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. THE USE OF A GEOGRAPHIC INFORMATION SYSTEM TO INVESTIGATE SOIL SLIP DISTRIBUTION AND THE LAND USE CAPABILITY CLASSIFICATION IN THE EAST COAST REGION, NEW ZEALAND. A thesis presented in partial fulfilment of the requirements for the degree of Master of Applied Science in Soil Science at Massey University SHERYL DENISE HENDRIKSEN 1995 ii ABSTRACT The land of the North Island East Coast region has such a severe erosion problem that in some places the current land use cannot be sustained. The expansion of exotic forestry in the region will provide protection for the land, regional growth and development, and employment, but it also brings competition for good land. The New Zealand Resource Management Act, 1991, aims to promote sustainable use of our resources and requires regulatory authorities to monitor the state of their natural resources and to follow the principles set in the RMA when developing land use policies. Remotely sensed data provides a timely and accurate assessment of surface features . Aerial photography provides a better delineation of soil slip erosion than satellite 1magery. Geographic Information Systems facilitate the storage and display of resource information. Through manipulation of GIS data layers, relationships between the distribution of soil slip erosion following Cyclone Bola, 1988, and other physical factors are investigated. The density of soil slip increases with increasing slope angle to a maximum on slopes of 300. The amount of soil slip depends on the underlying rock type with jointed mudstone having the highest density. Most soil slip erosion occurs on NE, N, NW, and E facing slopes, but the reason for this cannot be attributed to either slope angle or rock type. The Land Use Capability classification is currently used by land use managers and planners to describe the land in terms of its limitation to productive uses. The detail of information in the New Zealand Land Resource Inventory LUC classification can be improved by incorporating more detailed slope angle and slope aspect information derived from digital contour data. lll ACKNOWLEDGMENTS I would like to thank Massey University and my supervisors Mr Mike Tuohy, Dr Vince Neall and Dr Alan Palmer for the opportunity to explore GIS, image analysis and the sustainability of present land uses in the East Coast. I would also like to thank the Ministry of Research, Science and Technology for their study award that met the expenses of this project. I am extremely grateful to Mr Trevor Freeman of the Gisbome District Council and Dr Mike Marden of Forest Research Institute, Gisbome who provided me with information and an appreciation of the uniqueness of the East Coast and its land use Issues. During the completion of this thesis I have received friendship and assistance from many more people than I can thank individually here. Among these are Mr Hoole of Emerald Hills station and Mr and Mrs Shanks ofNgamarua station who kindly allowed me to wander over their properties; the staff of Landcare, Massey; and all my fellow postgraduates who so willingly provided assistance and moral support. Special thanks are owed to Mr Len Brown and Miss Jocelyn Young for their guidance in image analysis and GIS, and for their continued friendship, I would not have completed this thesis without them. Once again, I thank my best friends; John, Gabrielle and Julian for sharing with me the tears and joys of this project. ABSTRACT ACKNOWLEDGEMENTS CONTENTS LIST OF FIGURES LIST OF TABLES CHAPTER I: INTRODUCTION 1.1 INTRODUCTION 1.2 OBJECTIVES OF THIS STUDY CHAPTER II: CONTENTS IDSTORY OF LAND USE, POLICIES AND ASSOCIATED PROBLEMS IN THE EAST COAST REGION 2.1 INTRODUCTION 2.2 SETILEMENT AND LAND CLEARANCE 2.3 ACCELERATED EROSION 2.4 CONFRONTING THE EROSION . 2.5 THETAYLORREPORT . 2.6 THE EAST COAST PROJECT, 1970. 2. 7 THE RED REPORT 2.8 PRODUCTION FOREST DEVELOPMENT 2.9 CYCLONE BOLA 2.10 EROSION CONTINUES . 2.11 LAND USE INCENTIVES 2.11.1 Agricultural Incentives 2.11.2 Forestry Incentives 2.12 EAST COAST PROJECT CONSERVATION FORESTRY SCHEME, 1990 . 11 lll IV Vlll X 2 5 8 9 9 11 12 12 14 15 18 19 19 20 20 iv 2.13 NGATI POROU FORESTS LTD 2.14 EAST COAST FORESTRY PROJECT, 1993 CHAPTER III: DESCRIPTION OF THE STUDY AREA 3.1 FEATURES OF THE STUDY AREA 3.2 PHYSIOGRAPHY 3.2.1 Tectonic Setting 3.2.2 Geological Structure of the Te Arai River catchment 3.2.3 Lithologies of the Te Arai River catchment 3.2.4 Tephra cover in the Te Arai River catchment. 3.2.5 Soils of the Te Arai River catchment. 3.3 CLIMATE 3.4 EROSION IN THE TE ARAI RIVER CATCHMENT CHAPTER IV: DATA COLLECTION 4.1 INfRODUCTION 4.2 EXISTING DATA SETS . 4.3 FIELD SURVEYS 4.4 TOPOGRAPHIC DATA 4.5 REMOTELY SENSED INFORMATION 4.5.1 Satellite Imagery . 4.5.2 Aerial Photography 4.6 SPATIAL CORRELATION OF DATA FROM VARIOUS SOURCES CHAPTER V: 5.1 5.2 5.3 DIGITAL IMAGE ANALYSIS INTRODUCTION THEORY AND PREVIOUS RESEARCH DETERMINATION OF CYCLONE BOLA SOIL SLIPS 5.3.1 Cyclone Bola soil slips derived from Satellite Imagery 5.3 .2 Cyclone Bola soil slips derived from Aerial Photography 21 21 24 25 25 25 26 30 34 35 36 39 39 40 40 44 44 44 45 46 47 51 51 55 v 5.4 505 EVALUATION OF DIGITAL IMAGE ANALYSIS 0 50401 Satellite Imagery 0 50402 Aerial Photography DISCUSSION CHAPTER VI: 601 602 603 604 605 INVESTIGATING THE CHARACTERISTICS OF SOIL SLIP DISTRIBUTION USING A GEOGRAPlllC INFORMATION SYSTEM INTRODUCTION PREVIOUS WORK 60201 Aspect 0 60202 Slope Angle 60203 Lithology METHODOLOGY RESULTS AND DISCUSSION 6.401 Slope Aspect 60402 Slope Angle 60403 Slope Angle per Aspect 60404 Rock Type 6.405 Slope Angle per Rock Type 60406 Rock Type per Slope Aspect DISCUSSION CHAPTER VII: LAND USE CAP ABILITY MAPPING 701 INTRODUCTION 702 LAND USE CAP ABILITY MAPPING IN NEW ZEALAND 0 70201 New Zealand Land Resource Inventory 70202 Gisborne District Council Erosion Categories 703 THE SUIT ABILITY OF THE LUC CLASSIFICATION FOR DESCRIBING LAND WHICH IS PRONE TO SOIL SLIP EROSION 704 LUC MAPPING USING THE GIS 0 60 60 61 62 65 66 66 67 69 70 72 72 73 75 77 79 80 82 85 85 87 88 90 93 Vl 7.4 .1 Revised LUC mapping using NZLRllithology and erosion, and TIN generated slope angles 7.4.2 Udpating the revised LUC units using TIN generated slope aspect information 7.5 DISCUSSION CHAPTER VIII: 8.1 8.2 CONCLUSIONS AND SUGGESTIONS FOR FUTURE WORK CONCLUSIONS . SUGGESTIONS FOR FUTURE WORK REFERENCES APPENDIX 1 95 100 105 107 110 112 123 Vll Figure 1.1 1.2 LIST OF FIGURES The location of this study in the East Coast Region, North Island, New Zealand Emerald Hills station in the Te Arai River catchment 2.1 The distribution of rainfall during Cyclone Bola in the Te Arai catchment compared with the average annual rainfall . 3.1 Geologic structure of the Te Arai River catchment, from Brown (1961) . 3.2 The geology of Emerald Hills station mapped from field survey compared with the lithology information available on the farm plan 3.3 3.4 3.5 The Te Arai syncline in crushed mudstone with jointed mudstone on the right tilted up towards the east . To the west of the Te Arai syncline the banded mudstone and sandstone is tilted up towards the west Faults are observed in bands of crushed mudstone 3.6 Three tephras are found on the broader ridges and shoulders. 3.7 3.8 3.9 3.10 4.1 4.2 5.1 5.2(a) 5.2(b) 5.3 5.4(a) 5.4(b) 5.5 The Waiohau ash overlying Mudstone indicates that no soil parent material in the area is older than 11 ,200 years. The colluvium between the Waiohau and the Waimihia infers the instability of that period. Areas on Emerald Hills station where all three tephras may be found Earthflow erosion on Emerald Hills station Gully infilling following Cyclone Bola Soil slip erosion following Cyclone Bola Slope angles on Emerald Hills station derived from 20m digital contour data Slope aspects on Emerald Hills station derived from 20m digital contour data Satellite image of Emerald Hills station, registered to NZ metric grid A sample area from the satellite image Satellite image (a) with areas classified as bare ground overlaid Sample area from aerial photography (a), and the results of classification of the photograph scanned at !Om (b), 5m (c) and 1m (d) resolution A sample area from aerial photograph scanned at lOrn resolution Aerial photograph(a) with areas classified as bare ground overlaid A comparison of soil slip erosion derived from (a) satellite imagery and (b) aerial photography 6.1 Overlay ('intersection') procedures to provide soil slip per landscape feature information 6.2 Overlay ('Union') procedures and associated database polygon attribute tables Vlll Page 3 4 17 26 28 29 29 30 33 33 37 37 38 42 43 53 54 54 57 58 58 59 71 71 lX 6.3 The distribution of slope aspects on Emerald Hills station 73 6.4 The extent of soil slip in each slope aspect 73 6.5 The distribution of slope angles on Emerald Hills station 74 6.6 The distribution of soil slip per slope angle class 75 6.7 The distribution of slope class per slope aspect 76 6.8 The density of soil slip on Slope* Aspect class 77 6.9 The distribution of rock types on Emerald Hills station 78 6.10 The density of soil slip per rock type 78 6.11 The distribution of slope angle per rock type. 79 6.12 The distribution of rock type per slope aspect 81 7.1 A comparison of LUC mapping for Emerald Hills station on the NZLRI and on the GDC farm plan 89 7.2 The distribution of soil slip per LUC unit 90 7.3 A comparison of slope information for areas mapped as Vle3 . 92 7.4 A comparison of the revised LUC mapping with the NZLRI 98 7.5 A comparison of the revised LUC mapping with the GDC farm plan 99 7.6 A comparison of the LUC mapping further updated by the inclusion of slope aspect information with the farm plan 101 7.7 GDC farm plan LUC units are assigned in relation to landforms whereas the GIS method assigns them according to set criteria for lithology, slope and erosion 102 7.8 The distribution of Cyclone Bola soil slip on the revised LUC map 103 X LIST OF TABLES Table Page 2.1 History of the development and land use policies of the East Coast 7 2.2 Description of Gisborne District Council Erosion Categories 13 3.1 Tephras deposited in the Gisborne region, from Pullar ( 1973). 31 6.1 Calculation of the areal extent of each aspect which was affected by soil slip during Cyclone Bola, 1988 72 6.2 The extent of jointed, banded and massive mudstone per slope aspect 82 7.1 The Gisbome-East Coast rgion LUC classification description of the Units used in the revised schedule (below) 94 7.2 Revised description ofLUC units used in this study 95 7.3 Description of the LUC units mapped on the GDC farm plan 97 1 CHAPTER I INTRODUCTION 1.1 INTRODUCTION The New Zealand Resource Management Act , 1991 (RMA), requires the use of land to be managed in such a way as to ensure its potential to meet the reasonably foreseeable needs of future generations and to avoid, remedy or mitigate any adverse effects on the environment resulting from its use. The ability of the land in the East Coast of the North Island (Figure 1.1) to sustain pastoral land use has long been questioned. The combined factors of easily eroded, tectonically tilted and crushed soft rocks and periodic high rainfall storms produce spectacular flooding and erosion. The results of such events are erosion of the pastoral hill slopes and inundation of the intensely cropped downstream flats by floodwaters and silt. The mitigation and remedy of large scale erosion has been appreciated as being beyond the resources of the district with the formation of cental government funded schemes such as the East Coast Project 1970, and the East Coast Forestry Project, 1993 . These schemes recognise that certain classes of land do not have the capacity to sustain pastoral use and should be planted in production or protection forests to reduce their susceptibility to erosion. The classification and control of land in terms of sustainable land use was the responsibility of the Catchment Boards until they were abolished in 1991 . Soil conservation duties in the East Coast were incorporated into the restructured and resource limited Gisbome District Council (GDC). Because of the cost recuperation policies of the government, random land survey and research is no longer carried out. Land inventory mapping is completed only by interested parties for specific purposes. The only land use capability (LUC) information available for all of New Zealand is the 2 New Zealand Land Resource Inventory (NZLRI). The information contained in this land inventory is recommended for regional planning; it is often of too small a scale to be used for detailed land use planning. In carrying out their duties as defined in the RMA. local authorities are faced with the need to be able to investigate the state of their natural and physical resources; to classify the land in terms of its land use capability at the management-unit scale; and to predict the effects of present or proposed land use on the immediate and downstream environment. 