Predicting spatial patterns in riverbank erosion for catchment sediment budgets

Abstract

ABSTRACTRiverbank erosion is an important component of catchment sediment budget models but remains one of the least well-understood processes, particularly at large spatial scales. Here, we develop and test a new bank migration model in New Zealand for large catchment applications that (1) better represents spatial variability in factors influencing bank erosion and (2) improves predictive performance. We represent bank migration rates as a function of reach-scale stream power, channel sinuosity, soil texture, valley confinement, riparian woody vegetation and channel protection works. The new model significantly improves prediction compared to the SedNetNZ model. Comparison of measured bank migration rates with individual variables shows percent silt + clay derived from soil maps exhibited the strongest correlation, whereas other variables were non-significant. The model results demonstrate that improved prediction can be achieved by combining spatial representation of multiple factors over large areas, despite low correlation between individual variables and bank migration rates.