hoblitt, R.P. and Newman, S. 1999. Degassing and microlite crystallization during pre-climactic events of the 1991 eruption of Mt. Pinatubo, Philippines. Bulletin of Volcanology 60, 355-3 80. Hammer, J.E., Cashman, K.V. and Voight, B. 2000. Magmatic processes revealed by textural and compositional trends in Merapi dome lavas. Journal of Volcanology and Geothermal Research 100, 165- 192. Hoblitt, R.P., Wolfe, E.W., Scott, W.E., Couchman, M.R., Pallister, J.S. and Javier, D. 1996. The preclimactic eruptions of Mount Pinatubo, June 1991. In: Newhall, C.G. and Punongbayan, R.S. (eds.) Fire and mud: eruptions and lahars of Mount Pinatubo, Philippines. University of Washington Press. Seattle. pp. 457-511. Jaupart, C. and Allegre, C.J. 1991. Gas content, eruption rate and instabilities of eruption regime in silicic volcanoes. Earth and Planetary Science Letters 102, 413-4 29. Klug, C. and Cashman, K.V. 1994. Vesiculation of May 18, 1980 , Mount St. Helens magma. Geology 22, 468- 472. Chapter 6 Transition to Explosive Eruptions 210 Klug, C. and Cashman, K.V. 1996. Permeability development in vesiculating magmas: implications for fragmentation. Bulletin of Volcanology 58, 87-10 0. Klug, C., Cashman, K.V. and Bacon, C.R. 2002. Structure and physical characteristics of pumice from the climactic eruption of Mount Mazama (Crater Lake), Oregon. Bulletin of Volcanology 64, 486- 5 01. Laursen, T. and Lanford, W.A. 1978. Hydration of obsidian. Nature 276, 153- 156. Locke, C.A. and Cassidy, J. 1997. Egmont Volcano, New Zealand: three-dimensional structure and its implications for evolution. Journal of Volcanology and Geothermal Research 76, 149-16 1. Manga, M., Castro, J., Cashman, K.V. and Loewenberg, M. 1998. Rheology of bubble-bearing magmas. Journal of Volcanology and Geothermal Research 87, 15-2 8. Massol, H., Jaupart, C. and Pepper, D.W. 2001. Ascent and decompression of viscous vesicular magma in a volcanic conduit. Journal of Geophysical Research-Solid Earth 106, 1622 3-1 624 0. Matthews, A.J., Barclay, J., Carn, S., Thompson, G., Alexander, J., Herd, R. and Williams, C. 2002. Rainfall-induced volcanic activity on Montserrat. Geophysical Research Letters 29, 1644. McMillan, P., Peraudeau, G., Holloway, J. and Coutures, J.P. 1986. Water solubility in a calcium aluminosilicate melt. Contributions to Mineralogy and Petrology 94, 178-1 82. Melnik, O. and Sparks, R.S.J. 2002. Dynamics of magma ascent and lava extrusion at Soufrière Hills Volcano, Montserrat. In: Druitt, T.H. and Kokelaar, B.P. (eds.) The eruption of Soufrière Hills Volcano, Montserrat, from 1995 to 1999. Geological Society, London, Memoirs 21, 153 -1 71. Moore, J.G., Lipman, P.W., Swanson, D.A. and Alpha, T.R. 1981. Growth of lava domes in the crater, June 1980- January 1981. In: Lipman, P.W. and Mullineaux, D.R. (eds.) The 1980 eruptions of Mount St. Helens, Washington. U.S. Geological Survey Professional Paper 1250, 541 -54 8. Mueller, S., Melnik, O., Spieler, O., Scheu, B. and Dingwell, D.B. 2005. Permeability and degassing of dome lavas undergoing rapid decompression: an experimental determination. Bulletin of Volcanology 67, 526- 538. Neall, V.E., Stewart, R.B. and Smith, I.E.M. 1986. History and petrology of the Taranaki volcanoes. In: Smith, I.E.M. (ed.) Late Cenozoic volcanism. Royal Society of New Zealand Bulletin 23, 251-2 63. Newhall, C.G. and Self, S. 1982. The Volcanic Explosivity Index (VEI) – an estimate of explosive magnitude for historical volcanism. Journal of Geophysical Research-Solid Earth 87, 123 1-1 238. Paulick, H. and Franz, G. 1997. The color of pumice: case study on a trachytic fall deposit, Meidob volcanic field, Sudan. Bulletin of Volcanology 59, 171- 185. Platz, T. 2001. Mapping and characterisation of the vol caniclastic Maero formation deposits on the northwestern sector of Egmont volcano (Mt. Taranaki), New Zealand. Unpublished Diploma mapping thesis. Ernst-Moritz-Arndt Universitä t, Greifswald, Bundesrepublik Deutschland. Chapter 6 Transition to Explosive Eruptions 211 Polacci, M., Papale, P. and Rosi, M. 2001. Textural heterogeneities in pumices from the climactic eruption of Mount Pinatubo, 15 June 1991, and implications for magma ascent dynamics. Bulletin of Volcanology 63, 83-97. Polacci, M., Rosi, M., Landi, P., Di Muro, A. and Papale, P. 2005. Novel interpretation for shift between eruptive styles in some volcanoes. Eos, Transactions, American Geophysical Union 86, 333- 3 36. Pyle, D.M. 1989. The thickness, volume and grainsize of tephra fall deposits. Bulletin of Volcanology 51, 1-15. Richet, P., Whittington, A., Holtz, F., Behrens, H., Ohlhorst, S. and Wilke, M. 2000. Water and the density of silicate glasses. Contributions to Mineralogy and Petrology 138, 337-3 47. Rosi, M., Landi, P., Polacci, M., Di Muro, A. and Zandomeneghi, D. 2004. Role of conduit shear on ascent of the crystal-rich magm a feeding the 800-year-B.P. Plin ian eruption of Quilotoa Volcano (Ecuador). Bulletin of Volcanology 66, 307-3 21. Rust, A.C. and Cashman, K.V. 2004. Permeability of vesicular silicic magma: inertial and hysteresis effects. Earth and Planetary Science Letters 228, 93-1 07. Rust, A.C., Manga, M. and Cashman, K.V. 2003. Determining flow type, shear rate and shear stress in magmas from bubble shapes and orientations. Journal of Volcanology and Geothermal Research 122, 111- 132. Rutherford, M.J. and Devine, J.D. 2003. Magmatic conditions and magma ascent as indicated by hornblende phase equilibria and reactions in the 1995-20 02 Soufrière Hills magma. Journal of Petrology 44, 1433 -14 53. Saar, M.O. and Manga, M. 1999. Permeability-porosity relationship in vesicular basalts. Geophysical Research Letters 26, 111 -1 14. Sato, H., Nakada, S., Fujii, T., Nakamura, M. and Suzuki-Kamata, K. 1999. Groundmass pargasite in the 1991-1 995 dacite of Unzen volcano: phase stability experiments and volcanological implications. Journal of Volcanology and Geothermal Research 89, 197- 212. Schlinger, C.M., Rosenbaum, J.G. and Veblen, D.R. 1988. Fe-oxide microcrystals in welded tuff from southern Nevada; origin of remanence carriers by precipitation in volcanic glass. Geology 16, 556- 5 59. SEAN 05:06. Smithsonian Institution, 1980. Mount St. Helens. S cientific Event Alert Network vol. 5, no. 6. SEAN 05:07. Smithsonian Institution, 1980. Mount St. Helens. S cientific Event Alert Network vol. 5, no. 7. SEAN 07:03. Smithsonian Institution, 1982. El Chichon. Scientific Event Alert Network vol. 7, no. 3. Chapter 6 Transition to Explosive Eruptions 212 Sherburn, S. and White, R.S. 2005. Crustal seismicity in Taranaki, New Zealand using accurate hypocentres from a dense network. Geophysical Journal International 162, 494-5 06. Sparks, R.S.J. 1997. Causes and consequences of pressurisation in lava dome eruptions. Earth and Planetary Science Letters 150, 177-1 89. Sparks, R.S.J., Murphy, M.D., Lejeune, A.M., Watts, R.B., Barclay, J. and Young, S.R. 2000. Control on the emplacement of the andesite lava dome of the Soufriere Hills volcano, Montserrat by degassing-induced crystallization. Terra Nova 12, 14-2 0. Stasiuk, M.V., Barclay, J., Carroll, M.R., Jaupart, C., Ratté, J.C., Sparks, R.S.J. and Tait, S.R. 1996. Degassing during magma ascent in the Mule Creek vent (USA). Bulletin of Volcanology 58, 117- 130. Steen-McIntyre, V. 1975. Hydration and superhydration of tephra glass: a potential tool for estimating age of Holocene and Pleistocene ash beds. In: Suggate, R.P. and Cresswell, M.M. (eds.) Quaternary Studies. Royal Society of New Zealand Bulletin 13, 271-2 78. Stevenson, R.J., Dingwell, D.B., Webb, S.L. and Sharp, T.G. 1996. Viscosity of microlite-bearing rhyolitic obsidians: an experimental study. Bulletin of Volcanology 58, 298-3 09. Stewart, R.B., Price, R.C. and Smith, I.E.M. 1996. Evolution of high-K arc magma, Egmont volcano, Taranaki, New Zealand: evidence from mineral chemistry. Journal of Volcanology and Geothermal Research 74, 275-2 95. Taylor, G.A.M. 1958. The 1951 eruption of Mount Lamington, Papua. Australian Bureau of Mineral Resources , Geology and Geophysics Bulletin 38. Thomas, N., Jaupart, C. and Vergniolle, S. 1994. On the vesicularity of pumice. Journal of Geophysical Research-Solid Earth 99, 1563 3-1 564 4. Topping, W.W. 1972. Burrell Lapilli eruptives, Mount Egmont, New Zealand. New Zealand Journal of Geology and Geophysics 15, 476-4 90. Tuffen, H., Dingwell, D.B. and Pinkerton, H. 2003. Repeated fracture a nd healing of silicic magma generate flow banding and earthquakes? Geology 31, 1089- 109 2. Wellman, H.W. 1962. Holocene of the North Island of New Zealand: a coastal reconnaissance. Transactions of the Royal Society of New Zealand Geology 1 , 29-99. Wright, H.M.N., Cashman, K.V., Rosi, M. and Cioni, R. 2007. Breadcrust bombs as indicators of Vulcanian eruption dynamics at Guagua Pichincha volcano, Ecuador. Bulletin of Volcanology 69, 281- 300. Chapter 6 Transition to Explosive Eruptions 213 6.8 Appendix 6.8.1 Appendix A: Correlation of Burrell Deposits, NW Sector Figure 6-13: Correlation of pyroclastic flow deposits associated with the Burrell episode. Note that the major unit from medial to distal represents the Burrell Breccia (A). In sec tion S04-133, the thin pyroclastic pumice flow deposits represent Burrell Breccia (B ) units 1-3. Exposures are sorted by stream and planimetric distance from source. See figure legend for further details (RAD – rock-avalanche deposit). Magnetism was measured by a portable fluxgate magnetometer. For clarity, older exposed units were omitted. Outcrop numbers and profiles refer to Pl atz (2001 ) except S04-1 33. For list of samples and coordinates of outcrops see Appendix B. Chapter 6 Transition to Explosive Eruptions 214 6.8.2 Appendix B: Location of Pyroclastic Flow Deposits Table 6-5. Location of pyroclastic flow deposits on the NW and S sector of Mt. Taranaki. Chapter 6 Transition to Explosive Eruptions 215 6.8.3 Appendix C: Volatile Contents in Glass – Preliminary Results and Conclusions 6.8.3.1 FTIR Spectroscopy The total H 2 O concentrations (measured from the 3550 cm -1 peak; see Chapter 2) of glass inclusions hosted in clinopyroxene cr ystals vary considerably from 0.74 to 6.16 wt.% with one outlier at 9.15 wt.%, and app ear to correlate linearly with molecular H 2 O at 1630 cm -1 on a positive slope (Fig. 6-14a). It is noted that one matrix glass was analysed with 0.28 wt.% H 2 O. Molecular CO 2 peaks were only present in some samples and calculated concentrations range between 19 and 82 ppm. If total H 2 O and CO 2 concentrations are plotted, th e data appear to correlate linearly although on a negative slope, i.e. smaller H 2 O concentrations seem to contain more CO 2 . Two data points off the linear trend are noted (Fig. 6-14b). Figure 6-14: Estimates of total water and carbon dioxide contents in melt inclusions. a) total H 2 O at 3550 cm-1 vs. molecular H 2 O at 1630 cm -1 , and b) molecular CO 2 at 2350 cm -1 vs. total H 2 O at 3550 cm -1 . The FTIR data of these melt inclusions must be considered cautiously, particularly with respect to the host mineral and the shape of melt inclusions. Clinopyroxene exhibits good cleavage due to its crystallographic habit. It has been frequently observed that melt inclusions can leak during ascent, especially in crystals with good to very good fissility such as plagioclase and clinopyroxene (Cashman and McConnell, 2005; Blundy et al., 2006). Due to rapidly changing pressure conditions during ascent, melt inclusions can partially re-equilibrate by loosing water whic h can be possibly detected from the shape of the melt inclusion (Fig. 6- 15). Melt inclusions in clin opyroxene normally exhibit irregular outlines and are positioned along cleavage planes, making them mostly unsuitable for analysis (Fig. 6-15). These sheet -like shapes make it difficult to intersect the inclusions from both sides and therefore to measure their thicknesses. The measurement of 9.15 wt.% H 2 O in one inclusion ap pears too high, although Chapter 6 Transition to Explosive Eruptions 216 Figure 6-15: Different shapes of melt inclusions in clinopyroxene. a) overview of crystal 2 (sample SD20); note the many inclusions of glass, plagioclas e, apatite and Fe-Ti oxide which are mostly oriented along crystallographic planes, b) two close-ups as ma rked in a); I - represents common but very small melt inclusions found in clinopyroxene, which are un suitable for FTIR analysis. Their shape is near spherical to ovate; II - the bottle-neck shape is typical for leaked melt inclusions, c) overview of crystal 11 (sample SD32) showing a large irregular shaped melt inclusion, d) close-up of c) showing the impossibility of using these inclusions for FTIR analysis; it can be assumed that the inclusion extends further into the crystals as indicated by the diffuse outline of the melt inclusions further to the right, e) section of crystal 4 (sample SD20) showing two types of melt inclusions, the reddish-brown coloured inclusions are probably altered in comparison to the brown inclusions to the right; note again the irregular outline of the inclusions, f) section of crystal 8 (s ample SD9D); abundant sheet-like inclusions probably oriented along crystallographic planes; the inclusions around the Fe-Ti oxide inclusions (black) appear to be connected. Scale bars are 100 µm in a), c), and e), 50 µm in d) and f), and 10 µm in b). Chapter 6 Transition to Explosive Eruptions 217 concentrations of 10 wt.% in andesitic me lt inclusions have been reported (Anderson, 1979; Grove et al., 2003). In a ddition, the apparent negative co rrelation of (very small) CO 2 and H 2 O concentrations is contradictory becau se carbon dioxide is about one order of magnitude less soluble than water relating to melt and volatile composition, and prevailing pressure and temperature conditions, i.e. CO 2 is also released from the melt phase earlier (at greater depth). If the trend is real, it must be assumed that clinopyroxene host crystals ta pped melt over a large pressure range, or alternatively, those crystals are derived from different magma batches. Despite the difficulties experienced in this study, leaky melt inclusions can provide important information about changes of physical properties of the magma during ascent and therefore eruption dynamics (Blundy et al., 2006). 