1.2 OBJECTIVES OF THE STUDY This project investigates the usefulness of a geographic information system (GIS) and remotely sensed information in fulfilling these duties. Image analysis techniques enable the acquisition of information relating to the state of resources from remotely sensed data. This can be accomplished in much shorter time than land based surveys. Computerised geographic information systems facilitate the storage of large amounts of spatially referenced information which can quickly be retrieved and manipulated for display or analysis. Emerald Hills station in the Te Arai River catchment (Figure 1.2) was selected for this project because it has good vehicle and foot access; it has a variety of lithologies which are detailed on a 1:10,000 scale soil and water conservation farm plan along with other physical attributes; and it sustained considerable soil slip erosion during Cyclone Bola. Specific objectives of the project include: (1) the delineation of soil slip erosion on Emerald Hills station following Cyclone Bola, 1988; (2) the determination of any relationships between the distribution ofthat soil slip and site factors; (3) the investigation of the success with which the Land Use Capability mapping system describes the proneness of the land to soil slip erosion; (4) an attempt to improve the detail of the New Zealand Land Resource Inventory, Land Use Classification. / . Figure 1.1 20 15ml SCALE 1 l5Qtl00 The location of this study in the East Coast Region, North Island, New Zealand. 3 Figure 1.2 Emerald Hills station in the Te Arai River catchment. Photo view from farm down catchment to Gisborne. 4 2.1 CHAPTER II HISTORY OF LAND USE, POLICIES AND ASSOCIATED PROBLEMS IN THE EAST COAST REGION INTRODUCTION 5 The East Coast is on the active margin of a slab of continental crust, the Indian­ Australian Plate. Young unconsolidated marine rocks have been faulted, tilted and elevated up to l,OOOm above sea level, since the Miocene period, by pressure of the incoming subducting Pacific Plate. During the Quaternary, fluctuations in climate and sea level have added further instability leading to major cycles of natural geological eros10n. In recent times erosion rates have been among the highest in the world. This erosion and subsequent sedimentation has been occurring since well before the occupation by humans. After mapping recent soils on alluvium on the Gisbome Plains the dates of c.AD1650 and c.AD1450 were adopted for two periods of sedimentation by Pullar (1962). Erosional debris derived from the present hill country was deposited as a delta and later as a flood-plain to form the Poverty Bay flats . Using buried soils and tephra marker beds Pullar and Penhale (1970) described the infilling of the Gisbome Plains. From the time of the Taupo pumice eruptions (AD131) to c.AD1450 the rate of erosion of the hills adjacent to Gisbome City was low, and negligible infilling of the Gisbome Plains from c.AD1450 to c.AD1650 allowed the formation of Waiherere soils. Catastrophic erosion about AD 1650 caused widespread infilling over the Plains and in particular in the Te Arai Valley, and since 1932 infilling of the Waipaoa River meander trough has been very high. The accelerated rate of erosion since the settlement of the area by Europeans, in the late 19th century, has in times of flood events threatened the actual survival of the current settlers. 6 As early as 1896 Sir James Hector, founder of the New Zealand Institute, warned that the elimination of the East Coast forest cover would result in widespread erosion. His concerns, and those of others went unheeded as property rights were exercised in the clearance of native vegetation for pastoral use. Without its protective cover the crushed and faulted landforms became exposed to the effects of a punishing climate. Hot, dry summers reduce vegetative cover, shrink and crack the surface predisposing it to the effects of periodic tropical cyclones which often bring high winds and very heavy rainfalls. These saturate the ground resulting in erosion which delivers vast quantities of debris into streams and rivers where it may accumulate or be carried downstream and deposited on fertile flat land, shorelines and the sea floor at the rivers' mouths. Concern for the severity of erosion in New Zealand, especially in the East Coast region, has created a history of erosion control and land use policies and research projects, Table 2.1. The effects of some land management practices on the lives of others have never been so seriously considered as they have in recent years. Storms since 1960, especially Cyclone Bola in 1988, have graphically demonstrated their effects. This, and the resultant requests for financial assistance by the land users, in a period of low economic wealth for all New Zealanders fuelled a debate on the need for wise and sustainable land use practices that resulted in the enactment of the Resource Management Act, 1991. Under this Act, local authorities are required to manage the use, development and protection of natural and physical resources in a way that not only satisfies present requirements but also ensures their sustainability to meet the reasonably foreseeable needs of future generations. This chapter briefly outlines land use development and its effects in the East Coast. Some of the policies designed to assist development and mitigate erosion are also described. Late 1800s­ early 1900s late 1800 1941 1944 1948 1952 1960 1963 1967 1967 1970 1975 1976 1978 1982 1984 1984/85 1985 1988 1989 1990 1991 1992 Table 2.1 DEVELOPMENT AND POLICIES AFFECTING THE EAST COAST Native forest and bush cleared for pasture Warnings of the effects of deforestation Govt. geologists advised Waipaoa riverbed had risen 2m in 15 yrs as a result of deforestation Soil Conservation and Rivers Control Act administered by the Soil Conservation and Rivers Control Council (SCRCC) Poverty Bay Catchment Board formed Flooding and large-scale soil erosion -Waipaoa River peak flow 3500 cumecs SCRCC adopt modified American Land Use Capability Classification for soil and water conservation plans, catchment control schemes and regional planning. Exotic forest plantings began at Mangatu SCRCC set up a technical committee of inquiry into conservation problems Resultant report titled "Wise Land Use and Community Development" - referred to as the "Taylor Report" or the "East Coast Project" First Land Use Capability surveys undertaken Co-ordinating committee set up to administer the East Coast Project released a district plan for the years 1970- 1975 Land Use Capability Worksheets completed East Coast Catchment Board (ECCB) set up a local study group to reconsider the proposals in the Taylor Report ECCB produced "Report of Land Use P1aruting and Development Study for the Erosion Prone Land of the East Coast Region" - known as the "Red Report" Protection/Production Forestry Grants introduced End of Forestry Grant applications New Zealand Forest Service disbanded, abolishment of government subsidies Ngatapa storm Cyclone Bola- Waipaoa R. peak flow 5300 cumecs East Coast Project Conservation Forestry Scheme Reorganisation of regional authorities results in dissolution of catchment boards, duties maintained by regional/district councils New Zealand Resource Management Act East Coast Forestry Project History of the development and land use policies of the East Coast 7 8 2.2 SETTLEMENT AND LAND CLEARANCE Settlement of the East Coast by Europeans began near Gisborne (originally known as Turanga). This township was originally a whaling station because of the ease with which the small whaling ships could be navigated into the mouth of the Turanganui River. The whaling station later became a trading centre and supply base for the hill country stations, and is still the only large township in the region. The native vegetation on the East Coast prior to European settlement followed a pattern determined primarily by altitude, soils and drainage. Fern and scrub associations existed on soils derived from pumice on the lower hills adjacent to the Poverty Bay flats with manuka (Leptospermum scoparium) present on areas of deeper pumice. Light bush dominated by titoki (A/ectryon exce/sus), puriri (Vitex /ucens), kowhai (Sophora tetraptera), and karaka (Corynocarpus /aevigatus) extended up the moister valleys; while heavy bush of podocarp and broadleafed species dominated the higher country. At the time of settlement heavy bush of podocarp and broadleaf types existed in the Te Arai catchment. The main timber types were kahikatea (Podocarpus dacrydioides) and pukatea (Laurelia novae-zelandiae) with to tara (Podocarpus totara, P. hallii) and rimu (Dacrydium cupressinum) dotted about between tawa (Beilschmiedia tawa) and other species (Hamilton and Kelman, 1952). Early settlers describe this heavy bush as a 'black bush' and recall that the occurrence of 'hard red totara' trees was considered to indicate good soil. The luxuriance of the bush from a pioneer's point ofview indicated the fertility ofthe area. Early agriculture was established on the fertile flat lands with maize and wheat, and by 1858 there was a large export trade in ryegrass seed to Sydney out of the port at Gisborne. The potential of the region for sheep farming was soon realised and hill country settlement began in the 1870s. The population grew from 500 in 1879 to 2,300 in 1886 with large areas of land being 'broken in' by the burning of native vegetation followed by sowing with such grasses as ryegrass, cocksfoot and white clover. The development oflarge agricultural units proved very productive, with a high carrying capacity on fertile East Coast soils. 9 2.3 ACCELERATED EROSION Within 30 to 40 years of clearing the native bush and forests, erosion in the East Coast became the most severe in the country; severe even by world standards (Cumberland, 1944; Campbell, 1946; Bishop, 1968). A marked change in the appearance of the country after heavy rainfall and subsequent flooding during December 1893 and January 1894 was noted by Hill (1896) with " ... open and improved country seeming to have suffered most and bush country least." And once started, it accelerated. General opinion is that there was a gradual increase in the area affected by erosion up to 1936, and since then the rate of erosion and the area affected has increased rapidly (Hamilton and Kelman, 1952). The problems associated with erosion not only affected the farmers in loss of soil and those nutrients contained within it (a legacy from the cleared native vegetation) leading to reduced carrying capacity of the land, but were also to the detriment of the settlers downstream. Large slumps, earth flows and other mass-movements appeared damaging roads and buildings; streams and rivers flooded during heavy rains endangering stock and people on the flats and in the townships. Silts were deposited on fertile flat lands and infilled the port threatening the future of shipping. Periodic flood events have covered considerable areas of the flats (Pullar, 1962). 