Streambank erosion (M.Sc. thesis)
My M.Sc. research at the University of West Florida (UWF) centered around a U.S. Fish and Wildlife Service (USFWS) grant (no. 13058) to study streambank erosion in the Southeast U.S. with Dr. Johan Liebens (UWF) and Chris Metcalf (USFWS). This research resulted in 3 first-authored papers.
The first paper investigates the applicability of an existing empirical model for streambank erosion for the northern Gulf of Mexico Coastal Plain (NGMCP). The model, known as BANCS, has been widely adopted by the stream restoration community, but few papers evaluating BANCS have been published in the peer-reviewed literature. The objective of the USFWS grant was to calibrate the model for the NGMCP and to improve the model’s scientific validity or empirical fit based on local environmental factors. The paper builds on this work to show that BANCS had little to no correlation to observed erosion rates and reviews the fluvial geomorphology literature to suggest why this is so. The paper identifies the lack of a physics-based model for near-bank shear stress as a key weakness.
The second paper seeks to develop an empirical model for streambank erosion, to equip stream restoration practitioners with a model that is similar to BANCS in practicability and data requirements, but is based on established measurements rather than on visual estimations. Key physical and environmental measurements pertaining to streambank erosion were collected (e.g. biomass density, root density, soil shear strength, and soil composition). A two-tier statistical model selection approach was developed based on Akaike’s Information Criterion (AIC) and repeated cross-validation. The paper details the statistical model selection process and identifies three statistical models for streambank erosion in the NGMCP, each with a significant correlation to observed erosion rates (max. R² = 0.65 ).
The third paper presents a spatially-distributed, process-based framework for streambank erosion modeling and calibrates it for the NGMCP. The model is based on the “HIPS” equation (named for the coauthors of seminal papers), which assumes erosion rate proportional to near-bank velocity excess (i.e. perturbation above the bulk velocity). The proportionality constant is related to bank erodibility. The model uses a state-of-the-art hydrodynamic flow model to calculate near-bank velocity, and remotely-sensed tree coverage data to model bank erodibility. A monthly hydrological model (NLDAS-2 Noah 2.8) was integrated into the model to estimate monthly discharge. Using only 3 free parameters and requiring no field work, the calibrated model is significantly correlated (R = 0.55) to observed erosion rates. The model may be used in the future to predict the effects climate or land-use changes may have on streambank erosion rates. Furthermore the model is applicable anywhere on the planet, provided that stream channel geometry can be surveyed, and is available on GitHub.