A very high-resolution analysis of the influence of bank roughness on the rate of river bank erosion processes

River bank erosion is a significant contributor of sediment to our rivers and a changing climate poses new challenges for our understanding of these erosion processes. This study made use of very-high resolution terrestrial laser scanning and meteorological and flow observation data to attempt to identify the relative contributions of different erosion processes on a stretch of the River Arrow in Warwickshire, UK. Over 24 months, five sections of river bank were scanned seven times each, creating 6 time periods of change for the analysis. It was possible to identify subaerial erosion as the dominant erosion processes across all five of the study
banks, contributing up to 98% of the erosion recorded. Through a series of linear regression models it was possible to identify maximum discharge, mean stage, maximum stage and peaks above the Q10 stage as statistically significant contributors to erosion, explaining 36.7% of the volume
of erosion per m2 per year across the whole bank face. The most interesting feature of these models was the direction of the model coefficients, with most of the flow variables exhibiting negative coefficients, suggesting that as flow increases erosion decreases. High flows were not generating
erosion and were inhibiting other erosional processes, in particular that of subaerial erosion which was the most significant contributor to bank change during the study period. Meteorological variables were also modelled
via linear regression and maximum temperature, total rainfall and average rainfall were identified as significant contributors to erosion, but those models still only explained 19.7% of the subaerial erosion above the Q10 level and 23.7% of the total erosion volume. In addition to linear
models, a principal components analysis was also carried out to try to explain more of the erosion. The PCA model explained an additional 7% of the erosion above the Q10 level using two components spanning the full range of meteorological variables calculated, with component one
comprising positive contributions from cold hours, frost days, freeze thaw cycles, total rainfall, wetting and drying cycles, wet days and rain hours and negative contributions from mean and minimum temperature. Component two comprised positive contributions from maximum temperature and hot hours but negative contributions from average rainfall. The final part of this study sought to identify whether roughness had a significant effect on fluvial erosion. Roughness was calculated at three different scales - 0.5m, 0.25m, and 0.03m - and the effect of roughness on erosion was modelled using a series of further linear regression models. These models explained between 3.2% and 89.3% of the erosion value, when local erosion and local roughness were controlled for. Again, the coefficient values in these models were interesting, with greater roughness leading
to greater erosion in the majority of cases for all three roughness scales. Interaction models, that measure the effect of multiple levels of independent variables on the relationship between another independent variable and the dependent variable, were undertaken to try to understand the
combined effect of roughness at different scales. Fewer of these models were statistically significant - only 16 out of 30 - but the significant models frequently demonstrated that an increase in the roughness at the 0.5m scale resulted in a weakening of the relationship between roughness at the
0.03m scale and erosion. However, there still remain some inconsistencies cross the interaction models that require further analysis. Overall, the research was deemed successful, shedding new light on the process
of bank erosion and identifying numerous opportunities for further research. In particular, more temporally dense measurements of erosion are needed to better understand the relationship between flow events and erosion, as well as attempting to respond more directly to high flow events
by scanning before and after high flows to more directly attribute erosion directly to specific flow events.