2.4 CONFRONTING THE EROSION The Soil Conservation and Rivers Control Act, 1941 was passed " ... to make provision for the conservation of soil resources and for the prevention of damage by erosion, and to make better provision with respect to the protection of property from damage by floods." The Catchment Boards established under this Act recruited river engineers and trained soil conservators to implement schemes for the control of flooding and erosion. The Soil Conservation and Rivers Control Council also introduced a system of erosion control based on Land Use Capability surveys and farm plans which consisted principally of producing a map showing the land use capabilities, or limitations, of various parts of a farm. Erosion control measures were then integrated into management and financial plans for the whole property, often with financial assistance from the council for conservation measures. 10 In 1944, the Poverty Bay Catchment Board (one of the first Catchment Boards formed) was set up with one of its prime responsibilities being the stabilisation of the headwaters of the area, thereby safeguarding the alluvial flats of Poverty Bay. In 1948, the Board commenced a large-scale trial of counter-erosion afforestation which attempted to establish a tree cover of conifers on those catchment areas above eroding gullies in the upper Waipaoa catchment. These trials clearly showed that once a canopy had formed, runoff and erosion in the gully below were considerably reduced. The Waipaoa River, with its headwaters in the Raukumara Ranges, and draining 216,700 hectares is the main river of the Poverty Bay flats . Fallowing large floods in 1948 and 1950 a flood control scheme was sought for the Waipaoa River to protect the flats, the only substantial area of flat land in the region, and Gisbome city from further flooding. The scheme consisted of three river diversions involving a channel shortening of three kilometres, together with a system of parallel stopbanks designed to accommodate a '100 year' flood flow of 4,400 cumecs (the 1948 flood peak flow having been estimated as 3,900 cumecs). These measures would safely contain the flood waters of consequent storms including the 1985 Ngatapa storm which had a peak flow of 4,800 cumecs. During Cyclone Bola, in 1988, there were some minor overflows, and in many places only 50 rnillimetres of freeboard . The river had a peak flow of 5,300 cumecs, 43% greater than the 1948 flow and 20% greater than the predicted '1 00 year' peak flow. In contrast the normal summer flow of the river is 10 cumecs (Singleton eta!. , 1989). The Catchment Board realised that flood control measures alone could not succeed without controlling the land degradation and erosion upriver. In 1955, at the request of the Board, the Soil Conservation and Rivers Control Council set up a special committee to report on remedial measures. Based partly on the success of the tree planting trial referred to above, it recommended that the worst of the eroded areas in the Waipaoa catchment, about 2,500 hectares, should be afforested. This was to become the basis ofMangatu Forest, for which planting began in 1960 (Allsop, 1973). 11 For approximately twenty years the Catchment Board co-operated with fanners to implement conservation works on farms principally involving tree planting, damming and drainage. However, by 1963, recognising that large-scale erosion control was beyond their resources the Catchment Board issued a report that this could only be controlled by massive reafforestation, which led to the Taylor Enquiry. 2.5 THE TAYLOR REPORT In July 1963, the Soil Conservation and Rivers Control Council set up a Technical Committee of Enquiry to investigate the conservation problems of Poverty Bay - East Coast districts and to make recommendations on a comprehensive control programme. This report was published under the title 'Wise Land Use and Community Development' (SCRCC, 1967) but is colloquially referred to as the 'Taylor Report' after the chairman of the Committee, N.H. Taylor, a founding Director of N .Z. Soil Bureau. The Committee recognised the exceptionally severe land erosion and associated social and economic problems of the region which they felt required urgent large scale remedies. The costs of dealing with the erosion problems were of such magnitude that the farming community alone could not economically support them. Two zones were recognised and separated by a "blue line", one containing the worst erosion and highest potential for further erosion - termed the "critical headwaters area"; and the other being the balance of the district - referred to as the "pastoral foreland" . 1. The "critical headwaters area" of 140,000 hectares, consisted of the most eroded and erodible land in the headwaters of streams and rivers underlain mostly by crushed argillite and greywacke rocks. 2. The "pastoral foreland area" of 485,000 hectares, comprised the easier less eroded land underlain by mudstone, sandy mudstone and argillite. The Committee recommended that in order to control accelerated erosion and to ensure maximum enduring productivity from the Poverty Bay - East Coast District the "critical headwaters" be recognised as land needing the protection of a forest cover and that this area be progressively acquired and planted with dual purpose forests designed 12 to provide effective erosiOn control combined with maximum productivity. The Committee also recommended that the Poverty Bay Catchment Board's policy of farmer co-operation be continued in the "pastoral foreland" and that the soil conservation programs be expanded with more prominence being given to the treatment of whole catchments using preventative erosion control work and conservation farm plans. The whole of the Te Arai River catchment area fell within the pastoral foreland. 2.6 THE EAST COAST PROJECT, 1970 A Co-ordinating Committee was set up as a result of the Taylor Report. In 1970 it released a district plan for the years 1970-1975, called the East Coast Project. It recommended that 7,000 hectares of new forest be planted in the critical headwaters area by the 1973174 planting season along with increase in annual planting in State forests in the East Coast. The Taylor Report envisaged economic revival for the area with employment in forest planting, road building, port development and tourism resulting in expansion of the townships. However, land acquisition and forest planting was slow and not until 1973 was the planting target of2,000 hectares a year exceeded. There was anxiety from farmers behind the "Blue Line" who felt they would be forced off their land. There was also criticism of the "no compromise" stance of the Report in delineating the "Blue Line" because they felt that the potential of much of the land behind the "Blue Line" had been underestimated. 2.7 RED REPORT It became appreciated that the simple zoning between the Pastoral Foreland and the Critical Headwaters Area (the "Blue Line") was too general. In 1976 an interdepartmental committee was formed by the Poverty Bay Catchment Board to: - investigate modifications required for implementing the recommendations of the Taylor Report; - recommend best long-term use of different areas of land in three categories; large scale afforestation, farm scale afforestation and farming. - make recommendations on planning and administration policies and practices that should apply at a regional level so as to reasonably achieve a desired pattern of land use within 50 years. 13 The committee produced "The Report of Land Use Planning and Development Study for Erosion Prone Land of the East Coast Region, 1978" (PBCB, 1978). Informally referred to as the "Red Report", this report recommended that the principles of the Taylor Report be re-affirmed and it redefined areas requiring soil conservation works based on the capability of the land to sustain long term use. It developed four categories referred to as the Gisbome District Council Erosion Categories which describe sustainable land use capabilities (Table 2.2). Category 1 Category 2 Category 3 Category 4 Table 2.2 Gisborne District Council Erosion Categories Recommended Use (a) Arable Farming (b) Arable & Pastoral Farming (a) Conservation Fanning & Farm Scale Forestry (b) Conservation Fanning, Farm Scale and Large Scale Forestry on some areas (a) Large Scale Production Forest (low priority for protection) (b) Large Scale & Farm Scale Protection\Production Forest (moderate priority for protection) (c) Large Scale Protection\ Production Forest (high priority for protection) Protection Forest Classes II & III Classes IV & VI Units VIle 1, 2, 5, 7 & VIIw Units VIle 3, 4, 6, 8 19 & 21 Units VIle 9, 10, 11 & 17 Units VIle 12, 14, 20 Units VIle 13, 15, 18 Class VIII Description of Gisborne District Council Erosion Categories 14 These categories have been used in successive district plans to advise land use policies. The Report recognised that the Land Resource Inventory and Land Use Capability information prepared by the Ministry of Works and Development (NW ASCO, 1979) was valuable resource information but that it was more suited to indicative regional planning, while more detailed LUC surveys were necessary for farm scale land use planning. The Report also recommended that the Poverty Bay Catchment Board increase its annual programme of farm scale land use capability surveys; that the Board vigorously promote conservation farming methods, predominantly on category 1 and 2 land, afforestation in association with other conservation works in small areas of category 3 and appropriate category 2 land; and encourage the private sector to adhere to recommended land use policies and where appropriate, be given financial support to cany out afforestation of category 3 land. The Red Report concluded: 'The East Coast is one of the few areas in New Zealand where there need be no competition with pastoral and forestry interests. There is ample scope for development in farming and forestry. A balanced regional land use and development policy will achieve this." 2.8 PRODUCTION FOREST DEVELOPMENT Mangatu Forest was initiated to stabilise the severely eroding land of the Mangatu station. Planting began in 1960 and by 1970 about a million trees per year were being planted which had a significant effect on the labour force in the area (predominantly Maori). Within five years of the beginning of the Mangatu project the population within close proximity had increased by 44 percent. In all, a total of 36,000 hectares of the more severely eroded hill country was planted at Mangatu, other state forests and farm conservation works. In the mid-1970's the East Coast Project was abandoned because, even with harvesting, it did not promise a financial return. During the 1970's and 1980's afforestation by the New Zealand Forest Service continued along with forest plantings and on-farm conservation works by the East Coast Catchment Board, but it slowed after tax deductions for forest planting were disallowed from 1984, and after the Forest Service was disbanded in 1986/87. However, the impact of Cyclone 15 Bola, in 1988, would show the physical value of forests foretold by so many people in the past, and that forests should not be valued by financial return alone. It was particularly noticeable that Mangatu forest escaped with only minor damage. During the 1980's the economy of the East Coast plummeted. The revenue from farms was low as a result of droughts, low wool prices and the removal of subsidies for fertiliser and farm improvements. This and the cessation of forest planting was having a detrimental effect on employment and economic welfare in the region. Then in March, 1988 the East Coast region experienced a greater than 1 00 year return storm. 