Multi-scale morphodynamics of unconfined coarse-bedded rivers in the Ruamāhanga catchment : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Earth Science at Massey University, Palmerston North, New Zealand
Loading...
Files
Date
2023
DOI
Open Access Location
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Massey University
Rights
The Author
Abstract
Riverine floods are the most frequent global natural disaster and a fundamental behaviour of alluvial systems. Flood protection management tends to focus on constraining the spatial extent, frequency, and magnitude of inundation, but often neglects a second intrinsic alluvial behaviour: changes to the channels themselves. This omission can be particularly problematic for gravel-bed rivers, which are well-known for their change propensity across spatiotemporal scales due to complex sediment dynamics. Changes often vary asynchronously on timescales far greater than individual flow events with controls and processes that nest across spatial scales. Further, differential feedbacks mean that similar water discharge may result in two or more very different channel responses. This complex array of riverscape responses and biophysical feedbacks in space and time encapsulates riverscape dynamics, which lie at the heart of this thesis. The applied research of this thesis addresses important gaps regarding 1) suitability assessment of geospatial time-series data, 2) river avulsion hazard screening, 3) tectonic forcing of alluvial rivers, and 4) channel response to human forcing.
The gravel-bed rivers of New Zealand’s Ruamāhanga catchment provide an excellent real-world setting to explore these gaps. The short, steep rivers draining the high-relief Tararua Range experience frequent, intense rain-driven hydrology and transit the low-relief of an active forearc basin after emerging from the mountain front. The rivers cross numerous active geologic structures including the Wairarapa Fault, known for the world’s largest single-event land-based horizontal displacement (18.7 m) in 1855. Rivers draining the Tararua Range are known to have filled and expanded into the late-1860s as landslide sediments introduced to rivers during the 1855 shaking gradually worked downstream. Associated changes in river channel and floodplain forms would have generated more severe flood responses given a comparable pre-quake precipitation event. While the rivers eventually adjusted to a more relaxed state of response, their high intrinsic dynamism continues to challenge human habitation. With extensive riparian agriculture and 26% of all buildings occurring on Holocene alluvium, humans have exerted considerable control over the past sixty years using frequent (annual- to sub-annual) earth-moving to incrementally train multithreaded, wandering rivers into the narrower, straighter, and steeper forms seen today.
Confidence in localizing hazards and characterising river dynamics rests on certainty that the same locations is/are being compared through time and expressed as coregistration error for members of a time-series. I developed a numerical model to illustrate how root-mean-squared-error (RMSE) values, the customary form of error expression, from intercomparison of randomly-varied and systematically-biased datasets are substantially greater than RMSE values derived from comparison to a common reference. A case study from the Ruamāhanga catchment spanning a wide quality spectrum of archival aerial photomosaics (n = 5) indicates assumptions of residual normality are tenuous and RMSE consistently represents only 65-75% of error frequency and 30-36% of the maximum observed error magnitude. My results suggest that an empirically determined 95% confidence value is better suited to address simulated and real-world limitations. After reprocessing a subset of the historic data, some prior interpretations of channel movement were found erroneous. I propose a conceptual framework to aid fitness-for-purpose determination for different types of geospatial data quality, errors, and analyses to promote more suitable interpretations.
Globally, the direct observational record of major river relocations (avulsions) is highly limited given their infrequent occurrence on timescales from decades to millennia. Field- and model-based investigations over the last 45 years have identified a diverse array of contributing factors. These include sedimentation rates, conveyance capacities, and erodibility of existing channels relative to potential receiving areas, although to varying degrees across landscapes. My review identifies topographic advantage as a common denominator across landscape settings and develop a relative digital elevation model (rDEM) approach for rapid, landscape-scale screening in GIS. I propose revision of the traditional two-phase avulsion model to be consistent with other threshold phenomena (e.g., landslides) that divides factors of the set-up phase into static and dynamic components. I present a simple, dichotomous conceptual framework along a gradient of sensitivity (threshold proximity) to aid resource prioritisation for follow-on investigation and/or mitigation. Explicit inclusion of coseismic displacement adds novelty, particularly as adjacent streams appear to be out-of-phase.
Control of rivers by tectonic processes is traditionally investigated at millennial and orogen scales though some recent studies have explored annual and reach scales. Nonetheless, fluviotectonic dynamics operating between these timescales are relatively unexplored, especially regarding gravel bed rivers. I relate changes from a time-series of benchmarked cross-sections to surface deformation interpreted from a high-resolution DEM for a 16-kilometre study segment that crosses four active oblique strike-slip faults and several folds. Net and total bed change within and between cross-sections exhibit a high-degree of noise and lack reach-scale patterns. In contrast, patterns of total change accumulated over the time-series show strong spatial partitioning by intersecting geologic structures. The least dynamic cross-sections are generally in proximity to uplifted axes while the most dynamic cross-sections are generally downstream of such intersections and/or coincide with inferred back-tilting. This is the first analysis to suggest morphological forcing of an alluvial river by active geologic structures is detectable at decadal-scale and persistent during an interseismic period.
Human river management is a critically important control, but seldom addressed by research across spatiotemporal scales. I address this gap by evaluating a mix of short- and long-term records to assess congruence between two common management aims: increased channel stability and reduced active footprint. Active belt width at the riverscape scale (~16 km) interpreted from a 69-year aerial photo record shows decreasing trend (-48% mean) and increasing uniformity (-62% SD) that converges on the width of the contemporary design corridor (fairway). By contrast, ultrahigh-resolution, reach-scale morphological budgeting over a series of sub-annual events finds roughly sixfold greater volumetric changes in sub-reaches with recent in-channel flood protection earthworks than adjacent untreated reaches. Treated subreaches experienced up to 16 metres of lateral bank erosion, had more instances of increased activity outside the fairway, and propagated changes into untreated areas upstream and downstream. While management actions appear collectively successful in long-term constraint of the riverscape’s active belt, the same actions amplify bed changes from common flow events. I call this anti-pattern the “fairway paradox”. Increased magnitude and frequency of reach-scale movements over short time periods creates maintenance dependencies and makes river responses to flooding less certain. Such scale-dependent responses mean action-effectiveness can only be assessed if aims explicitly identify spatiotemporal scale. Greater certainty regarding a river’s location at any point in time may be gained by less frequent mechanical interventions. This study is the first rigorous evaluation of the effectiveness of NZ river fairway management with broad implications given the widespread application of similar river management throughout NZ.
Collectively, this thesis highlights importance of identifying different controls on river behaviour and the scales on which they operate. Substantive doubt is cast on broad application of theoretical or averaged design conditions over alluvial riverscapes with diverse and comingled controls. This is particularly important considering management has occurred under best-case conditions and a large sediment-generating earthquake is expected regionally every ~150 years. As the first work of its kind in NZ and possibly globally, this thesis provides a robust example for future multiscale investigations.
Description
Some Figures in this thesis may be copyrighted. Copyright holders are invited to use the Copyright Take Down Request link below to request their removal.
Keywords
Ruamahanga River Watershed (N.Z.), Fluvial geomorphology, River channels, New Zealand, Ruamahanga River Watershed