6.8.3.2 Thermal Analysis The study of water concentrations in natural glasses using FTIR spectroscopy demonstrated that this technique can only be applied to glass inclusions in minerals and quenched volcanic glasses (i.e. obsidian). Unfortunately, it cannot be applied to vesicular pumice rocks due to their thin bubble walls. Earlier attempts were made to measure water contents by thermogravimetric analysis of quenched synthetic glass (McM illan et al., 1986) and obsidia n (DeGroat-Nelson et al., 2001), however, the methods used may have not have been optimised to provide precise results. For example, (1) glass chips with/w ithout inclusions were used, which modifies the glass weight fraction, and/or (2) step-heating occurred e ither at constant rates or incrementally in 4-5 steps which may lead to erroneous results, since volatile diffusion rates in glass chips are higher than in glass powders. In this study, powderised pure glass samples were used for thermal analysis since it enabled uniform heating across the sample as well as quick volatile diffusion th rough the glass. An optimal methodology of heating the sample was developed through iteration (Fig. 6-16a). Preliminary results of this methodology are illustrated in Figure 6-16b. The low temperature experiments are most likely inco mplete, because further weight loss is indicated by other runs (Fig. 6-16a). It is noted that some samples give a very good reproducibility while other samples display major differences in weight loss, mainly Chapter 6 Transition to Explosive Eruptions 218 caused by the difference in applied maximum temperatures (Fig. 6-16b). Nonetheless, these preliminary results encourage further pursuit of this method. Hydration of glass is an inter-di ffusion process where hydronium ions (H 3 O + ) are substituted for mobile alkali ions, mainly Na + and K + (Laursen and Lanford, 1978). The process already starts once volcanic glass is exposed to the atmosphere and water is adsorpted on its surface. In earlier st udies the rate of obsidian hydration was investigated, for example, to indirectly da te obsidian artefacts based on the thickness of the hydration rind (S teen-McIntyre, 1975. Friedman and Long, 1976; Laursen and Lanford, 1978). Despite the large surface areas of pumice exposed to hydration compared to obsidian, no such effects (e .g. discolouration) were observed, and this method seems applicable, at least to historically erupted pumice. Figure 6-16: Preliminary results of the thermal analysis studies. a) trial and error series of sample SD20, b) reproducibility results of different pumice samples; it is noted that for the same sample the maximum weight loss is often observed at similar temperatures. One sample (SD35) was used to quantify the weight change and a thermal analysis (to a temperature of 950 °C) c oupled with infrared H 2 O and CO 2 detectors was carried out (by H.B. Lindgren, Geological Survey of Denmark and Greenla nd, Copenhagen). The resulting DTA-EWA curve (Differential Th ermal Analysis-Evolved Water Analysis) shows two prominent water release peaks at 355 °C and 635 °C, and a weak peak at 460 °C. However, if a larger amount of sample were used, the peaks would have been more intense (H.B. Lindgren, pers. comm., 2006). The analysis showed that only 6.5 ppm H 2 O was released. No peaks on the CO 2 curve were observed. The measured 6.5 ppm H 2 O of this sample is in gross contrast to the average weight change of 1.47 wt.% obtained by DSC-TGA (Fig. 6-16b). In addition, prominent water release peak at 635 °C only roughly coincides with the maximu m weight changes at 657 °C, 685 °C, and 687 °C as observed on the DSC-TGA curves (Fig. 6-16b). This implies that most water bound in volcanic glass may be released at temperatures well below 900 °C. Further Chapter 6 Transition to Explosive Eruptions 219 studies need to be pursued to confirm these preliminary results and examine whether loss of other volatile phases (such as F and SO 4 ) could explain the enormous weight change observed in erupted pumice clasts. Chapter 7 Reconstruction of Eruption Mechanisms 221 Chapter 7 Reconstruction of Eruption Mechanisms Using Physico-chemical Data 7 Chapter 7 Reconstruct Petrographic and geochemical data can be used for calculation of melt and magma conditions. In this chapter, physico-chemical data are interpreted to reconstruct eruption mechanisms during the Maero Erupt ive Period. Between eruption episodes and events significant differences in erupti ve style and processes occurred; through this physico-chemical dataset some of the contro lling influence on thes e will be discussed. 7.1 Introduction Reconstruction of eruption mechanisms is difficult if no observational datasets are available. Understanding the eruptive styles of a dormant volcano, deposits of historic eruptions need to be studied. In most cases , these studies are limited to petrographic and geochemical analyses. The datasets collected for the Maero Eruptive Period comprise petrography, glass and bulk rock chemistry, wh ich are used to derive conclusions about the most important physical property of erupted magma: its melt a nd magma viscosities. It is then debated what propos itions can be derived about each eruptive episode of the Maero Eruptive Period in respect of eruption style and mechanisms. Chapter 7 Reconstruction of Eruption Mechanisms 222 7.1.1 Viscosity of Magmas Understanding the viscosity of silicate melts and the knowledge of its governing factors are particularly important for igneous processes including melt and crystal segregation, convection in magma chambers, melt migration in source regions, and magma ascent and fragmentation. Processes occurring in sub-surface regions are particularly important where melt and magma changes in the conduit control the dynamics and style of eruptions (i.e. magma flow dynamics and fragmentation). Since the seminal works of Bottinga a nd Weill (1972) and Sh aw (1972) numerous studies were carried out on the physical properties of a variety of anhydrous and hydrous silicate melts: rhyolite a nd granite (Neuville et al., 1993; Dingwell et al., 1996), andesite (Richet et al., 1996; Liebske et al., 2003), basalt (Toplis et al., 1994; Giordano and Dingwell, 2003; Sato, 2005), also includi ng phonolite, trachyte, ba sanite, peridotite, and carbonatite (Dobson et al., 1996, Whitti ngton et al., 2001; Romano et al., 2003; Dingwell et al., 2004; Giordano et al., 2004a; Liebske et al., 2005; Misiti et al., 2006). There are three main factors controlling the viscosity of silicate melts: 1) melt structure and chemical composition, 2) temperature, and 3) volatile content. Most silicate melts are composed of tetrahedrally coordinated Si 4+ and Al3+ and minor Fe 3+ cations strongly bonded to four non-bridging oxygen anions, al ong with octahedrally coordinated Ca2+ , Mg 2+ and Fe 2+ cations weakly bonded to six bri dging oxygen anions (Best, 2003). Interlinked polymers of mainly (SiO 4 ) 4 - , (Al 3+ O 4 ) 5 - , and subordinate (Fe 3+ O 4 ) 5 - tetrahedra form a three-dimensional melt stru cture or network. The higher the degree of polymerisation, i.e. the longer the string-like polymers with interspersed octahedra, the higher is the melt viscosity. Thus, the ch emical melt components can generally be subdivided into 1) network-forming cations (Si 4+ , Al 3+ , Fe 3+ ), 2) bridging oxygens, 3) network-modifying cations (monovalent [Na + , K + ] and divalent [Ca 2+ , Mg 2+ ] cations including more highly charged high-field-strengt h cations [e.g. Ti 4+ , P 5+ ] and low abundance trace elements [e.g. Mn 2+ , Ba 2+ , Sr 2+ ]; Best, 2003). It is noted that some mono- and divalent cations are needed to ba lance the charge of trivalent cations in tetrahedral coordination. Ferrous iron occurs exclusively as a ne twork-modifying cation whereas ferric iron is either a network-former or netw ork-modifier (Best, 2003). According to this, the oxida tion state of a melt can influence its degree of polymerisation, and hence its viscosity. Chapter 7 Reconstruction of Eruption Mechanisms 223 The temperature is the second principal factor controlling the degree of polymerisation and thus viscosity. At higher temperatures polymers of silicon and aluminium cations with non-bridging oxygens (i.e. Si=O and Al =O bonds) can be stretched or broken due to the increase of kinetic energy of atoms. Silicate melts at high temperature behave as Newtonian fluids, whereas at low temper atures, their behavi our is strongly non- Newtonian due to polymerisation and crysta llisation. Pressure, by contrast, does not have a significant direct impact on melt viscosities, at least not be low 2 GPa (Pinkerton and Stevenson, 1992; Baker, 1996). The effect of volatiles on viscosity of melts is still poorly known. Baker and Vaillancourt (1995), Dingwell et al. (1996), and Hess and Dingwell (1996) showed a strong non-linear increase in viscosity with decreasing water contents, particularly below 1 wt.%. Dissolved water occurs in silicate melts as OH - ions and molecular water (Stolper, 1982). Hydroxyl ions are able to break polymers in melts, and hence act as network-modifiers. Consequently, the effect of water on viscosity is more drastic in silica-rich melts (i.e. andesi tes-rhyolites) compared to si lica-poor melts (i.e. basalts). Fluorine (F - ) is probably the second most effectiv e volatile in reducing melt viscosities. In contrast to water, fluorine remains in so lution even at atmospheric pressure. Lavas of Mt. Nyiragongo (DR Congo), for example, co ntain large abundances of fluorine and phosphorus (Platz et al., 2004), which may have contributed to their extremely mobile nature reaching speeds of up to 60 kmh -1 (SEAN 02:03). Recent studies investigated the effect of volatiles such as fluorine, chlo rine, phosphorus, and boron on melt viscosities; fluorine, for example, depolymerises the melt by replacing one oxygen with two fluorines (Toplis and Dingwell, 1996; Gi ordano et al., 2004b; Zimova and Webb, 2007). Because the viscosity of hydrous and anhydrous natural melts is sensitive to chemical composition (cf. silica-factor) and temp erature, a range in viscosity (log 10 η in Pas) over 15 orders of magnitude is obs erved. Melt viscosities (log 10 η ) relevant to th is study (i.e. andesitic to rhyolitic melts) commonly range between 2-15 Pas for rhyolites in the temperature range of c.600-120 0 °C (c ontaining various proportions of H 2 O and F) and 1-13 Pas for anhydrous andesites in the temp erature range of c.650- 1650 °C (Neuville et al., 1993; Baker and Vaillantcou rt, 1995). Hydrous andesites ha ve lower viscosities of log10 η = 0.7-13 Pas at temperatures between c.700-1650 °C and H 2 O concentrations of 2 and 3.5 wt.% (Richet et al., 1996). Magmas are even more complex, since they commonly represent three-phase mixtures of melt, crystals and bubbles (i.e. liquid-solid-gas), hence their rheological behaviour is Chapter 7 Reconstruction of Eruption Mechanisms 224 strongly non-Newtonian. In this case the “v iscosity” of two-phase (liquid-solid) or three-phase magmas is termed apparent viscosity, ηa. The dependence of viscosity on crystal-content is important for modelli ng conduit and dome-growth processes (cf. Melnik and Sparks, 1999, 2002). For estimating apparent viscosities of crystal-bearing magmas relative to their equivalent melt viscosities, the Einstein-Roscoe equation (Einstein, 1906; Roscoe, 1952, Brinkma n, 1952) can be used (Marsh, 1981): 5.2)1( −−= νηη BXa [ Eq. 7-1 ] where η is the melt visocisty (Pas ), B=1.35 representative for uniformly sized spherical crystals and X ν is the crystal volume fraction suspended in the melt. Because magmas contain various crystal shapes and sizes, B is larger and approximated to 1.7. There is a non-linear positive relationship between crystal volume and apparent viscosity (Pinkerton and Stevenson, 1992; Lejeune a nd Richet, 1995) and the maximum crystal content a magma can bear before acting as a ‘solid’ is 70 vol.% (Lejeune and Richet, 1995). The viscosity of melts (and ma gmas) is also influenced by the presence of suspended bubbles. The rheology of bubble-bearing me lts varies depending on the bubble volume fraction and the Capillary number. The dimensionless Capillary number, Ca, is defined as the ratio of viscous stresses and surface tension stresses (Manga and Loewenberg, 2001): σ μηa Ca = [ E q. 7-2] where μ, η , a, and σ are shear rate, suspending fluid viscosity (i.e. melt viscosity), bubble radius, and surface tension, respectively. If Ca is low (Ca<0.1), bubbles resist deformation and preserve their near spherical shape; if Ca is high (Ca>0.1), bubbles deform easily and elongate (Manga and Loewe nberg, 2001). As a result, an increase in the shear rate will increase the Capillary number, and consequentially the viscosity of the bubble-melt mixture will decrease (Pal, 2003). Chapter 7 Reconstruction of Eruption Mechanisms 225 7.2 Methods and Approach To understand differences between individual dome-forming episodes, petrographic and bulk rock chemical studies were carried out. Since eruptive conditions at the onset of dome emplacement are unknown, estimates of melt and magma viscosities are based on glass chemistry of studied tephra and pyroclastic flow samples with estimated constant parameters such as melt water contents, empl acement temperature, and crystal content. Using this approach, indi vidual eruptive episodes can be compared and some conclusions be derived. 