2.9 CYCLONE BOLA In early March, Cyclone Bola formed in the Pacific Ocean and moved across the Fijian Islands, west to Vanuatu, south to Raoul Island in the Kermadecs, and then southwestwards to New Zealand. Just north of New Zealand the Cyclone met an eastward-moving anticyclone in the Tasman Sea and came to a virtual halt. At 12 noon on Monday, March 7th, the centre of Cyclone Bola was 200 kms northwest of North Cape, trapped between the anticyclone and another anticyclone forming to the east. The cyclone, heavily laden with moisture from the humid tropics, began to release its load onto Northland, the Coromandel Peninsula and the East Coast of the North Island resulting in torrential rain and storm-force winds as it moved southwest towards the west coast around Taranaki. The Hawkes Bay and East Coast catchment authorities studied the weather forecast, checked rainfall and river level measuring systems and began issuing flood warnings. In the East Coast region, over the 96 hour period from 6th to 9th of March, it rained almost continuously. Up to 900mm of rain was recorded in the higher catchment inland of Tokomaru Bay towards the Raukumara Range, 600mm was common, with 500mm widespread. These figures were the highest rainfall recorded since records began in 1876. The estimated return period of this event was greater than 100 years (B Turnpenny, G.D.C. pers comm.). On average, 5,000 tonnes of rain fell on every 16 hectare of the region. In the Te Arai catchment the total rainfall during the storm was approximately one-third ofthe average annual rainfall, Figure 2.1. A state of civil defence emergency was announced on the March 7th and lasted until the March 25th. Whilst the Waipaoa River Flood Control Scheme contained most of the flood waters from its catchment, Cyclone Bola caused widespread landsliding and flooding over an area of650 Ian2 ofthe Raukumara Peninsula (Phillips eta!,. 1990), and 12,000 Ian2 of the North Island (Trotter eta!. , 1989). Every road in the region was affected to some degree, many suffering severe damage. State highways between Gisborne and Wairoa, Waikaremoana and Opotiki were closed due to slips, washouts, siltation or the collapse of bridges. The railway line was impassable due to slips and an 80 meter dropout on the western approach to the Waipaoa River bridge, 10 kilometres from Gisborne. There were widespread losses to telephone and power lines. The water pipeline which runs through the Te Arai River valley and supplies Gisborne city with domestic water was blocked by debris and a large section was washed out by the swollen Te Arai River. This resulted in 10% of the city having no water for 3 days. Storrnwaters inundated the sewerage system and a small number of houses in lower lying areas were flooded . Some residents in farming districts were isolated in flooded houses by road failures and slipped tracks, and were without adequate domestic water for 5 to 6 days. Severe erosion occurred in hill country, especially the Tauwhareparaeffutamoe area inland from Talaga Bay, the hill country area south and west of Poverty Bay and the Tutira area of the Hawkes Bay. Some hill country properties lost up to 30% of their grazing area, but about 70% of damaged properties lost between 5% and 10%. Although between 60% and 80% of the damaged area has recovered quite quickly, the more severely eroded areas will take decades to recover (Singleton et a!., 1989). Landslide scars may recover 70-80% of the production of remnant forest soils over the first 20-40 years but there is no further recovery up to 100 years (Trustrum et a!. , 1990). A study in the Waihora River catchment, revealed an increase in erosion C}CI~ Bo lo Ra i....foll ElXB a. R'ld:l Dal:.o (""") A_..~ ltr-N.Ja I R.a i .-.fa I I Met . S«-v i oa D<:ri. a (nn) Figure 2.1 The distribution of rainfall during Cyclone Bola in the Te Arai River catchment compared with the average annual rainfall. 17 18 severity of 72% over the 1978 records (ECCB & R WB, 1988). This was a cause for concern because that level of erosion had taken 120 years since forest clearance to develop, (T. Freeman, G.D.C.perscomm. 1992). Widespread erosion on steep hill surfaces resulted in the inundation of farmland on river terraces and flats by silts with severe siltation occurring to some extent on virtually all valley floors in the region, particularly the Hikuwai Valley. Siltation resulted in a loss of depth in Gisborne harbour with an estimated 75,000 cubic meters of infill. This restricted loading levels of ships with logs, meat, kiwifruit and maize. Dredging was required with an estimated cost of$468,000. Cyclone Bola caused in excess of$NZ120 million damage to the North Island (Trotter et a/., 1989). A survey of one large catchment in the northern East Coast estimated that for every $1 loss in on-site production sustained from Cyclone Bola, a further $1.34 was spent on repairs to farm assets. There was also $0.76 of downstream production loss and $0.23 of repairs to downstream assets (Trustrum and Blaschke, 1992). 2.10 EROSION CONTINUES The region has about 240,000 hectares of hill country of which 140,000 hectares is zoned as erosion category 2 land, having "moderate to severe" erosion. The remaining 100,000 hectares is zoned category 3 erosion land, having "extreme" erosion. Cyclone Bola-induced surface slipping on category 2 land destroyed much grazing and looked appalling yet it is the predominantly mudstone, category 3 land that has received the most attention by those concerned with the erosion problem. Cyclone Bola served to remind us of the dynamic nature of the land surface. The land is continually increasing in height above sea level every year with uplift of 3mm/yr uplift along the Raukumara Range and 1 mm/yr closer to the coast including the study area (Pillans, 1986). Deformational rates for the region, the Hikurangi subduction system, comprise plate convergence of 50mm/yr; with 40rnrnlyr contraction and 30mm/yr strike-slip motion (Walcott, 1978). This is resulting in a highly deformed landscape prone to erosional forces. Streams and rivers downcut with greater 19 gravitational power each year, soil productivity decreases each year as rains wash off the nutrient-containing topsoil leaving slow to heal patches on all slopes, yet still the land remains unclothed, unprotected. Erosion on the hill country of the East Coast is occurring constantly, Bola simply made it more obvious by condensing five years erosion into one (Trevor Freeman, G.D.C. pers comm., 1993). 2.11 LAND USE INCENTIVES 2.11.1 Agricultural Incentives Prior to 1985 an array of Government-funded incentives were available to companies and producers aimed at increasing agricultural production. Reasons for high assistance for the agriculture sector were twofold . Prices for major agricultural products, particularly meat, had been weak for a number of years and many domestic industries which provided the materials needed for agricultural production were heavily protected against competition from imports resulting in high costs to the farmer. A Supplementary Minimum Prices scheme (SMP) was introduced by the Government in 1978. It was designed to create greater long-term confidence in the profitability of pastoral farming by establishing new minimum prices for meat, wool and milkfat for the coming season, and providing 'support payments' to cover any short fall in market returns. In 1982, 1983 and 1984 $382m, $382m and $295m were spent on the SMP scheme in the East Coast region. Between 1971 and 1985, $247m was also spent by the Government on fertiliser and lime subsidies and applications in the region, $23m was spent on a land development interest subsidy and $48m on a livestock incentive scheme (New Zealand Statistics Yearbooks) . These incentives did not take into account the suitability of the land for agricultural practices but encouraged the clearance of native vegetation and exposure of unstable surfaces resulting in severe and often large-scale erosion, for example the Mangatu slip. N .Z. Statistics Department information shows a reduction on livestock numbers in the years following the Ngatapa and Bola storms (1985 and 1988). This would seem to 20 reinforce the opinion of many, that the land cannot sustain pastoral use. Farmers in the region however, prefer to place the emphasis for lowered stock numbers on a falling meat and wool market and the cessation of subsidies. 2.11.2 Forestry Incentives Under the Forestry Encouragement Act, 1962, the Government made loans to local authorities towards the cost of establishing and tending new plantations and the maintenance of those already in existence. In 1970, funding was extended to private landholders in the form of a grant equal to 50 percent of the qualifying costs (up to $200 per year) of establishing and developing new approved forests. This grant was an alternative to the tax concession already available to income-earning forest companies. Exotic forest plantings in New Zealand have increased steadily since 1950 (New Zealand Statistics Yearbooks) . Plantings in the East Coast region increased steadily from 10,000 hectares in 1971 to 86,000 hectares in 1990. Since 1975 private tree plantings have been greater than plantings by the State. 2.12 EAST COAST PROJECT CONSERVATION FORESTRY SCHEME, 1990. Since Bola, several schemes have been initiated by the Government following public and local body pressure to redefine long term sustainable land use in the region. The East Coast Project Conservation Forestry Scheme, administered in 1989 by the East Coast Catchment Board and subsequently, since local government amalgamation, by the Conservation Division of the Gisborne District Council, targeted 15,000 hectares of hill country with severe to extreme or potentially severe soil erosion problems. The Scheme applied to land upstream of Gisborne, the Poverty Bay flats and the Tolaga Bay flats . It offered a 95% subsidy to farmers to establish conservation forest. Central government would meet 66% of the cost ($7 million spread over five years) and 29% would be met by district ratepayers. 21 The Catchment Board estimated that 104,000 hectares of land in the East Coast region required conservation forestry and set a five year target for the planting of 13,600 hectares specifically the more severely eroded areas in the "critical headwaters" . Initially landowner participation was reluctant. Why commit scarce resources to a project with no guarantee that one can harvest the trees in the future? There was no comprehensive land use plan to indicate which land required a permanent vegetation cover and which land should be zoned as suitable for forest which can be harvested without compromising the down-stream economy. To date a total of 9,000 hectares of conservation forest has been established with 67 percent ofthis on land with severe erosion (GDC categories 2 and 3), and 12 percent on land with moderate erosion which under certain circumstances can readily achieve severe status. The balance being comprised of the adjoining land up to stable fencelines. The Scheme has been successful in reafforesting land with severe erosional problems. 2.13 NGATI POROU FORESTS LTD Another government initiative was specifically directed towards the multiple-owned land in the north. It offered government loans at favourable rates to companies willing to participate in joint ventures with the land owners. Ngati Porou Forests Ltd planted 600 hectares offorest in 1989 and continues planting today. However, in doing so they are removing mature manuka and kanuka which some contend is contrary to the objectives of the reafforestation strategies for the region. The removal of natural vegetation over 2 metres on Class VI or Class VII land in preparation for planting is subject to a Gisbome District Council resource consent under the Vegetation Removal and Earthworks Regional Plan. 2.14 EAST COAST FORESTRY PROJECT, 1993 In 1992, the Government announced a new forestry scheme with multiple objectives of commercial forestry, conservation forestry and regional employment. The Scheme, to be administered by the Ministry of Forestry, aims for the Government to fund the planting of predominantly radiata pine (Pinus radiata) on a total of 200,000 hectares 22 of the Gisborne/East Coast region over the next 28 years. The project targets NZLRI Class VII (GDC categories 2 and 3), land and therefore is designed to plant a mix of severely and moderately eroding or erodible land but excludes areas of natural forest. Clearance of established or substantially regenerated natural forest is counter to the project's objectives. The project places a restriction on the clearance of natural vegetation for forest planting funded under this project, 11 ••• No line cutting, gap clearing or total clearing of any indigenous tree species will be approved under the scheme. 11 Emerging indigenous tree species are defined within the scheme as those over 50 em in height at more than 50 stems per hectare that have reached or will reach 30 em in diameter. Areas that do not meet GDC resource consent criteria on vegetation removal are not eligible for grant approval. Although a successful tender is reliant upon a number of factors, a property with more than 60% of Class VII land is qualified to tender. To meet the projects conservation objectives of increased employment opportunites and financial return to the economy, high initial stocking rates are considered necessary. The units within Class VII are grouped according to a required initial stocking rate to qualify for tender- Group A, requiring a minimum initial stocking rate of 1250 stems per hectare, includes units VIle I, 2, 4, 6, 8, 9, 10, 11, 17, 19 & 21. Group B, which has a minimum initial stocking rate of 1500 stems per hectare, inc/udesunitsVI!e3, 5, 7,12,13,14,15,16,18 & 20. Preference for funding is given to those tenders with a more intensive planting regime (more Group A land) (MOF, undated) . This scheme has already come under serious criticism as it is seen as a subsidy to forestry companies who are purchasing the better of the eligible land and receiving government funding for normal operational costs resulting in higher profits. As the world demand for timber increases it is argued that production forestry will of its own accord become more widespread in the region and that Government initiatives should be targeted more directly at the most erodible land; i.e. that which contributes most to 23 sediment loading of streams and is not desired by forestry companies because of its difficult terrain. 24 CHAPTER III DESCRIPTION OF THE STUDY AREA 3.1 FEATURES OF THE STUDY AREA The Te Arai River catchment occupies an area of 19,000 ha, located approximately 25 kilometres south east of Gisborne, Figure 1.1 . Altitudes in the catchment range from 30m along the river flats to 718m at Te Rimuomaru in the south and 708m at Parikanapa in the west. There are small areas (c. 1 0% of the catchment area) of flattish land along the river, which expands towards the mouth of the catchment, that represent infilling of previous downcutting episodes within the physiography. Most of the catchment is composed of moderately steep to very steep slopes. About 65% of the land is of go to 250 slope and 13% has a slope of between 26° and 350_ The catchment is drained by the Te Arai River and its tributaries which flow into the Waipaoa River 4km inland from its mouth. The catchment has been settled by European fanners since the mid-1800's. Extensive bush-clearing for conversion to pasture occurred around the turn of the century and until 1994 the catchment was predominantly in pastoral vegetation except for small (less than 100 ha) areas of conservation and production forestry, and an 1,100 ha water supply bush reserve at the head of the catchment. Intensive production on the river flats downstream includes viticulture, citrus orchards and market gardening. Te Arai station, 10,691 acres, was the first property to be developed in the catchment. Fanned by Charles Westrup from 1867, it was later leased by John Clark ofOpou and settled by ballot in 1908. Emerald Hills station (Figure 1.2) has been fanned by the Parker family since 1898. In 1994, Emerald Hills and Y -wury sheep and cattle stations were sold in 1994 to Forest Enterprises Ltd. Planting of Pinus radiata on these properties, about 3,500ha within the catchment, commenced in the same year. 25 3.2 PHYSIOGRAPHY 3.2.1 Tectonic Setting The study area lies within the East Coast Fold Belt, east of the Wairoa syncline. Much of this region has been subjected to uplift of about 3rnmlyr (Pillans, 1986) during the last two million years (Quaternary). Broad synclinal and anticlinal folds have been faulted and refolded to tighter folds along a north to northeast trend. This structural deformation characterises the Fold Belt and determines the rock outcrop patterns. The anticlines are asymmetrical; with the longer and shallower limbs dipping to the west. Lithologies of the East Coast region are predominantly Cretaceous greywackes and argillites, and Tertiary sandstones, mudstones, argillaceous limestones and bentonitic clay shales. During the Cenozoic tectonic movements resulted in highly incompetent Cretaceous bentonitic deposits and claystones being intensely crushed. More porous but stronger Tertiary sandstones and mudstones lie above these crushed rocks. Quaternary tectonic movements, eustatic sea levels changes and periodic tephra showers have subsequently influenced the shaping of the present landscape. 3.2.2 Geological Structure of the Te Arai River Catchment The catchment extent is controlled by the western axis of Mangaone-Waingake uplift in the west; by the Rerepe anticline in the south and east; and by the Waerenga-0-Kuri fault complex in the north, Figure 3 .I. The Mangaone-W aingake uplift is a large compound anticline trending roughly NE-SW from the north end of the Morere anticlinorium of Hawkes Bay to the Waimata River valley where it terminates against a large cross fault ; the Waimata Valley Fault (Brown, 1961). This western axis of the Mangaone-Waingake anticline forms the fault controlled eastern flank of the Wairoa Basin. The Rerepe anticline is an extension of a structural platform which comes off the northwest flank of the Morere anticline. Between the Mangaone-Waingake anticline and the Rerepe anticline the surface dips to the Waingake-Te Arai synclinal. The Te Arai River roughly follows its direction. The syncline is broad, low dipping and simple, in contrast to the anticlines which are steeply dipping, asymmetric and faulted . This asymmetry fits the regional pattern of steeper eastern flanks and long low western flanks. The latter are well defined by sandstone strike ridges and dip slopes. 26 Figure 3. 1 Geologic structure of the Te Arai River catchment, from Brown ( 196 1) 3.2.3 Lithologies of the Te Arai River Catchment All the lithologies of the study area are marine sedimentary rocks of Tertiary age. By far the most predominant rock type is mudstone, either crushed, jointed or banded with thin layers of sandstone. There are some diapirs containing large blocks of Eocene 27 bentonites in the Waingake region. The high ridges which constitute the west and south boundaries of the catchment have been mapped as massive sandstone on the NZLRI, but as banded sandstone by Francis eta/. (1989). They were confirmed as the latter by this author. East of the western margin of the catchment lies a belt of alternating beds of sandstone and mudstone interbedded with thick-bedded sandstone and tuffaceous sandstone. Small areas of light-coloured, fine-grained, calcareous rocks occur in small areas. Along the levees and terraces of the river and its tributaries is Holocene alluvium. A field survey oflithologies and dip slopes was carried out on Emerald Hills station for familiarisation and to confirm the work of Francis eta/. (1989). Surface lithologies are also included in the inventory on the GDC soil and water conservation farm plan for Emerald Hills maps, at a scale of 1:10,000. The field mapping on the farm plan is based on any available outcrops or exposures and on the overall "look" of the land. The difference that the type of mapping has on the interpretation of lithology is shown in Figure 3 .2 Short range changes in the state of the mudstone, overlooked by geological mapping, are delineated on the farm plan. The Te Arai syncline trending NNE-SSW is clearly obvious within a band of crushed mudstone east of the Te Arai River. The jointed mudstone to the east of the syncline is tilted up towards the east (the Rerepe anticline) (Figure 3.3). West of the syncline banded sandstone and mudstone is tilted upwards towards the Waingake anticline, Figure 3.4. Faults were noted in bands of crushed mudstone also to the west of the syncline (Figure 3.5). Geology mapped by field survey Roelm mappd on GDC farm plan m :wot. ,....;.,., au~~. 1Juldlld 0 :kilt, 1Bxuiad 0 * · jtCn.W 0 ltrl. pin1.od """' .l.tb.rrium ~ lloi., pin1.od Gd .Ult O~rinlllllll'li ~ ltrl. """""""- I1Dl ioah 0 Se\ b!IIJdod. fm Topiln Scale 1 : 000011 Figure 3.2 The geology of Emerald Hills station mapped from field survey compared with the lithology information available on the farm plan. 28 Figure 3.3 Figure 3.4 The Te Arai syncline in c rushed mudstone with jointed mudstone on the right tilted up towards the cast To the west of the Te Arai syncline the banded mudstone and sandstone is tilted up towards the west 29 30 Figure 3.5 Faults arc observed in bands of crushed mudstone 3.2.4 Tephra cover in the Te Arai River catchment Although originally tephras from the Taupo and Okataina Volcanic Centres were deposited over the entire region, they have been removed or reworked by wind and rain erosion. The Taupo, Waimihia, Rotoma, Waiohau, Mangaoni and Rotoehu tephras were deposited in the region, Table 3 . I. In the region, the Taupo, Waimihia, Rotoma and Waiohau are the soil forming tephras which occur within the top 45cm of the soil surface. The tephras, buried paleosols and the soils derived from these are discussed by Pullar eta!., ( 1973). Taupo ash and pumice or Waimihia formation are the most likely soil forming parent material which would result in yellow brown pumice and yellow brown loam respectively. A study carried out by Veld and De Graaf ( 1989) recognised 31 two rhyolitic ash layers on Emerald Hills and a neighbouring station believed to be from the Taupo Volcanic Zone. They identified them as the Taupo pumice which was found overlying the Waiohau ash or more often depositied on colluvium or bedrock. Tephras deposited in the Gisborne region Tephra Taupo ashes + pumice Waimihia formation Rotoma ash Waiohau ash Mangaoni lapilli Rotoehu ash Source Taupo Vole. Centre " " " Okataina Vole. Centre M M M " " M H H Date 1,800 B.P 3,400 M 7,300 " 11,200 " 30,000 M >41,000 M Figure 3.1 Tephras deposited in the Gisbome Region, from Pullar (1973). This study found that three tephras are recognisable on Emerald Hills station. Taupo fine to coarse ash and pumice of up to 15mm in diameter occurs mixed in the top soil; dark orangey, brown Waimihia lapilli underlies the topsoil and overlies colluvium, weathered bedrock or the light yellow cream Waiohau ash deposit (Figure 3 .6). These deposits do not occur in a consistent manner. On the higher narrower ridges to the west of the farm, exposed to the prevailing south­ westerly winds there is a little Taupo ash and pumice mixed in the top soil and in many places no tephra at all. Exposures show small pockets of W aimihia beneath the top soil and overlying weathered bedrock. This is a result of infilling of small indentations in the microtopography at the time of the air fall. On the eastern shoulders of these ridges the Waimihia lapilli is found to be more extensive in cover although not consistent in depth or occurrrence. The depth of this deposit may be due to overthickening by erosion from the ridges. On broader ridges of lower elevation to the east of the farm all three tephras may still be found . On the broadest of these ridges, 1 0-20cm of Ah horizon contains Taupo ash and pumice overlying 10-30cm of