7.3 Results 7.3.1 Petrography 7.3.1.1 Pyroclastic Flow Deposits The mineral assemblage within the studied BAF rocks (Table 7-1) is identical for both phenocrysts and groundmass constituents, in cluding plagioclase, clinopyroxene, hornblende and Fe-Ti oxides. In add ition, biotite is a common but minor microphenocryst and groundmass phase. Or thopyroxene (<1%) is a groundmass constituent observed in all samples, however, it is difficult to detect where brown groundmass glass is present. Due to this similarity in mineral assemblage, three petrographic features were used to distinguish individual samples and highlight major textural features within the sample set: 1) hornblende texture, 2) degree of groundmass crystallinity, and 3) groundmass texture. Th e observations are summarised in Table 7-2. Hornblende crystals exhibit different de grees of decomposition and samples were subdivided into three categories showing 1a) fresh crystals with no reaction rim; 1b) distinct but relatively thin and continuous rim; and 1c) more advanced reaction with a thick reaction rim or entirely altered crystals (Table 7-2). The degree of groundmass crystallinity varies between individual samples and is differentiated into five subgroups based on microlite content and glass colour: 2a) nearly holocrystalline groundmass with occasional clear, tran slucent interstitial glass; 2b) semi- to holocrystalline groundmass with clear, translucent glass; 2c) semicrystalline to hyaline groundmass with clear, translucent gl ass; 2d) semi- to holocrystalline groundmass with brownish glass; 2e) semicrystalline to hyaline Chapter 7 Reconstruction of Eruption Mechanisms 226 Chapter 7 Reconstruction of Eruption Mechanisms 227 Chapter 7 Reconstruction of Eruption Mechanisms 228 groundmass with brownish glass. Semi- to holocrystalline glass has abundant microlites, predominantly plagioclase laths, with residual, interstitial glass and glass pockets still present. Semi-cry stalline to hyaline glass is used to highlight the abundance of glass in comparison to microlites. No nearly holocrystalline groundmass with remnants of brownish glass was observed (Fig. 7-1). The third category, groundmass texture, highl ights those samples that could represent hybrid rocks, based on the pr esence of two different groundmass glasses, or may imply other emplacement and growth mechanisms (Fig. 7-1). Table 7-2. Summary of petrographic studies of BAF samples. Chapter 7 Reconstruction of Eruption Mechanisms 229 Figure 7-1: Groundmass texture (a-d) and crystallinity (e-h) of clasts from pyroclastic flow deposits. a and c) two groundmass glasses, b-d) differences in degree of groundmass crystallisation in clear translucent and brown glasses, e) semi- to hyaline, clear translucent glass; note microvesicularity, f) same image as in e) under crossed polarised light, g) semi- to holocrystalline brown glass, h) same image as in g) under crossed polarised light. Chapter 7 Reconstruction of Eruption Mechanisms 230 7.3.1.2 Lava Flows The studied lava flows (see Table 3-4) contain phenocrysts of plagioclase and clinopyroxene with microphenocrysts and groundmass constituents of plagioclase, clinopyroxene, orthopyroxene, Fe-Ti oxides and tr aces of olivine. If hornblende crystals (<1%) are present either as phenocryst or microphenocryst they show thick decomposition reaction rims or are entirely decomposed. It is noted that cores of groundmass orthopyroxene crystals often represen t olivine. The Turtle is an exception since it contains hornblende (6-7%) with varying degrees of reaction as well as minor biotite crystals. In mineralogy, the Turtle is similar to MacKays Rock, a lava flow forming the SE crater wall (Cronin, 1991). 7.3.2 Bulk Rock Chemistry 7.3.2.1 Pyroclastic Flow Deposits The Maero eruptive sequence is unique in terms of bulk geochemistry within the erupted rock suite of Mt. Taranaki because they display the highest potassium contents known from the centre. Also within the succession of individual dome-forming events of this period, there is diversity in all major elements, especially SiO 2 , Al 2 O 3 , Fe 2 O 3 and MgO. Two eruptive events st and out: the remnant summit dome and the Turtle. The latter is a remnant coulée flow partially collapsed on its northern-southwestern side. Although it was speculated (Croni n, 1991) that the Turtle wa s part of the summit dome, this is not supported by differences in Sr, Mg#, and K 2 O between the two. The summit dome composition is further described in detail in Chapter 5. For the purpose of this sec tion, the Maero BAF composi tions are not distinguished individually but rather their compositional range within the Maero Eruptive Period are illustrated. Some of the BAF units are described in detail in subsequent sections. The dense, dome-derived clasts exhibit the highest silica and potassium contents of the Maero eruptive products although their SiO 2 contents are restricted between 56-60 wt.% (Fig. 7-2). Major element abundances of BA F rocks show positive correlation (e.g. for K 2 O) or negative correlation (CaO, FeO, MgO, Al 2 O 3 ) with increasing SiO 2 contents. In addition, considerable variation in trace element abundances (e.g., Sr, Zr, Ni) are observed within this group (Fig. 7-2). It is noted that one sample is distinct within this Chapter 7 Reconstruction of Eruption Mechanisms 231 Figure 7-2: Bulk rock composition of selected Maero eruptives. Block-and-Ash Flow deposits are not differentiated and the Pyramid Dome and the Turtle are omitted for clarity. For co mparison, selected lava flows of the upper main cone and Fanthams Peak are plotted. Mg#=100[ Mg 2+ /(Mg 2+ +Fe 2+ )]; all iron as Fe 2+ . group with a considerable lower Mg# of 43.1. For compar ison, one sample of a pre- Maero BAF deposit is shown th at is distinct in nearly all elemental abundances (Fig. 7-2). Three pumice samples collected from three BAF units showed nearly identical major and trace element compositions with silica values of 57.3-57.5 wt.%. These pumice clasts are 2 wt.% higher in silica, compared to the Burrell Lapilli fall deposits. It is noted that the collected samples were relatively small and also considering the porosity, the analysis may not represent their true bulk composition (e.g. the phenocryst assemblage may be underrepresented causing higher bulk silica conten ts due to higher proportions of high silica matrix glass). Chapter 7 Reconstruction of Eruption Mechanisms 232 7.3.2.2 Lava Flows Bulk rock compositions of lava flows app ear to arrange into distinct groups. The youngest lava flows (Summ it Group) show highest K 2 O contents. With increasing age, individual lava flow groups show progressively lower potassium contents and correlate positively with silica. Potassium is an incompatible element in lava flow rocks and its age-abundance ratio has already been recognised by Stewart et al. (1996) and Zernack et al. (2006). The scoria-and-ash flow deposit from Bobs Ridge (T04-56) cannot clearly be classed into either the Summ it or Staircase Group, although its chemical signature leans more toward the former. Likewise, two lava flows in the upper Pyramid Stream have chemical similarities to the Staircase Group. Chemical compositions of the Fanthams Group are also displayed to illustrate their differences to the central-vent Mt. Tara naki rocks. They are characterised by ≤53 wt.% SiO 2 . Interestingly, Fanthams Peak lava flows de scribe a similar inclined linear slope as main cone lava flow groups (Fig. 7-2). If Mg# is plotted against SiO 2 , it is observed that both main cone and Fanthams Peak Groups are clearly se parated illustrating a complex petrogenetic evolution of Mt. Taranaki rocks. Investigated lava flows show very simila r trace element abundances if compared, for example, to Maero-aged BAF deposits (Fig. 7-2). Only a slightly greater range in abundances (e.g., Sr and V) are noted which ma y be owed to the fact that studied lava flows cover a much greater time span and therefore, exhibit greater variations. 7.3.3 Physical Properties As described above, an understanding of er uption dynamics requires that the physical properties of the erupting melt (and magma) need to be estimated. Of these water content is probably the most influential constituent on the properties of silicate melts since it controls melt density and viscosity, and also the explosivity of a magma. Chapter 7 Reconstruction of Eruption Mechanisms 233 7.3.3.1 Water Estimates of Volcanic Glasses The determination of water contents in silicate glasses is limited to two precise methods, Karl-Fisher-Titration (cf. Scholz, 1984; Behrens, 1995) and Fourier Transform Infrared spectroscopy (FTIR; cf. Mackenzi e, 1988; Stuart. 2004). Other methods such as loss on ignition and the water-by-di fference (WBD) method are widely used, however, their resu lts are less precise. Since the WBD method is the difference from an electron microprobe analytical total and the ideal of 100% (Fig. 7-3), it also includes a summation of analytical errors and is highly dependent on the state of glass weathering, post-depositional hydration and th e quality of shard polish. In addition, Roman et al. (2006) demonstrated that water contents in glass inclusions estimated by WBD are approximately 1 wt.% higher than when measured by Fourier Transform Infrared spectroscopy. Figure 7-3: Analytical totals of all EMP glass analyses (a) and sample averages (b) are plotted agai nst silica content. Estimated glass water contents using the water-by-difference method (WBD) are shown on the right axis. The terms andesite, dacite, and rhyolite refer to the TAS-classification scheme of Le Maitre et al. (1989). Chapter 7 Reconstruction of Eruption Mechanisms 234 Although the WBD method appears to overestimate water contents in silicate glasses, it represents a simple method to approximate wate r contents in glass shards at least to the first order. The typical high microlite content and small grain size of andesitic glass makes it difficult to apply FT IR or other techniques such as Ion Microprobe with confidence. The totals of uncontaminated glass analyses are highly variable and range between 92.5% and 100.5% w ith a mode at 98.0-98.5% (F ig. 7-3). The distribution curve is also skewed toward lower tota ls (left skewed by -1.3). Using only the difference to 100%, the mean value of 97.8% of all analyses (n= 875) would correspond to an estimate of 2.2 wt.% H 2 O. It is also observed that the sample mean analytical totals roughly negatively correlate with their mean silica contents (Fig. 7-3). 7.3.3.2 Viscosity Melt viscosities are calculat ed after the models of Sh aw (1972) and Hui and Zhang (2007). In the former model, melt viscosity, η , is determined following the equation: 4.65.1 10 log303.2 4 10 −−= sTsη [Eq. 7-3] where T is the temperature in K. The parameter s is calculated as )1/()( 22 SiOSiOii XXSXs −=∑ o [Eq. 7-4] where X i represents the mole fractions of oxides and °iS is the partial molar activation energy of oxides (Shaw, 1972). The model of Hui and Zhang (2007) uses the following formula: )( log T D C e T B A +++=η [ Eq. 7-5] where T is the temperature in K, e is Eu ler number, and A, B, C, and D are linear functions of oxide mole fractions (except for H 2 O). The parameters A-D are calculated as follows: 222 3222 ),(),( 43.826.15938.14031.34 76.1901.1871.1479.17038.6 AlOKNaOHOKNa CaOMgOOAlTiOSiO XXZX XXXXXA ex −+−+ −−−−−= [ Eq. 7-6] 222 3222 ),(),( 12.1655.4884.3829.68 64.2296.2561.3293.24814.18 AlOKNaOHOKNa CaOMgOOAlTiOSiO XXZX XXXXXB ex ex +−+− ++++= [ Eq. 7-7] Chapter 7 Reconstruction of Eruption Mechanisms 235 22 232 ),( ),(),( 16.322.43201.332 67.8592.6953.10598.6173.21 AlOKNaOH OKNaCaOMgOOMnFeOAl XXZ XXXXXC exex −−+ −−−−= [ E q. 7-8] OHOPOKNaCaO MgOOMnFeOAlTiOSiO XZXXX XXXXXD ex ex 2522 3222 75.51397.40477.38401.5812.67 83.11056.3810.2205.14316.2 ),( ),( +−+++ ++−−= [ Eq. 7-9 ] with Z being: )]/797.185(1/[1)( 2 T OHXZ += . [ Eq. 7-10 ] Al2 O 3ex and (Na,K) 2 O ex are excess oxides after forming (Na,K)AlO 2 . The 2 σ deviation is 0.61 log 10 η units (Hui and Zhang, 2007). For both models the melt composition and its temperature are required as variables. Lava dome temperatures at other volcanoes were measured at >900 °C (Santiaguito, Sahetapy-Engel et al., 2004), c.850 °C (Unzen volcano, W ooster et al., 2000), 720 °C (2004 Mt. St. Helens lava dome, Driedger et al., 2004), and 800 °C (1999 lava dome of Soufrière Hills Volcano, Taron et al., 2007). Since the majority of studied Taranaki glasses are derived from dome-forming eruptions a temperature of 900 °C is assumed. This estimate is based on applying the amphibole-plagioclase thermometer of Holland and Blundy (1994) to the rock suite (see section 5.5.3.1) . Thus, using a consistent temperature for all melt viscosity calculations, comparison can be made between dome- forming and sub-Plinian erup tions. A similar approach was undertaken regarding the melt water content. Melt viscosities were determined ‘anhydrous’ (i.e. 0.1 wt.% H 2 O) and using 1 wt.% H 2 O. Glass water contents as es timated by WBD were also used, although the resulting melt viscosities are lower because WBD often overestimates the ‘true’ glass water content (see above). Fo r calculation purposes only the mean sample compositions of studied glass analyses were used resulting in mean sample melt viscosities. Calculated melt viscosities are plotted against silica contents (Fig. 7-4). In both models it is observed that with increasing silica contents, melt viscosities increase linearly with fixed water contents (i.e. 0.1 and 1 wt.% H 2 O). Using estimated water contents (i.e. by WBD), the Shaw-model also shows increases in melt viscosities with increasing SiO 2 content, however, on a shallower slope, whereas melt viscosities decrease with increasing SiO 2 abundances in the Hui- and-Zhang model. In th e silica range 58.2-73.2 wt.% melt viscosities (log 10 η ) of Shaw (1972) range betw een 5.1-8.0 Pas (anhydrous), 4.5-7.0 Pas (1 wt.% H 2 O), and 4.5-6.0 Pas (WBD). Melt viscosities (log 10 η ) calculated Chapter 7 Reconstruction of Eruption Mechanisms 236 after Hui and Zhang (2007) are consistently higher in the same silica range varying between 7.3-9.0 Pas (anhydrous), 5.7-7.0 Pas (1 wt.% H 2 O), and 5.6-4.7 Pas (WBD). It is noted that one outlier at 71.85 wt.% SiO 2 has elevated melt viscosities in the Hui and Zhang (2007) model which is mainly ascr ibed to a comparatively low potassium content. Figure 7-4: Calculated melt viscosities, η, are plotted against silica abundances. Values of the models of Shaw (1972 ) and Hui and Zhang (2007 ) are shown for H 2 O contents of 0.1 wt.%, 1 wt.% and WBD at T=900 °C and P=1 bar. Solid and dashed lines are regression lines of η at WBD for the Hui-and-Zhang- and Shaw-models, respectively. If the apparent viscosities of magmas ( ηa) are to be estimated, the volume fraction of crystals and vesicles must be known. Rock s of the present summit dome exhibit an average crystallinity (excluding microlites) of 42 vol.% with lower and upper limits of 36 and 49 vol.%. Other studied rocks (mai nly derived from BAF deposits) show crystallinities between 27 and 42 vol.%. Th e microlite content of rocks has not been determined because microlite nucleation and growth continues syn- and post-eruptively which makes it difficult to estimate its proportion at the time of eruption or just prior to it. Higgins (1996) measured lava flow crys tallinities (only crys tals larger than 150 μm) of 9.8-30.2 vol.