Waimihia lapilli. Between the Waimihia lapilli and the weathered bedrock (usually mudstone) there may be up to 50cm of light yellow/brown to white tephric material. It was thought to be the Waiohau tephra from 32 previous descriptions of field inspections. Ferromagnesian mineralogy and electron microprobe analyses indicated that this is Waiohau ash, although the presence of a small amount of cummingtonite in some of the samples analysed suggested that there is a mixed tephra of Waiohau and Rotoma (Appendix). There is no distinct evidence of the Rotoma tephra. It is described as a thin deposit in Gisborne by Vucetich and Pullar (1964), with a sharp boundary between it and the underlying Waiohau buried soil. Possibly, the Rotoma is present in small amounts which cannot be visually distinquished from the Waiohau ash. Erosional processes following the deposition of the Waiohau may have removed any developing soil so that the Rotoma was deposited onto the Waiohau, and intennixed by pedogenic processes, or it was stripped from the landscape by erosional processes during the 4,000 yrs after its deposit and before the Waimihia airfall. On the ridges shoulders there is often a layer of colluvium, of varying thickness, beneath the Waimihia lapilli and above the Waiohau ash (Figure 3.6). This material has been redeposited from upslope. Once again, this indicates that there was a period of slope instability sometime between 11,200 and 3.400 yrs. B.P. Further down the slopes of broad ridges the tephras thin and in most places the Waiohau, or the Waiohau and Waimihia, are not present. There are some places on the shoulders where pockets of Waimihia or Waiohau occur directly beneath the Ah horizon. On less broad ridges the tephras are also thinner and often only the Taupo, or the Taupo and Waimihia, are present. Between the ridges the landscape is deeply incised by drainage channels. Associated slips and earthflows have cut into the slopes to within metres of the ridges. In these areas an occasional pocket of Waimihia can be found but nonnally the only tephric evidence is the Taupo pumice mixed in the Ah horizon. On these slopes the Ah horizon is very thin, or missing in sites of recent erosion. 33 The most revealing exposures on Emerald Hills station were found on the shoulders of stable ridges not exposed to the strong south-westerlies. The following sequence was found to be common in these sites : Ah!Taupo pumice Waimihia fapilli Waiohau ash 10-20 em 10-30 em 10-30 em Cw (bedrock/collu vium) Figure 3.6 ----Taupo pumice ----Waimihia lapilli ----Waiohau ash Three tephras are found on the broader ridges and shoulders. The Waiohau ash overlying Mudstone indicates that no soil parent material in the area is older than 11 ,200 years. The colluvium between the Waiohau and the Waimihia infers the instability of that period. 34 On the 2,900ha property there are few areas where the tephras are preserved in their true sequence. A field survey revealed that all three tephras are found in only a few stable sites on the broader parts of the ridges delineated in Figure 3. 7. Large areas mapped as Mj I Kt or Kt /Mj Uointed mudstone and ash) on the farm plan, are usually the more stable slopes where Taupo pumice occurs mixed in the Ah horizon. Figure 3. 7 Areas on Emerald Hills station where all three tephras may be found. 3.2.5 Soils in the Tc Anti River catchment The NZLRI worksheets have soil information based on the General Survey of Soils of the North Island (N.Z. Soil Bureau, 1954) which was mapped at l :250,000 scale. The soils are described as: Manawatu silt loams along the river fl ats ofthe Te Arai River, Mahoenui silt loams on the mudstone slopes to the west of the river with Hangaroa sandy loam interspersed in parts (presumably where the mudstones are interbedded with sandstones) , and Wharerata sandy learns on the steep banded sandstones ofthe western boundary ridge. To the east ofthe river are Turakina silt loams, Pakarae sandy loams and Wharerata and Waihua sandy loams on the western and southern boundary ridges . 35 A good description of the soils of the Gisbome Plains which includes the flood plains of the Te Arai River mapped at 1:15,840 was presented by Pullar (1962). In contrast to the NZLRI legend he established local soil series and mapped the T e Arai River flood plain soils as Matawhero and Waihirere silt loams with Makauri clay loam occurring in subsidiary drainage valleys. Gisbome District Council soil and water conservation farm plans do not include soils information because they are not considered as important in their relationship to land use and land stability as the underlying rock types. 3.3 CLIMATE The climate of the Gisbome region is classed as mild, with a large number of sunshine hours and low mean wind speed. Its position at the easternmost tip of the North Island often results in differing weather conditions from those current elsewhere. Meteorological conditions over the ocean to the east of New Zealand sometimes affect the Gisbome region alone. Temperatures of the area are mild. At elevations less than 500m the annual range of mean temperatures lies between 70C and 2QOC; although temperatures greater than 240C can occur on average for 65 days per year and exceed 3QOC on six days, during foehn winds. At higher altitudes temperature variation is greater and there is more likelihood of frost in winter. The distribution of rainfall in the region is influenced by the physiography. The Poverty Bay flats and their extensions are protected by adjacent hills and receive a lower rainfall than the hill country where rainfall increases from south to north and from east to west. Rainfall isohyets (New Zealand Met. Service, 1973) show coastal yearly normals are about l,OOOmm compared with 1,400mm at Waerenga-0-Kuri (314m a.s.l.) to the north of the catchment, 1,800mm at Parikanapa (708m a.s.l.) on the western boundary, and 1,400mm at Waingake (60m a.s.l.) located in the centre of the catchment. For any one month the variability of rainfall from year to year is high compared with most parts of New Zealand and this is most pronounced from 36 December to April. This variability of rainfall combined with the drying effects of the northerly winds leads to frequent moisture deficiency for plant growth not only in the drier lowlands but also on the eroded hills where topsoil is thin. Two extremes of rainfall occurrence are common in the east coast. Dry periods in the wannest part of the year sometimes cause a very serious reduction in the amount of pasture available for stock. The other extreme is the flood-producing rainfalls of which there are two types. First, there are the short period high intensity storms usually associated with intense cold front conditions. These storms produce local centres of torrential rain affecting relatively narrow belts in their path resulting in flood conditions in only one or two watersheds. Secondly, there is the more frequent occurrence of a generally high rainfall over a much larger area. This situation exists under cyclonic conditions when the centre of low pressure is situated to the east, north-east or north of the area (refer to section 2. 9). When the depression moves slowly or remains stationary the continuous rain saturates the catchment resulting in very high run-off figures . 3.4 EROSION IN THE TE ARAI RIVER CATCHMENT Throughout the catchment the physiography reflects historic erosional processes. Large areas have been altered by progressive earthflows, Figure 3.8. Gully erosion is evident along drainage channels and is serious in some places especially in the crush zones associated with faults and the syncline. Poplar and willow plantings, and debris dams protected some, but not all, gullies during Cyclone Bola, Figure 3.9. Stream bank erosion occurs along the Te Arai River and its 3 major tributaries, the Waimata, Kauwaewak.a and Waingak.e Streams. Where these erosional processes remove the toes of slopes, slumping often occurs. The erosion form most obvious in the catchment is the soil slip and sheet erosion induced by Cyclone Bola in 1988, Figure 3 . I 0. 37 Figure 3.9 Gully infill ing following Cyclone Bola 38 Figure 3. 10 Soil slip erosion following Cyclone Bola Slip scars were estimated to be approximately 5% of the land area (C.M. Trotter, Landcare Research, Palmerston orth. pen; com.) . The erosion inflicted by Cyclone Bola appeared to be restricted to pasture land with little evidence of disturbance in the bush reserve or pine blocks, except for deeply incised gullies. That the catchment has undergone considerable erosion since about 11,000 years ago is evident by the lack of Ohakean terraces along the river and gravels and by the absence of tephras known to have been deposited in the area since that time. A layer of gravels exposed in a high bank of the Te Arai River bank about 30m above the present river level indicates the level of the valley floor during the Ohakean. The gravels are presumed to be of Ohakean age because of the overlying Waiohau tephra. 39 CHAPTER IV DATA COLLECTION 4.1 INTRODUCTION An objective of the project was to investigate the availability of existing suitable data and to utilize this data wherever possible. Thematic layers of information were collected, or developed, and stored as coverages in the vector structured GIS database ofPC ARC/INFO v3 .4D. This chapter will begin by discussing how the information for this research was collected, including field surveys and remotely sensed data. Next, a section will describe the processing of the data collected in preparation for inclusion in the GIS database. This ranges from the construction of a digital elevation (DEM) from which the topographic variables; slope angle and slope aspect are derived, to the geometric co-alignment of all layers of information. 4.2 EXISTING DATA SETS The New Zealand Land Resource Inventory data pertaining to the catchment was purchased from the Landcare Research in digital format. It included information relating to rock type, soil, slope angle, erosion type and severity, vegetation and a land use capability assessment published originally at a scale of 1:63,360 (1 inch to 1 mile). This information was obtained on diskette and loaded directly into the ARC/INFO database. The data was coordinated to the latitude and longitude geometric system and it was necessary to transform it to the New Zealand Metric Grid system using a transformation program purchased from the Department of Survey and Land Information (DOSLI), Wellington. Land resource information was also available at a larger scale (1 : 1 0,000) for Emerald Hills station in the form of a Gisbome District Council soil and water conservation farm plan. This resource inventory was prepared by soil conservators from the District 40 Council as part of a comprehensive conservation plan for the farm. Areas with consistent resource characteristics are delineated on aerial photographs along with a description of those characteristics. This information is denoted in a consistent manner with the NZLRI. The information from the farm plan for Emerald Hills station was collated into the database by digitizing the resource inventory boundaries, developing a coverage, then typing the relational information of LUC, rock type, slope angle, erosion types and degree and vegetation, into the associated database table. 4.3 FIELD SURVEYS The distribution of tephra on Emerald Hills station was mapped by field investigation and delineation on aerial photographs. During this field investigation the areas mapped on the farm plan as having a tephra cover were investigated. The extent of the tephra cover was noted and the slope angle of the land mapped within those LUC units containing tephra was measured by abney level. Erosion on the farm was inspected and verified with that evident on the farm plan or aerial photographs. Sample areas of soil slip were inspected more closely for confirming the accuracy of the image classification. A survey of dip slopes, geologic features and lithologies on the station enabled the confirmation ofthe geology ofEmerald Hills station as mapped by previous workers in the Te Arai catchment. 4.4 TOPOGRAPHIC DATA Digital 20 metre contour information for the catchment was purchased from DOSLI. To obtain slope and aspect information a digital elevation model was created by a triangulated irregular network model (TIN) using the digital contour information. The development of a TIN for the entire catchment was beyond the memory capability of the 486 PC computer being used for this project. Therefore, a TIN was created for Emerald Hills station. To reduce the size of the resultant TIN and ARC/INFO coverages of slope and aspect classes, weed tolerances for vertices along the contour and z (altitude) values were set to 40m and 20m respectively. A proximal tolerance of 20m was also stipulated for triangle labels (centres). The TIN was made up of 43,026 41 triangles and comprised 2mb. Slope angle and slope aspect are derived from a TIN coverage by the ARC/INFO, 'TIN to polygon' , conversion facility. In the process of converting each triangle to a polygon the maximum rate of change in z values across the polygons is calculated (as percent or degrees) . Also, the compass direction of the maximum rate of descent across the polygon is calculated and expressed in degrees from 0 to 360. Triangles having a percent-slope ofless than 1, or a degree-slope of less than 34, are considered to be flat and, having no aspect, are given an aspect value of- 1. Lookup tables are used during the 'TIN to polygon' conversion to group the slopes and aspects into classes. The slope angle values were grouped into slope angle classes of5 degree intervals (0-5, 5-10 ...... >40 degrees) . Slope aspects were grouped into the eight octants normally used in aspect representation (N, NE, E, SE, S, SW, W, NW). The slope angle (Figure 4.1) and slope aspect (Figure 4.2) coverages derived from the TIN were 3mb and 2mb respectively. Figure 4.1 ~ gJope ClaS1le8 (degrees) tll tl (o-3) 0 B (4-7) 0 c (8-15) 0 D (16~0) 0 E (21-~} 0 F (2s-35) ~ G (over 55) Scale 1 : 4{1000 Slope angles on Emerald Hills stat1on derived from 20m digital contour data .J:>. N 43 § ~ -=.:.z:- r-1 q, ~ ro n; 1;/J "0 ..... ::J 0 c 0 u 2 '§I "0 E 0 N E 0 ~ "0 ~ ~ Qj "0 c 0 ~ "' ~ I ::2 ro Qj E w c 0 "' u ~ a. "' ro ~ a. 0 U5 44 4.5 REMOTELY SENSED INFORMATION 4.5.1 Satellite Imagery A panchromatic SPOT (Systeme Pour !'Observation de Ia Terre) satellite image of the entire catchment, acquired on March 26 1988, was obtained from the Landcare Research in EPIC format, on six floppy diskettes; each subimage being 1132 rows and 1024 pixels per rows. The panchromatic sensors on the SPOT satellite consist of a high-resolution-visible (HR. V) imaging system designed to operate with a 1 0 metre resolution over the wavelength range 0.51 to 0. 73 urn. Because of the large amount of radiometric data relating to the study area and the limited screen capabilities of the image analysis programs available, the smaller area of Emerald Hills station was selected for analysis. Using the DRAGON image analysis package data was selected from four of the diskettes and combined to provide a single image of the farm (Figure 5. 1). 4.5.2 Aerial Photographs A set of aerial photographs were purchased from Aerial Surveys Ltd, Nelson. Following a decision to use Emerald Hills station as a window in which to investigate erosion during Cyclone Bola eleven photographs were acquired in order to have full coverage of the farm. They had an endlap and sidelap of 30 percent. These black and white photographs with a format of 23 em by 23 em were acquired at a scale of 1:27,500 on March 27, 1988. This was within three weeks of Cyclone Bola (March 6- 9) and therefore the soil slips were clearly identifiable on the photographs by the characteristics of tone, shape and texture (Fig 5.4a). Digital information from the aerial photographs was obtained by scanning them on a Microtek Scan Maker liSP flat bed scanner using the Adobe Photoshop (ver. 2.5) program operating in Windows. The information was stored in Tiff format which was a compatible format for importing into the DRAGON image analysis program. To minimize the geometric inaccuracy resulting from tilt and relief displacement, as little as possible of the outer regions of each photograph was used to obtain coverage of the farm. 4.6 SPATIAL CORRELATION OF DATA FROM VARIOUS SOURCES 45 All data layers were registered to the New Zealand Metric Grid to facilitate overlay analysis with ground control points (GCPs) determined by digitizer from the NZMS topographic maps, 1:50,000 scale. The base GCP coverage had a positional error RMS of 2.4m, the features used for geometric registration had a horizontal map accuracy of within 15m (DOSLI). While a polynomial affine transformation was used for all geometric registrations carried out on PC ARC/INFO, it could not correct the error in the Emerald Hills station farm plan. Because the farm plan information had been delineated on enlarged mosaiced aerial photographs there was considerable relief and processing distortion. 'Rubbersheeting' was used to correct the geometric error by aligning the farm plan with GCPs identified on the NZ metric grid. During the transformation process the features of the coverage being rubbersheeted are moved using a piecemeal transformation that preserves straight lines. It is very powerful in stretching the coverage therefore a large number of GCPs are required to ensure the accuracy of the resultant coverage. Accuracy was ascertained by overlaying the arcs of the rubbersheeted coverage on the registered satellite image which was known to be geometrically consistent with the contour and farm boundary coverages. The registration of the aerial and satellite imagery was carried out in IDRISI and is discussed in more detail in the next chapter. 46 CHAPTER V DIGITAL IMAGE ANALYSIS 5.1 INTRODUCTION The monitoring of the state of the natural resources under the administration of a local authority has become a statutory requirement since the enactment of the RMA, 1991 . While many studies of erosion have been carried out in New Zealand using manual methods of aerial photograph interpretation and ground survey to delineate erosional features (Trustrum and Stephens, 1978; Crozier et a/., 1980; Stephens et a/. , 1981 ; Trustrum and De Rose, 1988), more recently, computerised image analysis (Benny and Stephens, 1985; Trotter et a/. , 1989; Pain and Stephens, 1990; Wilde, 1992) and GIS techniques have been utilised to provide a rapid, cost-effective method of assessing the extent of degradation of the land resource. This chapter investigates the feasibility of utilizing a simple image analysis software package and a PC computer to delineate, quickly and relatively cheaply, soil slip scars and convert that information from raster to vector data format for use in the GIS . The ability ofFrench SPOT panchromatic satellite imagery (lOrn resolution) and 1:27,500 black and white aerial photography, to assist in the delineation of bare ground, 1s compared. To determine the extent of soil slip erosion on the pastoral land following Cyclone Bola, to investigate the relationships between these eroded areas and other physical characteristics of the land and, to evaluate these areas for land use suitability, all the eroded areas needed to be delineated and incorporated in the ARC/INFO database. Because of the large amount of radiometric data which would need to be processed to study the entire catchment only a window, Emerald Hills station, was analyzed. 47 5.2 THEORY AND PREVIOUS RESEARCH Air photograph interpretation has been an integral part of land surveys since the 1950s. Typically, erosion features have been mapped by visual interpretation and manual delineation on aerial photographs along with field survey (NW ASCO, 1979; Crozier et a/., 1980; Phillips eta/., 1990; Veld and de Graf, 1990). Many of these studies have used time sequential photographs to study changes in the extent of erosion (O'Loughlin, 1969; James, 1973; Trustrum and Stephens, 1981 ). More recently, computerised systems have been available and these enable automated classification of surface features from remotely sensed information (Stephens, 1985). Digital image analysis involves the storage, retrieval, interactive processing and display of digital images using computerised systems. Digital data may be obtained either directly from satellite and airborne scanners, or indirectly from aerial photographs. The latter may be accomplished by scanning the aerial photograph at a set resolution (pixel size) and recording the radiometric values of each pixel in digital format. Where the information from more than one radiometric band is used, the range of reflectance values pertaining to a feature is termed its 'spectral signature'. Many earth surface features of interest can be identified, mapped and studied on the basis of their spectral characteristics (Lillesand and Kiefer, 1987). However, this study used only black and white imagery and the pixel values describe their reflectance. The advantage of digital image analysis over manual delineation of surface features is that, once the reflectance values of the surface features under study are identified, an automatic classification of the data can be carried out very quickly. The results of this classification are usually portrayed as a colour coded image along with a table showing the area of each representative class. The name remote sensing was first coined in 1960 (Fischer, 1975) and referred to the observation and measurement of an object without touching it. By the late 1960s photographic methods were being used regularly and investigations using thermal infrared and microwave sensors onboard aircraft and cameras onboard satellites were being reported in the literature (Curran, 1985). From the launching of the first Earth 48 Resources Technology satellite (later named Landsat 1) in 1972 (Fischer eta/. 1976) digital image data became more widely available for land remote sensing applications (Lillesand and Kiefer, 1987). In New Zealand, the EPIC software was developed in the early 1980s allowing the use of digital land resource data (McDonnell, 1986). Digital image analysis has become a well recognised method of investigating erosion in New Zealand and elsewhere. In a storm damage assessment using digitized aerial photographs Pain and Stephens (1990) found that digital methods gave objective and rapidly obtained results. They concluded that the digital methods used in the study were superior to other objective methods of point sampling aerial photographs because oftheir ability to provide areal measurements as well as a map of the spatial location of landslides. Black and white aerial photographs have been extensively used because they are the most readily available at low cost and are easy to handle. Whilst these photographs allow the boundaries of erosion to be readily delineated, other types of photographs (e.g. colour and colour infrared) are especially valuable in displaying moisture, drainage, and vegetation conditions. Colour infrared film was found to be more useful for land use and erosion mapping than conventional film types such as natural colour, and black and white (Stephens eta/. 1984). It has been used extensively for land resource mapping, for example in land use studies (Stephens eta/. 1984) and for vegetation and erosion mapping (Birnie et a/, 1982). It was found to be the most suitable film type for detecting old landslide scars revegetated with pasture (Trustrum and Stephens, 1981) although, black and white aerial photography was found to be as good as multispectral data for the assessment of erosion type and extent except where only the topsoil was eroded (Cuff and Trustrum, 1983). The use of remote sensing for mapping bare ground and vegetation was described by Benny and Stephens (1985). They found that topography had a significant effect on the spectral signature. When only surface features on sunny or flat sites were selected it was found that almost all features had unique spectral signatures. However, variation in sun angle across an image (and therefore brightness value), in particularly in hill and steepland, results in overlap in the spectral ranges for different surface features making distinct definition of the features difficult. 