% with an estimated av erage crystallinity of about 20 vol.%. Decompression-induced microlite crystallis ation is an important process that significantly impacts on magma viscosities (Sparks, 1997). However, petrographic studies showed that rocks derived from BAF deposits can be semi-crystalline to hyaline (see Fig. 7-1). Hence, the microlite content can be very low, at l east in some individual Chapter 7 Reconstruction of Eruption Mechanisms 237 eruptive deposits. On this basis, end-member magma viscosities (log 10 ηa) were calculated with total crystal volume fractions of 30% and 55%. The Shaw-model predicts upper an d lower apparent viscosity [log 10 ηa in Pas] at 58.2 wt.% SiO 2 of 5.9-8.1 (anhydrous), 5.3-7.5 (1 wt.% H 2 O), and 5.3-7.4 (WBD) and at 73.2 wt.% SiO 2 of 8.8-11.0 (anhydrous), 7.8-9.9 (1 wt.% H 2 O), and 6.8-9.0 (WBD; Fig. 7-5). In contrast, upper and lower apparent viscosities of the Hui and Zhang model using the same crystal volume fractions of 30 and 55%, respectively, result at 58.2 wt.% SiO 2 in log10 ηa of 8.1-10.3 (anhydrous), 6.5-8.7 (1 wt.% H 2 O), and 6.4-8.6 (WBD), and at 73.2 wt.% SiO 2 in log10 ηa of 9.8-12.0 (anhydrous), 7.8-9.9 (1 wt.% H 2 O), and 6.6-8.7 (WBD; Fig. 7-5). Figure 7-5: Calculated magma viscosities, ηa, are plotted against SiO 2 contents. Lower and upper crystal volume fractions of 30% (a) and 55% (b), respectively are used for the calculation based on the calculated melt viscosities (see Fig. 7-4). Viscosities are calculated using H 2 O contents of 0.1 wt.%, 1 wt.% and WBD at constant T=9 00 °C and P=1 bar. Solid and dashed lines are regression lines of ηa at WBD for the Hui-and-Zhang- and Shaw- models, respectively. Chapter 7 Reconstruction of Eruption Mechanisms 238 7.4 Discussion 7.4.1 Comparison of Physico-chemical Properties 7.4.1.1 Bulk Rock Composition Rocks erupted during the Maero Eruptive Pe riod show bulk compositional variations, especially in SiO 2 and K 2 O. Pumiceous clasts of the sub- Plinian Burrell Lapilli eruption exhibit lowest silica abundan ces in a tight range of 55- 55.5 wt.% whereas the pre- climactic dome shows a broader compositional range (55.9-59.5 wt.% SiO 2 ). Products of other dome-forming eruptions also exhibit similar compositional variations, however, not as large as, for example, the Pyrami d and Tahurangi eruption episodes. Silica contents and Mg# are plotted for individual erupt ions (Fig. 7-6). It is noted that no bulk rock compositions are available for the erup tive events of Hooker Lapilli, Waingongoro Ash, and Mangahume Lapilli because identifi ed fall deposits did not contain sufficient amounts of sample material. In addition, onl y 1-2 analyses are av ailable for Puniho Ash and Te Popo Ash, which cannot reveal any compositional variations. No systematic compositional variation in time of Maer o-aged products is observed. The only noteworthy observation is that rocks of Newall Ash show homogeneous Mg# (46.2- 46.8) despite variations in SiO 2 (55.7-59.1 wt.%). This may be indicative of a homogenous source region and may exclude mixing/mingling processes during magma ascent. Figure 7-6: Bulk SiO 2 contents (a) and Mg# (b) are plotted against K 2 O in chronological appearance of eruption episodes. Although no systematic geochemical variations are apparent in this dataset, some remarks are made as to why the last erup ted products of the Pyramid eruption do not show highest silica and potassium abundances as inferred from the overall geochemical behaviour of the Mt. Taranaki rock suite. It has been recognised that the major Chapter 7 Reconstruction of Eruption Mechanisms 239 processes at mid- to shallow crustal magma storage levels involve fractional crystallisation and mixing and mingling of compositionally distinct or similar magmas. Influx of primitive or less evolved magma into a source region containing a more evolved magma often causes eruption (Huppert and Sparks, 1980; Druitt et al., 1999). Time is required for the replenished primitive magma at mid- to upper crustal levels to differentiate to a more evolved magma. In the case of a long period of repose, the magma is able to differentiate at depth, how ever, in the case of two eruptions separated by a short period of repose, the second er uption may produce less evolved magma due to a lack of time for differentiation. In ot her words, the absence of eruptions allows magma differentiation in the crustal reservoir. Cyclic variations of low and high eruptive frequencies are observed at Merapi volcano, Indonesia (Gertisser and Keller, 2003). Each magmatic cycle spans several hundreds of years and is characterised by the initial eruption of more evolved magmas (low eruption frequency) which then gra dually decline to less evolved magmas due to a higher eruption frequency (Gertisser and Keller, 200 3). Although the data of this study are insufficient to support or disprove such cyclic magmatic variation at Mt. Taranaki, it is noted that the latest two eruptions of Taranaki show d eclining silica abundances (Fig. 7-6). 7.4.1.2 Melt and Magma Viscosities Viscosities of natural silicate melts have b een frequently estimated using the model of Shaw (1972). However, this model extremel y underestimates predicted viscosities at very low H 2 O (<1 wt.%) contents and slightly underestimates viscosities of melt with high H 2 O (>3.5 wt.%) contents (D ingwell et al., 1996; Hess and Dingwell, 1996). Thus, the model of Hui and Zhang (2007) is preferre d for the following reasons. First, it is applicable to all natural hydrous and anhydrous silicate melts. Second, the model has a high accuracy with a 2 σ deviation of 0.61 log 10 η units. Third, the model should not be used for viscosities above 10 5 Pas, temperatures below 573 K, and for non-rhyolitic melts with >5 wt.% water which makes this model suitable for volcanological applications, and covers the entire range of prevailing conditions at andesite volcanoes. Fourth, it uses all major element oxides including trace abundances of MnO and P 2 O 5 . Fifth, iron is only used in the model as ferrous iron which makes an estimate of the iron Chapter 7 Reconstruction of Eruption Mechanisms 240 oxidation state redundant. Althoug h it is known that the ir on oxidation state has an impact on melt viscosity, especially at ve ry low temperatures (Dingwell and Virgo, 1987), there is currently no information of how to estimate the ferric/ferrous ratio of melts, especially if analyses are derived by EMP. The only potential disadvantage of the Hui-and-Zhang model are the large amount of parameters involved. Published matrix glass compositions from Merapi volcano, Indonesia (Bardintzeff, 1984; Berthommier, 1990; Ha mmer et al., 2000) and Soufrière Hills Volcano, Montserrat (Harford et al., 2003) were used to calculate melt viscosities using the same parameters (T=900 °C, 0.1 and 1 wt.% H 2 O) in order to compare to calculated Mt. Taranaki melt viscosities. Here only the m odel of Hui and Zhang (2007) is used (see above). This shows that Mt. Taranaki melt viscosities are intermediate between the slightly lower melt viscosities (at similar SiO 2 abundances) of Merapi volcano and the higher melt viscosities of Soufrière Hills Volc ano (Fig. 7-7). It is pointed out that the majority of Merapi dome clasts have glass silica contents above 71 wt.% and therefore exhibit similar melt viscosities as the majority of Taranaki glasses at lower SiO 2 contents. In contrast, Soufrière Hills glas s silica abundances are exclusively above 76 wt.% and thus displaying highest melt visc osities of approx. 1 log unit above Mt. Taranaki melt viscosities. If magma viscosities, ηa, (T=900 °C, crystal fractions of 30% and 55%, 1 wt.% H 2 O) between the three andesite volcanoes are compared then similar observations are made in comparison to melt viscosities. However, eruptive conditions (i.e. emplacement temperature, crystal and mi crolite content, residual water in matrix glass) differ from volcano to volcano. Viscosit ies of natural lava (either as flow or dome) are difficult to estimate and reliab le viscosity measurements under natural conditions are rare. Fig. 7-8 illustrates lava viscosities of selected andesite, dacite, and rhyolite centres. Calculated Mt. Taranaki magma viscositi es of the Maero Eruptive Period are based on matrix glass compositions. In order to compare Taranaki lavas to other centres, the range in magma viscosity with crystal fractions of 30% and 55% is used and plotted against the range of bulk SiO 2 abundances of Maero rocks. A linear relationship between bulk silica content and lava viscosity as observed for calculated melt and magma viscosities is assumed. In this context, studied Mt. Taranaki rocks appear to have lowest lava viscosities with estimated maximum viscosities (i.e. at crystal fractions of 55%) of log 10 ηa=9.2-9.8 Pas (T=900 °C; 1 wt.% H 2 O). It is noted that if melt water estimates are used from WBD then even lower lava viscosities result with a negative relationship between log10 ηa and SiO 2 . Only the 1979 Soufrière volcano Chapter 7 Reconstruction of Eruption Mechanisms 241 (St. Vincent) lava dome and lava flows fr om Colima volcano, Mexico, have similarly low viscosity. It is noted that Merapi lava dome rocks seem to possess lowest viscosities of log10 ηa=6-7 Pas (which would roughly co rrespond to the estimated minimum Taranaki dome viscosities at similar SiO 2 contents and 1 wt.% H 2 O) but these data were calculated on the basis of average extrus ion rate and T=1000 °C (Siswowidjoyo et al., 1995). On this basis, reported Merapi dome viscosities may be regarded as minimum values. Figure 7-7: Calculated Mt. Taranaki melt viscosities are compared to calculated melt viscosities of Merapi volcano (Indonesia), Soufrière volcano (St. Vincent), and Soufrière Hills Volcano (Montserrat), using the same parameters. Solid lines ar e regression lines for Taranaki data. The significance of calculated Mt. Taranaki melt and magma viscosities is that the magma erupted at any time during the Maer o Eruptive Period had lower viscosities compared to recent eruptions (i.e. Soufrièr e Hills Volcano, Montse rrat; Mt. St. Helens, USA). This means that emplacement and grow th of Taranaki lava domes is limited in height and total volume. Therefore, the form ation of spiny or Pelean domes within the Taranaki crater appears unlikely. Instead , lobate dome morphologies and coulée and block lava flows are expected (cf. Chapter 5). This conclusion is supported by petrographic rock textures (Table 7-2). The dominance of fresh and pristine hornblende crystals observed in rock s of pyroclastic flow deposits clearly shows that magma ascent and extrusion o ccurred at relatively fast rates. It also demonstrates that dome collapses occurred syn-eruptively since resulting pyroclastic Chapter 7 Reconstruction of Eruption Mechanisms 242 flow deposits cooled quickly to preserve th e observed pristine hornblende crystals. The observation of semi-crystalline groundmass text ures also indicate fast magma ascent and extrusion rates also implying low strain rates and low magma viscosities due to presence of ‘large’ quantities of melt. Do me geometries with heights of c.100 m may represent the upper limit before deforming. The presence of semi-crystalline groundmass textures also support syn-er uptive dome collapses (see above). Figure 7-8: Calculated Mt. Taranaki magma viscosities plotted against SiO 2 are compared to other andesite to rhyolite volcanoes. Since Taranaki viscosity calculations are based on glass chemical compositions of the Maero Eruptive Period, the range in bulk silica contents of rocks erupted during this period are used to allow comparison to published data. Upper and lower viscosity abundances are taken from Fig. 7-5. Taranaki data are illustrated by two parallelograms with upper and lower limits representing crystal volume fractions of 55% and 30%, respectively. The grey parallelogram corresponds to 1 wt.% melt water content, whereas the dashed parallelogram relates to water contents determined by WBD. Data source: silicic lava flows (Murase and McBirney, 1973 ; Fink, 1980 ; Navarro-Ochoa et al., 2002 ; Manley, 1996 ; Harris et al., 2004 ; and McKay et al., 1998 ); Mt. St. Helens (Murase et al., 1985 ; Scandone and Malone, 1985 ) ; Unzen volcano (Suto et al., 1993 ; Goto, 1999 ; Sato et al., 1999 ); Soufrière Hills Volcano, Montserrat (V oight et al., 1999 ; Sparks et al., 2000); Soufrière volcano, St. Vincent (Huppert et al., 1982 ) ; Merapi volcano (Siswowidjoyo et al., 1995). Chapter 7 Reconstruction of Eruption Mechanisms 243 7.4.2 Course and Eruption Styles of the Maero Eruptive Period 7.4.2.1 Types of Lava Domes The formation of either a dome shaped body or a lava flow is governed by prevailing magma viscosity and extrusion rates. By comparing magma viscosities of different volcanoes, inferences can be made about lava dome morphologies at Mt. Taranaki. Three basic dome types appear to have been formed during the Maero Eruptive Period: 1) composite lava domes, 2) coulée and block lava flows, and 3) homogenous lava domes. Composite lava domes (type 1) preferentially form within an intact summit crater where short-lived lava extrusion episodes of differe nt eruption periods amalgamate to a large dome complex or where a long-term extrus ion episode forms a heterogeneous dome by ongoing vent migration, alternating grow th episodes, and the extrusion of compositionally distinct magma domains. The best examples for this dome type are the 1980-1986 Mt. St. Helens and the 1995-present Soufrière Hills Volcano dome-forming eruptions. Type 2 lava domes comprise coulée and block lava flows. The formation of a block lava flow occurred during the most recent summit dome producing eruption (see Chapter 5). The Turtle was also interpreted as a coulée (Cronin, 1991). The columnar-jointed thick remnants on the upper NW flank exte nd down to 770 m over a 450 m altitude difference. Coulée and block lavas are generate d either once a lava lobe is spilled over the confining crater rim onto the upper flanks or , in the case of a breached crater, if they are extruded directly onto the upper flanks (cf. early stage of the summit dome formation). Recent block lava flows were observed, for example, at Santiaguito, Guatemala and Augustine volcano, USA). The third lava-dome type represents th e case where a simple homogenous dome is formed during a brief extrusion event. T ypically a hemispherical shape develops with/without individual lava lobes being emplaced on the upper flanks. An example for this type is the remnant summit dome of Mt. Taranaki. Chapter 7 Reconstruction of Eruption Mechanisms 244 7.4.2.2 Causes of Dome Collapse and Associated Deposits During recent and historic dome-forming erupt ions at andesite and dacite volcanoes, different dome collapse types are recognised. At Mt. Taranaki, three end-member types of dome collapse generated diverse properties and magnitudes of associated deposits: gravitational collapse of 1) parts of indivi dual lava lobes and 2) a major portion of the lava dome (sector dome collapse), and 3) dome disruption (or e xplosion) caused by over-pressurisation. Gravitational dome failure is also known as Merapi-type dome collapse. It is here distinguished between two end-members of a spectrum; the gravitational failures of a lava lobe/dome margin vs. failure of a majo r portion of the dome (sector collapse). An advancing lava lobe on a steep slope will behave depending on a combination of the slope and its tensile strength. Within th e lava body vertical velocity and hence deformation gradients occur with highest velocities and deformation rates at the top along with highest shear rates at the base near contact with substrate. Cracks develop preferentially at the surface due to cooling and are also induced by localised stress during flowage. Crack propaga tion is favoured by progressive deformation and hence flowage. Rainfall quenching is also another co ntributing factor to extend surface cracks causing further degradation of rock st rength (Elsworth et al., 2004). Once the gravitational force exceeds the tensile stre ngth of the lava, porti ons of the lobe can break off, forming either rock falls or BA Fs depending on the volume lost in each event. Similar processes are involved during a sector dome collapse. A dominant factor also involved in gravitational instabilities of lava domes is the increasing pressure load due to extrusion of large masses of lava. Incr easing weight acting on the basal dome portion causes progressively higher tensile stress. The delicate balance between downward- and upward-acting forces can be disrupted by volcano-tectonic ear thquakes triggering failure. Over-steepening and subsequent collap se of the dome front is also achieved by inflation of the dome interior or inflation of the upper edifice due to increasing magma supply rates. The collapse of an over-pressurised lava dome causing a lateral blast/explosion is known as Pelean-type dome collapse. Pressuri sation of the dome interior is caused by increasing fluid pressures within the melt fraction or increasing gas pore pressures, both caused by ongoing microlite growth when the outer part of the dome is sealed. Pressurisation within a dome results at high melt and magma viscosities where over- pressures in bubbles can be developed. High melt viscosities are the result of degassing Chapter 7 Reconstruction of Eruption Mechanisms 245 and groundmass crystallisation. A dramatic increase in melt viscosity occurs if dissolved water contents are reduced; a change from 1 wt.% to 0.1 wt.% H 2 O results in an increase in viscosity of more than 3 or ders of magnitude (Hess and Dingwell, 1996). Viscosity and bubble over-pressure s are apparently greatest in domes with minor water contents (Massol and Jaupert, 1998). Pore pressures increase with increasing viscosities, i.e. the more viscous the lava is, the more pressure builds up in pores or vesicles due to the increasing forces resisting pore/vesicle expansion. Large fluid (melt) pressures can develop by rapid nucleation, crystallisation and growth of anhydrous microlites (Sparks, 1997). If the lava is already highly viscous, exsolution of residual dissolv ed water is reduced or inhibited, causing excess fluid pressures up to several tens of MPa (Sparks, 1997). Pressure gradients within the dome and between dome and upper conduit result. If more pressure is added to the system, for example by intrusion of (volatile-rich) magma into the dome, the tensile strength of rocks can be exceeded. It is still to be investigated to what extent meteoric water vapour impacts on dome failure. It may be possible that rain water percolates through the porous crater rim rocks, heats up and vapour is trapped underneath the emplaced, weakly permeable dome cap. Resulting phreatic explosions by instant vapour release either weaken metastable domes, or themselves generate/trigger lateral blasts and major dome collapses. Large-scale failure of a pressurised dome, i nduced by either gravitational instability or excess internal overpressures, may cause late ral blasts or explosions. This is common when a dome is partially emplaced in a crater and partially onto the upper flanks. A sector collapse may expose th e pressurised dome interior, while much still-hot lava remains directly above the vent. Sudde n pressure reduction may induce rapid fragmentation by vesicle expansion and th is near-instantaneous exsolution and expansion of pressurised melt pockets trigge rs an explosion(s). Unroofing of a gas- pressurised inner dome or conduit may also generate ejection of ballistics, fountaining or magmatic fragmentation (e.g. Newhall an d Melson, 1983; Robert son, et al., 1998; Platz et al., 2007). The observe d single lapilli lithics scattered around the volcano were most likely generated by such mechanisms. Fragmentation of highly viscous, vesicular rocks by rapid decompression was investigated by Alidibirov and Dingwell (1996) who showed that a decompression wave propagates downward at sonic speed. Sub-parallel fractures perpendicular to the wave front are generated due to large pressure gradients. Chapter 7 Reconstruction of Eruption Mechanisms 246 A variety of deposits are observed at Mt. Tara naki, which are associated with the three end-member types of dome collapse. Syn-er uptive gravitational lava dome collapses (1 and 2) produce (hot) rock falls, BAFs an d pyroclastic surges. These deposits are primarily composed of magnetically aligned angular to subangular clasts. Other indicators for hot deposition are the presence of charred or charcoalised organic fragments (i.e. grass, twigs, or logs) or pervasive red-colour isation of matrix and clasts in the upper portion of the deposit [e.g. Tahur angi Breccia (a); cf. Chapter 5]. Block- and-Ash Flows can also be generated by post-eruptive gravitational dome collapses, however, if the dome has already cooled to ambient temperature then rock avalanches are triggered by gravitational failure (e.g. Pyramid Rock Avalanche; cf. Chapter 5). The disruption of a pressurised lava dome (type 3) can involve the generation of a directed blast, BAFs, pyroc lastic surges, and pumice fl ows depending on the magnitude of dome failure and whether the dome interior or the upper conduit is exposed. A blast deposit [Newall Breccia (a)] id entified by Cronin et al. (200 3) is confined to the W-NW sector of Mt. Taranaki (cf. Chapter 3 a nd Fig. 3-4i). The explosive removal of a pressurised lava dome occurred during the Burrell episode where a large BAF was initiated travelling more than 13 km from so urce (cf. Chapter 6). The generation of at least one pumice flow (exposed in Maero Stream) during the Puniho eruptive episode indirectly proves firstly, the failure of a pressurised dome triggering rapid magma vesiculation and fragmentation due to rapid decompression, and secondly, also the failure of a sustained explosive phase since no pumice fallout is recorded during Puniho times. The extent of Maero-aged BAF deposit va ries; only three BAF units are recorded outside the National Park boundary [Waiwera nui Breccia, Burrell Breccia (A), and Tahurangi Breccia (a)]. Since all BAFs produced during the M aero Eruptive Period were initiated by lava domes located in the summit crater, the initial potential energy can be considered as constant (except for mo re advanced lava lobes/coulées collapsed at significantly lower altitude). Consequentially, only two factors can significantly influence the runout of BAFs (assuming similar travel pa ths): the initial volume of collapsing material and added kinetic energy into the system. Kinetic energy could have been added at the beginning through a directed blast or explosive removal of the dome. The latter case is envisaged to have occurred during the Burrell episode (cf. Chapter 5). In summary, the common Merapi-type dome collapse as well as Pelean-type failure occurred during the Maero Eruptive Period. Chapter 7 Reconstruction of Eruption Mechanisms 247 7.4.2.3 Reconstruction of the Maero Eruptive Period The stratigraphy of the Maero Eruptive Period ha s been established in Chapter 3. It was shown that this period of activity is characterised by at least 10 eruptive episodes (see Fig. 3-6). All of these episodes are represented by fall deposits, however, exposed pyroclastic flow deposits on the NW sector are only recognise d for seven of the eruptive episodes. A complete reconstruction of the Maero Eruptive Period is difficult, because lava flows thought to have been emplaced during the period cannot be clearly correlated to any of the known episodes nor can they be distinguished as individual eruptive events. Despite this, conclusions can be dr awn about eruption styles and mechanisms based on field observations and measured and calculated physical rock and magma properties. Inferences about the course of each eruptive episode of the Maero Eruptive Period are discussed in chronological order. The Maero Eruptive Period starts with th e Hooker eruptive epis ode thought to have occurred at around AD 930 (Table 3-5). Only a discontinuous pumice and lithic lapilli fallout layer (Hooker Lapilli) is preserved on the eastern flank of Mt. Taranaki. This implies that only a low volume of magma erupted causing a spatially limited tephra layer. Although the main dispersal axis could have been towards the W-NW (contrasting to the main NE-S dispersal axis), no fall deposits were found on the NW sector. The production of pumice and dense lithic clasts during this event may suggest the explosive removal of a lava dome or plug. Due to the apparent limited volume of erupted pumice, vesiculation and fragmentati on of magma occurred probably only in the upper conduit by sudden decompression. Sustained fragmentation was probably prevented by limited magma supply rates. After a period of c.150 years, the Te Popo episode produced fall- (Te Popo Ash), BAF- (Te Popo Breccia) and multiple pyroclastic surge deposits dated at 878 ± 39 yrs B.P. (Cronin et al., 2003). During this episode, the NW flank of Mt. Taranaki was extensively inundated by pyroclastic surges and BAFs with their deposits directly overlying a thick paleosol of Oakura Tephra. The confinement of pyroclastic flow deposits to the NW sector has three implica tions: (1) either a low-point in the western crater rim enabled the direction of dome-related BAFs and surges towards the west, or (2) parts of the western rim collapsed, for example, caused by excess load pressure due to emplacement of large volumes of lava within the crater, or (3) dome growth exceeded Chapter 7 Reconstruction of Eruption Mechanisms 248 the dimensions of the crater area and individual lava lobes were emplaced onto the upper NW flanks and successively collapsed forming BAFs and surges. The partial rim collapse scenario appears to be the simplest explanation since about 30% of the clast population found in the BAF deposit (Te P opo Breccia) are orange -coloured clasts likely representing parts of the altered crater rim lava flows (cf. Fig. 3-8). Following an estimated 200-year-period of repose, the Waingongoro episode appears to have produced only a small fine ash layer on the eastern flanks; no pyroclastic flow deposits associated with this episode were found on the NW sector. The eruptive style producing this ash layer is unknown. Phreatic eruptions co mmonly produce thin, 2-3 cm thick, pale to dark grey ash layers of dens e lithic ash (i.e. shards of crystalline rock). This style can be excluded for this erup tion because the shards of the tephra were analysed and shown to be glassy, i.e. j uvenile magma was erupted. The presence of abundant microlites in these glass shards and their dense and angular shape are similar to glass shards erupted during dome-formi ng events. In addition, Waingongoro Ash has a very similar appearance to Te Popo Ash also indicating a similar predominant extrusive eruption style. The absence of pyr oclastic flow deposits on the NW sector may indicate that either the ex truded lava dome did not collapse or if it failed it only produced rock falls and minor BA Fs preserved at proximal areas. The Newall eruptive episode initiated a period of increased activity following an estimated period of repose of c.150 years. The episode started in c.AD 1430 and was initiated by a directed blast. This event and subsequent generated pyroclastic surges and BAFs devastated the NW sector as eviden t by abundant charcoalised organic fragments present in those deposits. The blast deposit [Newall Breccia (a)] is strongly confined to the W-NW area of Mt. Taranaki (Cronin et al., 2003). Medial and distal pyroclastic surge and BAF deposits are exposed, for exampl e, in Maero, Pyramid, and Waiweranui Streams. Observed fine to medium ash layers on the northern through to southern flanks along with exposed BAF [Newall Breccia (b)] and pyroclastic surge deposits on the W- NW sector clearly indicate a lava dome-f orming episode. The deposition of a single lithic lapilli layer [Newall Ash (a)] followed by two fine to medium ash layers [Newall Ash (b) and (c)] point towards three phases where the lapilli layer presumably represents a vent-clearing phase prior to lava extrusion. Th e two individual fall deposits are probably related to the lateral blast and later to ongoing dome growth and collapse events. The clear distinction of both ash layers may indicate that the Newall episode was a longer-term eruption. The evidence of a blast deposit implies failure of a Chapter 7 Reconstruction of Eruption Mechanisms 249 pressurised lava dome (type 3 dome collapse) . The exposure of the dome interior or un- roofing of the upper conduit caused instantaneous magma vesiculation and fragmentation. This is supported