49 Another problem existing with digitized aerial photographs is the spatial distortion which occurs towards the edge of the photograph due to relief displacement and radial distortion. This reduces the spatial accuracy of analyses when information derived from aerial photographs is overlayed on other geographically registered information. While most image analysis systems provide polynomial mapping facilities (i.e. affine, quadratic or cubic transformations) to correct geometric distortions, relief displacement can only be corrected by differential rectification (Dymond, 1991 ). Differential rectification of aerial (and satellite) imagery results in digital orthophotography which can be used in digital image analysis or GIS . In a GIS it can be used as an accurate base on which map overlays containing an array of pertinent information can be accurately registered. Apart from providing all the information of a photograph, digital orthophotographs also facilitate direct measurements such as lengths and areas interactively on the computer screen, profiling of the ground surface quickly and accurately, enhanced feature extraction (using image analysis systems), change detection, and efficient and accurate GIS updating (Dall, 1991 ). Digital image analysis of satellite digital data has also been used in the mapping of areas of soil erosion. The earliest available satellite imagery was the Landsat MSS data with a ground resolution of 79m and four spectral bands: green (0.5-0.6um), red (0.6- 0.7um), reflective infrared (0.7-0.8um), and thermal infrared (0.8-1.1um). However, some have found the MSS data has limited application potential for erosion mapping because of its insufficient resolution and its unavailability over large areas (Sauchyn and Trench, 1978; Gupta and Joshi, 1990). Landsat Thematic mapper (TM) data has been available since 1982 with higher spectral (seven bands) and spatial resolution (120m for the thermal and 30m for the rest). Another form of high resolution satellite data, SPOT with a ground resolution of 20m for the three multispectral bands and 1Om for the panchromatic (black and white) band became available in 1986. Both panchromatic (PAN) and multispectral linear array (XS) images of SPOT data were successfully utilised in the generation of an erosion map in a severely-eroded area in South China (Gao and Luk, 1989). In New Zealand, Trotter eta/., (1989) found the use of SPOT satellite imagery to provide quantitative data on landslide damage, on a 50 farm-by-farm basis, had its difficulties. This was mainly due to image mis-calculation resulting from the high degree of variation in radiance from landslides in steep hill country, exacerbated by the low sun angle at the time of image acquisition. The typical topographic variation within the area covered by one farm meant that it was necessary to classify the image as a series of small segments, essentially at the hillslope scale. In the MacKenzie Basin, satellite imagery provided a rapid and cost effective method for quantifying the extent of land degradation, due in part to intensive grazing by rabbits (Wilde eta/., 1991). By ground survey they established a relationship between percentage vegetation cover and normalised vegetation index (NVI=[IR-RIIR+R]) of SPOT XS data. Results indicated that soil types have a strong influence on the areal pattern of degradation with areas of more than 80% cover having good moisture­ holding capacity. Digital aerial and SPOT imagery is being utilised in an ongoing study of the change in bare ground in an area of sand dunes and plains within the Manawatu coastal sand country (Wilde, 1992). NOAA-A VHRR imagery has become the prime source of remotely sensed data for monitoring land cover in the context of global change, because of its near global coverage, frequent data acquisition schedule, and its proven ability to detect land cover changes (Millington, 1991 ). However, its resolution (1 .11an) does not assist detailed land cover mapping. Airborne multispectral scanner data is proving to be very useful for the detection of erosional features . They provide enhanced spatial and spectral resolution data over that which may be achieved from satellite data (Agar, 1991). In New Zealand, the Bay of Plenty Regional Council has utilized airborne remote sensing to build a database of its environmental resources. A Daedalus MSS was fitted to a Gates Lear Jet to obtain imagery of 30m resolution. 51 5.3 DETERMINATION OF CYCLONE BOLA SOIL SLIPS 5.3.1 Cyclone Bola soil slips derived from Satellite Imagery Initial statistics of the imagery showed that brightness values fell between 12 and 136 resulting in a dark image. This is due to the low solar illumination associated with the time of day and the season (1 0 am, March 26th). A linear contrast stretch image enhancement procedure was applied to the image in order to expand, uniformly, the range of pixel values to 0- 255 . The greater contrast in the image assisted subsequent visual detennination of ground control points (GCPs), and brightness range setting for each land cover. Because DRAGON's registration module had a bug at the time the registration was carried out in IDRISI. The satellite image was registered (geometrically corrected) to the NZ metric map grid, Figure 5 . I, by selecting GCPs which were clearly identifiable on both the image and the NZMS 260 map series (1 :50,000) from which eastings and northings were read. These GCPs were road intersections, field boundaries, stream confluences or ridge peaks as suggested suitable by Benny, 1983. IDRISI uses a linear polynomial fit and a nearest neighbour resampling method (Richards, 1986) to transform the image geometrically. The functional relationship between image X and Y, and map easting and northing is determined by a least squares regression. The transformation error for the satellite image was 3.06 pixels, 30.6m. By overlaying contour lines and by running a cursor over the registered image checking the geometric coordinates with the topographic map it was detennined that the registration was very successful. An accurate transformation of an image covering this area of land was expected to be difficult when it is understood that one pixel must be selected to represent the GCP, a subjective process; that a pixel in the SPOT image represents a 1Om by 1Om surface area; and IDRISI uses the bottom comer of the cell for the transformation process. Minor transformation error was due to relief displacement although of much less scale than on the aerial photography because of the far greater elevation of the sensors. The registered image was converted to the DRAGON format (using a TIFF intermediary) and then classified. Unsupervised (clustering) and supervised 52 (parallelepiped) classification procedures were carried out with the parallelepiped (boxcar) classification being the most successful at discriminating bare ground from other surface cover because of the ease with which the spectral range could be manipulated. The fundamental difference between these two techniques is that in the clustering process the image data is classified by aggregating the pixels into natural spectral groupings present in the image, while supervised classifications involve a training step where the analyst identifies pixels in the image which represent the features under study. A set of statistics (signatures) is then assembled which describes the spectral response pattern for each land cover that is being studied on the image; each spectral class represents one information class which can be discriminated by the classifiers. For a successful classification of the image into classes representing each land cover, the signatures of each class should not overlap each other. Representative areas (training areas) were delineated on the image for the two classes; bare ground (representing slips from Cyclone Bola) and pasture/trees. An attempt was made to classify for trees, pasture and bare ground. However, when the image was classified for the three land cover classes the resultant image showed that trees and pasture could not be discriminated. This was due partly to the amount of shading on the image and also to the characteristics of the trees themselves. The trees that were of most interest were space planted poplars and willows which were used for conservation works in eroding areas. These had similar reflectance values to pasture. Being autumn the tree leaf colour would be subdued and it is possible the poplars were affected by rust. Also, where there were only one or two small trees occupying a pixel, which represented 1Om by 1Om of ground area, the pixel would record the dominant reflectance character which was pasture. 53 Figure 5.1 Satellite image of Emerald Hills sta tion, registered to the NZ metric grid. A parallelepiped classification was chosen as the mtmmum distance to mean and maximum likelihood methods are not suited to a single band of information. The classification process was a very simple discrimination of two spectral classes; the lighter and the darker, with the distinction being such that the classes accurately represented the surface characteristics of bare ground or vegetation. The classification procedure involved adjusting the range of reflectance values in the feature signature file until it resulted in a distinct delineation of the land cover classes. This is often referred to as density slicing. The classified image was compared with the original image and the signature ranges adjusted until a satisfactory classification was obtained, Figures 5.2 (a) & (b). 54 Figure 5.2(a) Sample area from satellite image Figure 5.2(b) Satellite image (a) with areas classified as bare ground overlaid Because of DRAGON screen limitations the satellite imagery was processed as four -- subimages. To retain attribute information during conversion from raster to vector data systems the images were transferred to EPPL 7 (Environmental Planning and 55 Programming Language) in TIFF format then imported to ARC/INFO using the ARC/INFO GRIDPOL Y facility. In ARC/INFO the image was clipped to the boundary ofEmerald Hills station. 4.7% ofthe station had been classified as bare ground. This decreased to 4.5% after editing to remove areas of river bed erosion, roads and farm tracks, Figure 5.5. 5.3.2 Cyclone Bola soil slips derived from Aerial Photography Firstly, an area from a 1:27,500 black and white aerial photograph was scanned at lOrn, 5m and lm resolution to determine whether there was great advantage in scanning at higher resolution. The higher the resolution the greater the amount of data generated for processing and the more likely the need to subsample the imagery to accommodate the image data and screen limitations of the software. It was expected that the higher resolution (lm) would facilitate the selection of intensity value ranges and that the classification would result in smaller slips being identified. However, there was no change in the spectral limits assigned to each class from the 5m or 1m resolution images because the spectral infonnation was from the same source image. Nor was there any change in the percentage of bare ground classified on each of the three images, Figure 5. 3. The classified features appeared smoother on the 1m resolution image but any advantage in delineating smaller slips was limited by the reflectance variation in the image. Therefore, the classification of the aerial photographs was carried out on five photographs covering Emerald Hills station scanned at 70 dots per