by the presence of up to 15% pumice clasts within BAF and surge deposits, often rounded and con centrated near the top of the deposits. Since no pumice fall deposit is observed, it is assumed th at only a small magma volume vesiculated and fragmented explosively. In addition, in parts significant portions of orange-coloured clasts inferred to be derived from crater-rim lava flows are also present in these deposits possibly indicating excavat ion and broadening of the breached western crater rim. Less than 100 years later, another dome-f orming eruption took place, the Waiweranui eruptive episode. Deposits associated with this episode comprise a fine to medium ash bed (Waiweranui Ash) on the eastern fla nks and a distal BAF deposit (Waiweranui Breccia) along with a pyroclastic surge de posit on the NW sector. Based on deposition characteristics, it appears that this erupti on is marked by ongoing lava extrusion forming a summit dome which subsequentially collapsed . The extent of the Waiweranui Breccia (>13 km from source) and the absence of significant amounts of pumice clasts within the deposit may be indicative of a simple gravitational sector dome collapse (i.e. type 2). The next eruptive episode occurred about 70 years later (Lees and Neall, 1993) in c.AD 1590. The Puniho episode is also recorded on the eastern flank as fine to medium ash (Puniho Ash) and partially as a surg e deposit. On the contrary, a BAF deposit (Puniho Breccia), a pumice flow- as well as pyroclastic surge deposits are exposed on the NW flank of Mt. Taranaki. A predominant dome-forming eruptive style is assumed for this period. The collapse of the Puniho dome triggered a short-lived, presumably sub-horizontal directed, e xplosive fragmentation (type 3 dome collapse), which generated minor pumice flow(s). The failure to cause sustainable fragmentation could be due to insufficient time for the magma below to vesiculate to greater depth and possibly low magma extrusion rates during this period of dome growth. The course and eruptive style of the Burrell episode of AD 1655 is described in detail in Chapter 6. The Burrell episode is characte rised by an initial lava dome extrusion followed by explosive removal of the dome triggering a large BAF. Un-roofing the vent caused a switch to an explosive, sub-Plin ian eruption style, wh ich itself can be distinguished into three phases each producing individual pumice flows. From a sub- Plinian eruption column widespread pumice lapilli fall was produced. Chapter 7 Reconstruction of Eruption Mechanisms 250 About 40 years later, the Ma ngahume eruption occurred, a presumably small eruption based on the observation of only a single lithic lapilli layer (Mangahume Lapilli) found on the southern flank. No correlatives were found on the NW sector. The presence of a scattered dense lithic layer alone only indicates an explosive event, perhaps a vent- clearing phase with the formation of ballistics (originating from a former dome/plug?). Therefore, due to the absence of ash particles associated with the lapilli layer other eruption styles such as Vulcan ian explosions can be excluded . However, in the light of results derived from the Pyramid eruption (cf. Chapter 5), it is also possible that a lava dome was erupted in a single (short-term?) event and did not collapse, i.e. no pyroclastic flow and fine ash fall deposits were produced during this eruption. The Tahurangi eruptive episode, tentatively dated at AD 1755 (Druce, 1966), appears to be one of the bigger dome-forming episodes of the Maero Eruptive Period judged by the thickness of the fall deposit. Tahurangi Ash is a fine to medium, up to 10 cm thick ash deposit exposed on the NE-E flanks and is even in pockets found beyond the National Park boundary. Block-and-Ash Flow deposits [Tahurangi Breccia (a) and (b)] on the NW sector were correlated with Tahurangi Ash; charred tussock [Tahurangi Breccia (a)] proved to be too young for radiocarbon age determination (i.e. <250 yrs B.P.). Tahurangi Breccia (a) is exposed at distal reaches near Hangatahua River more than 12 km from source. The exposure of two ma jor BAF deposits may be indicative for ongoing dome extrusion and simultaneous dome failure. Rock falls associated with dome growth are likely be only depos ited at proximal reaches. The homogenous appearance of Tahurangi Ash does not indicate any majo r breaks in time (i.e. by bedding or grain size variati ons) and may therefore point to an ongoing extrusive event of constant intensity. The last eruption of Mt. Ta ranaki, the Pyramid eruption, produced the summit dome whose remnants (Pyramid Dome ) are still preserved within the crater (cf. Chapter 5). The preservation of a scattered lithic lapilli layer (Pyramid Lapilli) on the southern flank shows that this eruption occurred post-Tahurangi since there is a 2-3 cm thick paleosol between Tahurangi Ash and Pyramid Lapill i. A post-eruptive (cold?) Pyramid Dome collapse produced a rock avalanche (Pyramid Rock Avalanche) on the NW sector which is exposed >5 km from source. The date of the Pyramid eruption is poorly constrained but is thought to have occurred between AD 1839 and AD 1866 (cf. Chapter 5). Chapter 7 Reconstruction of Eruption Mechanisms 251 7.5 Conclusions The Maero Eruptive Period is characterised by contrasting eruption styles and mechanisms. It appears that every eruptiv e episode (see Table 3-5) began with the extrusion of viscous lava forming domes. In some instances, an initial vent-clearing phase producing ballistics occurred as evident by the record of scattered single lithic layers at medial reaches [Newall Ash (a), Mangahume Lapilli, Pyramid Lapilli]. The type of dome failure (Merapi-type vs. Pe lean-type dome collapse) then controlled successive eruptive phases. In the case of a Merapi-type dome collapse, dome growth continued or magma extrusion ceased shortly after the event (cf. Tahurangi episode). In contrast, Pelean-type dome fa ilures can trigger a directed lateral blast-explosion (cf. Newall episode), instantaneous but short-lived magmatic fragmentation (cf. Newall and Puniho episodes), and sustained sub-Plinian eruptions (cf. Bu rrell episode). In addition, the emission of lava flows thought to have occurred during the Maer o Eruptive Period is yet another eruptive style where also compositionally distinct magmas were erupted. During the Maero Eruptive Period, arrival of viscous and degassed andesite magma in the upper conduit led to the formation of three types of lava domes described above. The dome type depended on melt and magma viscosity and the rate of extrusion where either lava lobes, block lava flows or lava domes are generated. Estimates of melt and magma viscosities at Mt. Taranaki have b een derived in this study, based on glass chemistry, predominantly derived from shards of fall deposits. Important factors controlling viscosity are H 2 O content and temperature; these parameters were estimated by glass WBD and amphibole-plagioclase ther mometry, respectively. Upper and lower crystal volume fractions were quantified unde r petrographic microscope. Magma ascent rates were also indirectly estimated by hornblende decomposition reactions; if the dimensions of the lava dome are known, extr usion rates can also be evaluated using the model of Lyman et al. (2004). For the late st Pyramid eruption, minimum magma ascent and extrusion rates of 0.012 ms -1 and 6.0 m 3 s-1 (5.2×10 5 m3 d-1 ), were estimated, respectively. Pre- or syn- eruptive magma viscosity are unknown, however, the first- order viscosity can be estimated using the above parameters. Erupted rocks initially forming a lava dome have assumed maximum syn-eruptive viscosities in the order of log10 η =9.2-9.8 Pas (T=900 °C, 1 wt.% H 2 O, 55 vol.% crystals). If water estimates based on WBD are applicable, much lower maximum viscosities between log 10 η =8.3-8.0 Pas could have occurred. Chapter 7 Reconstruction of Eruption Mechanisms 252 7.6 References Alidibirov, M. and Dingwell, D.B. 1996. Magma fragmentation by rapid decompression. Nature 380, 146- 148. Baker, D.R. 1996. Granitic melt viscosities: Empirical and configurational entropy models for their calculation. American Mineralogist 81, 126- 134. Baker, D.R. and Vaillancourt, J. 1995. The low viscosities of F+H 2 O-bearing granitic melts and implications for melt extraction and transport. Earth and Planetary Science Letters 132, 199- 211. Bardintzeff, J. 1984. Merapi volcano (Java Indonesia) and Merapi-type nuée ardente. Bulletin of Volcanology 47, 433-4 46. Behrens, H. 1995. Determination of water solubilities in high-viscosity melts: an experimental study on NaAlSi 3 O 8 and KAlSi 3 O 8 melts. European Journal of Mineralogy 7, 905 -9 20. Berthommier, P. 1990. Etude volcanologique du Merapi (Centre-Java): téphrostratigraphie et chronologie - produits éruptifs. Unpublished PhD thesis. Université Blaise Pascal, Clermont-Ferrand, France. Best, M.G. 2003. Igneous and metamorphic petrology. Blackwell Science. Malden. B GVN 29:09. Smithsonian Institution, 2004. Mount St. Helens. Bulletin of the Global Volcanism Network , vol. 29, no. 9. Bottinga, Y. and Weill, D.F. 1972. The viscosity of magmatic silicate liquids: a model for calculation. American Journal of Science 272, 438 -47 5. Brinkmann, H. 1952. The viscosity of concentr ated suspensions and solutions. The Journal of Chemical Physics 20, 571. Cronin, S.J. 1991. The geology of the summit of Mt. Egmont. Unpublished BSc Honours thesis. Massey University, Palmerston North, New Zealand. Cronin, S.J., Stewart, R.B., Neall, V.E., Platz, T. and Gaylord, D. 2003. The AD1040 to present Maero eruptive period of Egmont Volcan o, Taranaki, New Zealand. Abstract. Geological Society of New Zealand Miscellaneous Publication 116A, 43. Dingwell, D.B. and Virgo, D. 1987. The effect of oxidation state on the viscosity of melts in the system Na 2 O-FeO-Fe 2 O 3 - SiO 2 . Geochimica et Cosmochimica Acta 51, 195- 205. Dingwell, D.B., Romano, C. and Hess, K.-U. 1996. The effect of water on the viscosity of a haplogranitic melt under P-T-X conditions relevant to silicic volcanism. Contributions to Mineralogy and Petrology 124, 19-28. Dingwell, D.B., Courti al, P., Giordano, D. and Nichols, A.R.L. 2004. Viscosity of peridotite liquid. Earth and Planetary Science Letters 226, 127 -13 8. Chapter 7 Reconstruction of Eruption Mechanisms 253 Dobson, D.P., Jones, A.P., Rabe, R., Sekine, T., Kurita, K., Taniguchi, T., Kondo, T., Kato, T., Shimomura, O. and Urakawa, S. 1996. In-situ measurement of viscosity and density of carbonate melts at high pressure. Earth and Planetary Science Letters 143, 207- 215. Driedger, C.L., Schneider, D., Wessels, R. and Ramsey, M. 2004. Airborne thermal infrared observations during the first month of the 2004 eruption of Mount St. Helens volcano, Washington. Abstract. American Geophysical Union V31E- 05. Druce, A.P. 1966. Tree-ring dating of recent volcanic ash and lapilli, Mt Egmont. New Zealand Journal of Botany 4, 3-41. Druitt, T.H., Edwards, L., Mellors, R.M., Pyle, D.M., Sparks, R. S.J., Lanphere, M., Davies, M. and Barreiro, B. 1999. Santorini Volcano. Geological Society London Memoirs 19, 1-176. Einstein, A. 1906. Eine neue Bestimmung der Moleküldimension. Annalen der Physik 19, 289-3 06. Elsworth, D., Voight, B., Thompson, G. and Young, S.R. 2004. Thermal-hydrologic mechanism for rainfall-triggered collapse of lava domes. Geology 32, 969-9 72. Fink, J. 1980. Surface folding and vi scosity of rhyolite flows. Geology 8, 250-2 54. Gertisser, R. and Keller, J. 2003. Temporal variations in magma composition at Merapi Volcano (Central Java, Indonesia): magmatic cycles du ring the past 2000 years of explosive activity. Journal of Volcanology and Geothermal Research 123, 1-23. Giordano, D. and Dingwell, D.B. 2003. Viscosity of hydrous Etna basalt: implications for Plinian-style basaltic eruptions. Bulletin of Volcanology 65, 8-14. Giordano, D., Romano, C., Papale, P. and Dingwell, D.B. 2004a. The viscosity of trachytes, and comparison with basalts, phonolites, and rhyolites. Chemical Geology 213, 49-61. Giordano, D., Romano, C., Dingwell, D.B., Poe, B. and Behrens, H. 2004b. The combined effects of water and fluorine on the viscosity of silicic magmas. Geochimica et Cosmochimica Acta 68, 5159 - 5 168. Goto, A. 1999. A new model for volcanic earthquake at Unzen Volcano: Melt rupture model. Geophysical Research Letters 26, 2541- 254 4. Hammer, J.E., Cashman, K.V. and Voight, B. 2000. Magmatic processes revealed by textural and compositional trends in Merapi dome lavas. Journal of Volcanology and Geothermal Research 100, 165- 192. Harford, C.L., Sparks, R.S.J. and Fallick, A.E. 2003. Degassing at the Soufrière Hills Volcano, Montserrat, recorded in matrix glass compositions. Journal of Petrology 44, 1503- 152 3. Harris A.J.L., Flynn L.P, Matias O., Rose W.I. and Cornejo J. 2004. The evolution of an active silicic lava flow field: an ETM+ perspective. Journal of Volcanology and Geothermal Research 135, 147 - 1 68. Chapter 7 Reconstruction of Eruption Mechanisms 254 Hess, K.-U. and Dingwell, D.B. 1996. Viscosities of hydrous leucogranitic melts: a non-Arrhenian model. American Mineralogist 81, 1297 -13 00. Higgins, M.D. 1996. Crystal size distributions and other quantitative textural measurements in lavas and tuff from Egmont volcano (Mt. Taranaki), New Zealand. Bulletin of Volcanology 58, 194-2 04. Holland, T. and Blundy, J. 1994. Non-ideal interactions in cal cic amphiboles and their bearing on amphibole - plagioclase thermometry. Contributions to Mineralogy and Petrology 116, 433-4 47. Hui, H. and Zhang, Y. 2007. Toward a general viscosity equation for natural anhydrous and hydrous silicate melts. Geochimica et Cosmochimica Acta 71, 403-4 16. Huppert, H.E. and Sparks, R.S.J. 1980. The fluid dynamics of a basaltic magma chamber replenished by influx of hot, dense ultrabasic magma. Contributions to Mineralogy and Petrology 75, 279 -28 9. Huppert, H.E., Shepherd, J.B., Sigurdsson, H. and Sparks, R.S.J. 1982. On lava dome growth, with application to the 1979 lava extrusion of the Soufrière of St. Vincent. Journal of Volcanology and Geothermal Research 14, 199- 222. Le Maitre, R.W., Bateman, P., Dudek, A., Keller, J., Lameyre, J., Le Bas, M.J., Sabine, P.A., Schmid, R., Sorensen, H., Streck eisen, A., Woolley, A.R. and Zanettin, B. 1989. A classification of igneous rocks and glossary of terms. Blackwell Scientific. Oxford. Lees, C.M. and Neall, V.E. 1993. Vegetation response to volcanic eruptions on Egmont Volcano, New Zealand, during the last 1500 years. Journal of the Royal Society of New Zealand 23, 91-1 27. Lejeune, A.M. and Richet, P. 1995. Rheology of crystal-bearing silicate melts: an experimental study at high viscosities. Journal of Geophysical Research 100, 4215 -42 30. Liebske, C., Behrens, H., Holtz, F. and Lange, R.A. 2003. The influence of pressure and composition on the viscosity of andesitic melts. Geochimica et Cosmochimica Acta 67, 473-4 85. Liebske, C., Schmickler, B., Terasaki, H., Po e, B.T., Suzuki, A., Funakoshi, K., Ando, R. and Rubie, D.C. 2005. Viscosity of peridotite liquid up to 13 GPa: implications for magma ocean viscosities. Earth and Planetary Science Letters 240, 589 -60 4. Lyman, A.W., Koenig, E. and Fink, J.H. 2004. Predicting yield strengths and effusion rates of lava domes from morphology and underlying topography. Journal of Volcanology and Geothermal Research 129, 125- 138. Mackenzie, M.W. 1988. Advances in applied Fourier tran sform infrared spectroscopy. Wiley. New York. Manga, M. and Loewenberg, M. 2001. Viscosity of magmas containing highly deformable bubbles. Journal of Volcanology and Geothermal Research 105, 19-24. Chapter 7 Reconstruction of Eruption Mechanisms 255 Manley, C.R. 1996. Physical volcanology of a voluminous rhyolite lava flow: the Badlands lava, Owyhee Plateau, southwestern Idaho. Journal of Volcanology and Geothermal Research 71, 129 - 1 53. Marsh, B.D. 1981. On the crystallinity, probability of occurrence, and rheology of lava and magma. Contributions to Mineralogy and Petrology 78, 85-9 8. Massol, H. and Jaupart, C. 1998. Gas overpressures in lava dome eruptions. Abstract. American Geophysical Union V32C-07. McKay, M.E., Rowland, S.K., Mouginis-Mark, P.J. and Garbeil, H. 1998. Thick lava flows of Karisimbi Volcano, Rwanda: insights from SIR-C interferometric topography. Bulletin of Volcanology 60, 239-2 51. Melnik, O.E. and Sparks, R.S.J. 1999. Nonlinear dynamics of lava dome extrusion. Nature 402, 37-41. Melnik, O. and Sparks, R.S.J. 2002. Dynamics of magma ascent and lava extrusion at Soufrière Hills Volcano, Montserrat. In: Druitt, T.H. and Kokelaar, B.P. (eds.) The eruption of Soufrière Hills Volcano, Montserrat, from 1995 to 1999. Geological Society, London, Memoirs 21, 153 -1 71. Misiti, V., Freda, C., Taddeucci, J., Romano , C., Scarlato, P., Longo, A., Papale, P. and Poe, B.T. 2006. The effect of H 2 O on the viscosity of K-trachytic melts at magmatic temperatures. Chemical Geology 235, 124-1 37. Murase, T. and McBirney, A. 1973. Properties of some common igneous rocks and their melts at high temperatures. Geological Society of America Bulletin 84, 3563- 359 2. Murase, T., McBirney, A.R. and Melson, W.G. 1985. Viscosity of the dome of Mount St. Helens. Journal of Volcanology and Geothermal Research 24, 193- 204. Navarro-Ochoa, C., Gavilanes-Ruiz, J.C. and Cortes-Cortes, A. 2002. Movement and emplacement of lava flows at Volcan de Colima, Mexico: November 1998- February 1999. Journal of Volcanology and Geothermal Research 1 17 , 155-1 67. Neuville, D.R., Courtial, P., Dingwell, D.B. and Richet, P. 1993. Thermodynamic and rheological properties of rhyolite and andesite melts. Contributions to Mineralogy and Petrology 113, 572- 581. Newhall, C.G. and Melson, W.G. 1983. Explosive activity associated with growth of volcanic domes. Journal of Volcanology and Geothermal Research 17, 111- 131. Pal, R. 2003. Rheological behavior of bubble-bearing magmas. Earth and Planetary Science Letters 207, 165- 179. Pinkerton, H. and Stevenson, R.J. 1992. Methods of determining the rheological properties of magmas at sub-liquidus temperatures. Journal of Volcanology and Geothermal Research 53, 47-6 6. Chapter 7 Reconstruction of Eruption Mechanisms 256 Platz, T. 2001. Mapping and characterisation of the vol caniclastic Maero formation deposits on the northwestern sector of Egmont volcano (Mt. Taranaki), New Zealand. Unpublished Diploma mapping thesis. Ernst-Moritz-Arndt Universitä t, Greifswald, Bundesrepublik Deutschland. Platz, T., Foley, S.F. and André, L. 2004. Low-pressure fractionation of the Nyiragongo volcanic rocks, Virunga Province, D.R. Congo. Journal of Volcanology and Geothermal Research 136, 269-2 95. Platz, T., Cronin, S.J, Cashman, K.V., Stewart, R.B. and Smith, I.E.M. 2007. Transition from effusive to explosive phases in andesite eruptions - a case-study from the AD1655 eruption of Mt. Taranaki, New Zealand. Journal of Volcanology and Geothermal Research 161, 15-34. Richet, P., Lejeune, A. - M., Holtz, F. and Roux, J. 1996. Water and the viscosity of andesite melts. Chemical Geology 128, 185–1 97. Robertson, R., Cole, P., Sparks, R.S.J., Harford, C., Lejeune, A.M., McGuire, W.J., Miller, A.D., Murphy, M.D., Norton, G., Stevens, N.F. and Young, S.R. 1998. The explosive eruption of Soufriere Hills Volcano, Mont serrat, 17 September, 1996. Geophysical Research Letters 25, 3429 - 3 432. Roman, D.C., Cashman, K.V., Gardner, C.A., Wallace, P.J. and Donovan, J.J. 2006. Storage and interaction of compositionally heterogeneous magmas from the 1986 eruption of Augustine Volcano, Alaska. Bulletin of Volcanology 68, 240- 254. Romano, C., Giordano, D., Papale, P., Mincione, V., Dingwell, D.B. and Rosi, M. 2003. The dry and hydrous viscosities of alkaline melts from Vesuvius and Phlegrean Fields. Chemical Geology 202, 23-3 8. Roscoe, R. 1952. The viscosity of suspensions of rigid spheres. British Journal of Applied Physics 3 , 267- 269. Sahetapy-Engel, S.T.M., Flynn, L.P., Ha rris, A.J.L., Bluth, G.J., Rose, W.I. and Matias, O. 2004. Surface temperature and spectral measurements at Santiaguito lava dome, Guatemala. Geophysical Research Letters 31, L1961 0. Sato, H. 2005. Viscosity measurement of subliquidus magmas: 1707 basalt of Fuji volcano. Journal of Mineralogical and Petrological Sciences 100, 133-1 42. Sato, H., Nakada, S., Fujii, T., Nakamura, M. and Suzuki-Kamata, K. 1999. Groundmass pargasite in the 1991-1 995 dacite of Unzen volcano: phase stability experiments and volcanological implications. Journal of Volcanology and Geothermal Research 89, 197- 212. Scandone, R. and Malone, S.D. 1985. Magma supply, magma discharge and readjustment of the feeding system of Mount St. Helens during 1980. Journal of Volcanology and Geothermal Research 23, 239 - 2 62. Scholz, E. 1984. Karl Fischer Titration. Springer Verlag. Berlin. Chapter 7 Reconstruction of Eruption Mechanisms 257 SEAN 02:03. Smithsonian Institution, 1977. Mount Nyirag ongo. Scientific Event Alert Network vol. 2, no. 3. Shaw, H.R. 1972. Viscosities of magmatic silicate liquids: an empirical method of prediction. American Journal of Science 272, 870 -89 3. Siswowidjoyo, S., Suryo, I. and Yokoyama, I. 1995. Magma eruption rates of Merapi volcano, Central Java, Indonesia during one century (1890 -199 2). Bulletin of Volcanology 57, 111-1 16. Sparks, R.S.J. 1997. Causes and consequences of pressurisation in lava dome eruptions. Earth and Planetary Science Letters 150, 177-1 89. Sparks, R.S.J., Murphy, M.D., Lejeune, A.M., Watts, R.B., Barclay, J. and Young, S.R. 2000. Control on the emplacement of the andesite lava dome of the Soufriere Hills volcano, Montserrat by degassing-induced crystallization. Terra Nova 12, 14-2 0. Stewart, R.B., Price, R.C. and Smith, I.E.M. 1996. Evolution of high-K arc magma, Egmont volcano, Taranaki, New Zealand: evidence from mineral chemistry. Journal of Volcanology and Geothermal Research 74, 275-2 95. Stolper, E. 1982. The speciation of water in silicate melts. Geochimica et Cosmochimica Acta 46, 2609 - 2 620. Stuart, B.H. 2004. Infrared spectroscopy: fundamentals and applications. John Wiley and Sons Ltd. Chichester. Suto, S., Sakaguchi, K., Watanabe, K., Saito, E., Kawanabe, Y., Kazahaya, K., Takarada, S. and Soya, T. 1993. Dynamics of flowage and viscosity of the 1991 lava of Unzen Volcano, Kyushu, Japan. Bulletin of the Geological Survey of Japan 44, 609- 62 9. Taron, J., Elsworth, D., Thompson, G. and Voight, B. 2007. Mechanisms for rainfall-concurrent lava dome collapses at Soufrière Hills Volcano, 2000-2 002. Journal of Volcanology and Geothermal Research 160, 195- 209. Toplis, M.J. and Dingwell, D.B. 1996. The variable influence of P 2 O 5 on the viscosity of melts of differing alkali/aluminium ratio: implications for the structural role of phosphorus in silicate melts. Geochimica et Cosmochimica Acta 60, 4107- 412 1. Toplis, M.J., Dingwell, D.B. and Libourel, G. 1994. The effect of phosphorus on the iron redox ratio, viscosity, and density of an evolved ferro-basalt. Contributions to Mineralogy and Petrology 117, 293- 304. Voight, B., Sparks, R.S.J., Miller, A.D., Stewart, R. C., Hoblitt, R.P., Clarke, A., Ewart, J., Aspinall, W.P., Baptie, B., Calder, E.S., Cole, P., Druitt, T.H., Hartford, C., Herd, R.A., Jackson, P., Lejeune, A.M., Lockhart, A.B., Loughlin, S.C., Luckett, R., Lynch, L., Norton, G.E., Robertson, R., Watson, I.M., Watts, R. and Young, S.R. 1999. Magma flow instability and cyclic activity at Soufriere Hills Volcano, Montserrat, British West Indies. Science 283, 113 8-1 142. Chapter 7 Reconstruction of Eruption Mechanisms 258 Whittington, A., Richet, P., Linard, Y. and Holtz, F. 2001. The viscosity of hydrous phonolites and trachytes. Chemical Geology 174, 209 -22 3. Wooster, M.J., Kaneko, T., Nakada, S. and Shimizu, H. 2000. Discrimination of lava dome activity styles using satellite-derived thermal structures. Journal of Volcanology and Geothermal Research 102, 97-118. Zernack, A., Stewart, R.B. and Price, R.C. 2006. Slab or crust - K 2 O enrichment at Egmont volcano, New Zealand. Abstract. Geochimica et Cosmochimica Acta Supplement 70, 614. Zimova, M. and Webb, S.L. 2007. The combined effects of chlorine and fluorine on the viscosity of aluminosilicate melts. Geochimica et Cosmochimica Acta 71, 1553- 156 2. Chapter 8 Conclusions 259 Chapter 8 Conclusions Over the last 1000 yrs of activ ity at Mt. Taranaki, a range of eruption products were generated: from pyroclastic flow deposits (i.e. Block-and-Ash Fl ow deposits, pumice flow units and surge deposits), to ash, lapilli, and pumice ous tephra layers, along with lava flows and domes. The previous eruptive reconstruction of this period, (here newly defined as the Maero Eruptive Period), have been revised to comprise at least 10 eruptive episodes, including predominantly extrusive, lava dome- and lava flow- producing eruption episodes and one explosive, sub-Plinian erupti on. Field-stratigraphy and glass chemical studies were used to correlate tephra units across different sectors of the mountain. A new stratigraphy of the Maero Formation is introduced, which correlates fall units of the eastern and southern flanks to pyrocla stic flow and fall deposits of the western to northwestern sector. A replacement type section for the Maero Formation is defined in Maero Stre am with an additional 9 reference sites positioned around the edifice. Despite the de tailed knowledge of lava dome-forming episodes, the stratigraphic clas sification of lava flow-produc ing eruptions is still poorly constrained. Also, it remains unknown whethe r lava flow-forming eruptions represent individual eruptive events or were related to lava-dome eruptions. As a result of analysing glass compositions of ash particles derived from flow and fall deposits, a new evaluation procedure has been developed that examines glass datasets for contaminated mineral-glass analyses. Th ese hybrid analyses can be identified by using bivariate compositional diagrams where an incompatible and a compatible element oxide of the mineral phase is pl otted. Because this is a simple mixing relationship, hybrid analyses are detectable if linear data trends point towards the composition of the mineral phase. Once potenti al hybrid analyses have been detected they must be checked for all other remain ing element oxides. Although exclusion of contaminated glass analyses improves and clusters the glass dataset by reducing the sample variance, analysed ash/lapilli particle textures also need to be examined. It was Chapter 8 Conclusions 260 demonstrated that during any one eruption episode a range of particle types form during individual eruption phases. Hence, tephras at Mt. Taranaki comm only contain two or more particle populations, each having matrix glass of differing compositions. Hence, particle glass chemistry from the combined unit may be unusable for tephra correlation purposes. Through the study of particle textures , glass chemistry and the application of the glass data evaluation procedure, a higher degree of resolution of the tephrostratigraphic record can be achieved. Detailed studies on the remnant summit dome of Mt. Taranaki we re carried out to understand lava dome emplacement and destruction mechanisms. It has been shown that the remnant dome (Pyramid Dome) was not formed during the Tahurangi eruptive episode but extruded post-AD1755. The main evidence for this conclusion stems from bulk rock geochemistry and the internal st ructure of the dome. Identified Tahurangi Block-and-Ash flow (BAF) deposits have a different composition from the Pyramid Dome, particularly in Al 2 O 3 , Mg#, Cr and Zr contents. The homogenous dome composition, its simple structure, and the ab sence of former dome remnants in the crater, clearly indicate an individual eruption, the Pyramid eruptive episode. Reconstruction of the inferred original dome geometry implies extrusion of a maximum volume of 5.9×10 6 m3 (4.9×10 6 m3 DRE) of lava. Due to the crater morphology, the dome was emplaced partially onto the crater floor as well as on the upper steep north- western flank, where some lava lobes advanced up to 300 m downhill. The observation of individual lava lobes on the upper flank, th e hemispherical eastern dome side within the crater and the occurrence of blocky la va flows imply simultaneous endogenous and exogenous dome growth. In consideration of th e texture of hornblende crystals and the formation of a thin vesicular dome carapace, it is further concluded that emplacement duration was rapid. By comparing the dome hornblende compositions with other well- studied volcanoes, it is assume d that Taranaki hornblende is stable up to 1 kbar or c.2 km depth. The arrival of pristine hornblende within the lava dome suggests minimum magma ascent rates of 0.012 ms -1 . Hence, the Pyramid Dome emplacement could have taken place within days. The magma storage depth(s) are only poorly constrained and based only on the hornblende geobarometer. According to this geobarometer, hornblende crystallises at pressures between 3-5 kbar (c.9.5-17 km). The lower crystallisation limit coincides with depth estimates of the Brittle -Ductile Transition Zone beneath Mt. Taranaki. Hence, magmas ma y stagnate at this level due to reaching Chapter 8 Conclusions 261 the level of neutral buoyancy. However, it is emphasised that temporary shallower magma stalling levels may also exist, as evidenced from co-magmatic xenoliths. The rock-avalanche deposit generated by the Pyramid (summit) Dome failure was interpreted to have occurred at or below c.350 °C, based on the random magnetic alignment of clasts. This also implies that the dome was at similar temperatures, i.e., cool. Hence, an important conclusion is th at the dome probably failed years after its formation, implying a potentially unexpected post-eruptive hazard. By applying simple conductive cooling models, a period of about 19 yrs is required to cool the dome to c.350 °C, or 74 yrs for it to cool completel y. Although the date of the Pyramid eruption cannot be well constrained, it most likely took place at some time in the period AD 1839- 1866. Transitions from initially extrusive, dome-f orming eruptions to explosive, sub-Plinian phases can occur suddenly and without any warning. This occurred during the AD 1655 Burrell episode, which produced widespread deposits of pumice fallout, pyroclastic pumice flows, BAFs and surges. The fallout, which was initially directed towards E and then NE is dominated by grey, highly ve sicular pumice clasts and a large (c.14%) component of dense lithic fragments. In co ntrast, pumice flow de posits show a variety of vesicular clasts ranging from brown and grey coloured pumice to semi-vesicular to vesicular black scoria. Dense lithics, breadcr ust bombs, and accidental wall rock clasts are also common. Pumice flow deposits are restricted to the upper to mid flanks of the volcano, and are predominantly preserved on the steep volcano flanks. Block-and-Ash flow deposits are found predominantly on the NW sector; the major BAF unit during the Burrell episode travelle d down the Hangatahua River more than 13.5 km from source. Pumice clasts at c.6 vol.% are the second most abundant component in this deposit. The total volume of erupted magma during the Burrell episode is estimated at 0.12 km 3 DRE. The formation of a variety of dens e to vesicular clasts during this event was related to differing eruption phases. The grey lithic clasts likely represent the initial extrusive phase, where a dome within the summit crater was formed. Dome removal generated BAFs and triggered an explosive, sub-Plinian eruption. The sub-Plinian phase was further subdivided into three pulses in which grey, br own and grey pumice clasts erupted. Each pulse corresponded to individual layers in the conduit caused by different degrees of vesiculation and crystallisation. The magma producing brown pumice clasts contained more microlites and had a more evolved matrix glass composition compared to magmas above and below it, which prod uced grey pumice. Since there are no Chapter 8 Conclusions 262 geophysical indicators for large magma batches in the upper crust, magma ascent is assumed to start at depths of c.9.5 km. Th e bulk compositional variation of erupted clasts (pumice to dense dome rocks) can be modelled by fractionation of hornblende, plagioclase, clinopyroxene, and Fe-Ti oxides. The differentiation to more silicic dome rocks occurred during magma ascent up to dept hs of 2 km, where the inferred stability limit of hornblende is reached, i.e. at shallower depth hornblende is not a fractionating phase and hence, rock differentia tion is significantly lowered. The Maero Eruptive Period consists of at least 10 eruptive episodes of contrasting eruptive style. It appears that every erup tive period produced lava dome(s), some of them preceded by a vent-clearing phase produci ng single lithic layers around the edifice (Newall Ash a, Mangahume Lapilli, Pyramid Lapilli). The type of dome failure controlled successive eruptive phases in most instances. Gravitational dome failures did not cause a change in eruptive style, as inferred for the Tahurangi episode. By contrast, the destruction of a pressurised dome caused instantaneous but short-lived magmatic fragmentation (Newall and Puniho episodes) or triggered a directed blast-explosion (Newall episode), or initiated sustained magmatic fragmentation (Burrell Episode). From a hazard and emergency management pers pective, the above described results are important in showing the variety of courses a future eruption may take at Mt Taranaki. They also point out that eruptions at this volcano are often more than just “one-off” occurrences. Instead, some of the recent episodes (e.g. the Bu rrell Episode), involved a long and complex chain of events that may have taken place over several months of activity. If Mt Taranaki continues to behave as it has over the last 1000 years, the next eruption will very likely involve dome extr usion, probably preceded by vent clearing blasts. The course of events thereafter will be highly dependent on (1) the state of the dome and upper conduit – particularly its ab ility to degas and the supply rate of new magma, and (2) the stability of the crater area. These features should be intensely monitored, and samples of ash or dome mate rial taken on many occasions to compare and contrast to the described Maero Period ev ents. The results of this study show that a highly explosive event is a possibility, but by no means certain during any future eruption. In addition, since the most widespread and destructive events are likely to be dense pyroclastic flows or BAFs, the geometry of the summit area will be highly critical to determine areas at greatest risk. Chapter 8 Conclusions 263 8.1 Avenues of Future Research In completing this study, it is recognised that it is only a step along the way to understanding the overall scientific questions of andesitic volcanism, as well as answering questions of likely volcanic risk at Mt Taranaki. Upon completion of this work, the following issues and questions are raised for subsequent studies: 1) The knowledge of water concentrations in silicate glasses is essential in order to understand degassing processes prior to and during volcanic eruptions. What is the total water budget dissolved in the pre-eruptive magma? How much water escapes during magma ascent through permeable wall rocks? How much water is released during the eruptions, and how much water remains in volcanic glasses? What other volatile components (e.g. F, Cl, CO 2 , SO 3 ) are released during eruptions and what is their impact on surrounding areas. Preliminary results of using thermal analysis for water estimates in matrix glasse s are encouraging and may answer some of the above questions. 2) For a better understanding of magmatic fragmentation, e.g. during sub-Plinian eruptions, porosity determinations of pumice clasts need to be complemented by studies on bubble aperture widths and tortuosities of vesicular samples (e.g. by electrical conductivity and high-resolution X-ray tomogr aphy). These results will help constrain timescales of degassing and fragmentation. 3) Tectonic aspects, such as the surrounding stress regime of volcanoes, need further consideration in understanding volcanic processes. At Mt. Taranaki, for example, this has not yet been concentrated on in detail. The occurrence of repeated edifice failure and the presence of faults near, and presumably beneath the volcano, may have impacts on the eruptible magma volume and composition. Sudden edifice collapse and hence, pressure release may promote the ascent and eruption of larger magma batches at depth. Changes in the prevailing crustal stress regime, for example from compressional to extensional, may enable and ease dyke initiation and propagation and thus trigger magma movement in the crust. Seismic monito ring techniques are the most reliable (and only) methods to identify precursors of ma gma movement within the mid-upper crust. Using the petrographic results from this study, critical areas to be focussed upon are potential mid-crustal ponding areas, where risi ng magmas may initially stall due to their initially high density. Hence, the collabo ration between geophysicists, geologists, and Chapter 8 Conclusions 264 volcanologists need to be intensified and focussed on geophysical monitoring techniques. 4) What is the maximum load pressure of lava that Mt Taranaki’s summit crater can resist before failure? The determination and assessment of the mechanical rock properties of crater-rim lava flows will aid hazard mitigation during future volcanic eruptions. For example, should a lava dome slowly grow within a summit crater, the hazard zone could be extended once the estimated threshold value for the structural stability of the crater walls is exceeded. Thus, the loss of lives can be prevented (or reduced) by sudden dome collapse due to flan k instabilities and th e possible triggering of a blast-explosion or sub-Plinian eruption. 5) During the Maero Eruptive Period many Block-and-Ash flows were generated by the destruction of lava domes but only a few of them travelled >10 km from source. What are the physical aspects controlling pyroclastic-flow runouts? What effect do topography, vegetation, and channe lisation have on the runout of pyroclastic flows? Do the physical properties of the lava dome (e.g. temperature, rock strength, density) influence the flow behaviour of Block-and-Ash Flows? I Appendices Type and Reference Sections V Type Section of the Maero Formation – Maero Stream V Reference Section 1 – Pyramid Stream XV Reference Section 2 – Hangatahua River XXI Electron Microprobe Analysis XXV Amphibole XXV Apatite XXXI Biotite XXXIII Clinopyroxene XXXIV Fe-Ti Oxides XLI Glass Inclusions XLVII Matrix Glass L I I I Olivine LXXVII Orthopyroxene LXXVI I I Plagioclase LXXIX X-ray Fluorescence Spectroscopy LXXXIII Major Element Oxides LXXXI I I Trace Elements XXXVI Laser Inductively Coupled Plasma Mass Spectroscopy LXXXIX Trace Elements LXXXIX II Appendices on DVD Digital Copy of Thesis Fourier Transform Infrared Spectroscopy Graphs Images Geochemistry Electron Microprobe Analysis Original Data Files and Images Amphibole Apatite Biotite Clinopyroxene Fe-Ti Oxides Glass Inclusions Matrix Glass Olivine Orthopyroxene Plagioclase Unknown Minerals Laser Inductively Coupled Plasma Mass Spectroscopy Original Data Files Laser ICP-MS Data X-ray Fluorescence Spectroscopy Original Data Files XRF Analyses Permeability Graphs Original Data Files Permeability_2003&2004 III Porosity Massey_2006 Oregon_2003&2004 Reference List Selected References Scanning Electron Microscopy Massey Oregon Thermal Analysis Original Data Files DSC-TGA Analysis Thin Section Photographs Type and Reference Sections Reference Section 1 Reference Section 2 Type and Reference Sections Type Section Electron Microprobe Analysis Amphibole XXV Electron Microprobe Analysis Amphibole XXVI Electron Microprobe Analysis Amphibole XXVII Electron Microprobe Analysis Amphibole XXVII I Electron Microprobe Analysis Amphibole XXIX Electron Microprobe Analysis Amphibole XXX Electron Microprobe Analysis Apatite XXXI Electron Microprobe Analysis Apatite XXXII Electron Microprobe Analysis Biotite XXXII I Electron Microprobe Analysis Clinopyroxene XXXIV Electron Microprobe Analysis Clinopyroxene XXXV Electron Microprobe Analysis Clinopyroxene XXXVI Electron Microprobe Analysis Clinopyroxene XXXVII Electron Microprobe Analysis Clinopyroxene XXXVII I Electron Microprobe Analysis Clinopyroxene XXXIX Electron Microprobe Analysis Clinopyroxene XL Electron Microprobe Analysis Fe-Ti Oxide XLI Electron Microprobe Analysis Fe-Ti Oxide XLII Electron Microprobe Analysis Fe-Ti Oxide XLIII Electron Microprobe Analysis Fe-Ti Oxide XLIV Electron Microprobe Analysis Fe-Ti Oxide XLV Electron Microprobe Analysis Fe-Ti Oxide XLVI Electron Microprobe Analysis Glass Inclusions XLVII Electron Microprobe Analysis Glass Inclusions XLVIII Electron Microprobe Analysis Glass Inclusions XLIX Electron Microprobe Analysis Glass Inclusions L Electron Microprobe Analysis Glass Inclusions LI Electron Microprobe Analysis Glass Inclusions LII Electron Microprobe Anal ysis Matrix Glass LIII Electron Microprobe Anal ysis Matrix Glass LIV Electron Microprobe Anal ysis Matrix Glass LV Electron Microprobe Anal ysis Matrix Glass LVI Electron Microprobe Anal ysis Matrix Glass LVII Electron Microprobe Anal ysis Matrix Glass LVIII Electron Microprobe Anal ysis Matrix Glass LIX Electron Microprobe Anal ysis Matrix Glass LX Electron Microprobe Anal ysis Matrix Glass LXI Electron Microprobe Anal ysis Matrix Glass LXII Electron Microprobe Anal ysis Matrix Glass LXIII Electron Microprobe Anal ysis Matrix Glass LXIV Electron Microprobe Anal ysis Matrix Glass LXV Electron Microprobe Anal ysis Matrix Glass LXVI Electron Microprobe Anal ysis Matrix Glass LXVII Electron Microprobe Anal ysis Matrix Glass LXVIII Electron Microprobe Anal ysis Matrix Glass LXIX Electron Microprobe Anal ysis Matrix Glass LXX Electron Microprobe Anal ysis Matrix Glass LXXI Electron Microprobe Anal ysis Matrix Glass LXXII Electron Microprobe Anal ysis Matrix Glass LXXIII Electron Microprobe Anal ysis Matrix Glass LXXIV Electron Microprobe Anal ysis Matrix Glass LXXV Electron Microprobe Anal ysis Matrix Glass LXXVI Electron Microprobe Analysis Olivine LXXVI I Electron Microprobe Analysis Orthopyroxene LXXVI II Electron Microprobe An alysis Plagioclase LXXIX Electron Microprobe An alysis Plagioclase LXXX Electron Microprobe An alysis Plagioclase LXXXI Electron Microprobe An alysis Plagioclase LXXXI I X-ray Fluorescence Spectrosc opy Major Element Oxides LXXXI II X-ray Fluorescence Spectrosc opy Major Element Oxides LXXXI V X-ray Fluorescence Spectrosc opy Major Element Oxides LXXXV X-ray Fluorescence Spectroscopy Trace Elements LXXXV I X-ray Fluorescence Spectroscopy Trace Elements LXXXV II X-ray Fluorescence Spectroscopy Trace Elements LXXXV II I Laser Inductively Coupled Plasma Mass Spectrometry Trace Elements LXXXI X Laser Inductively Coupled Plasma Mass Spectrometry Trace Elements XC