Modeling streambank erosion and failure along protected and unprotected composite streambanks

Streambank retreat can be a significant contributor to total sediment and nutrient loading to streams. Process-based bank stability models, such as the Bank Stability and Toe Erosion Model (BSTEM), have been used to determine critical factors affecting streambank erosion and failure such as riparian vegetation and to estimate retreat rates over time. BSTEM has been successfully applied on a number of cohesive streambanks, but less so on composite banks consisting of both cohesive and noncohesive soils in highly sinuous streams. Composite streambanks can exhibit rapid and episodic bank retreat. The objectives of this research were twofold: (i) develop and apply simplified procedures for estimating root cohesion based on above-and below-ground biomass estimates and (ii) systematically apply BSTEM to a series of 10 composite streambanks distributed along the Barren Fork Creek in eastern Oklahoma to assess model sensitivity to root cohesion and model performance in predicting retreat. This research aimed to document the influence of riparian conservation practices on bank retreat rates and evaluated simplistic methods for incorporating such practices into such process-based models. Sites modeled included historically unprotected sites with no riparian vegetation and historically protected sites with riparian vegetation present during all or part of the 2003 to 2010 study period. The lateral retreat ranged from 4.1 to 74.8 m across the 10 sites and was largest at the historically unprotected sites in which retreat averaged 49.2 m. Protected sites had less bank retreat but with more variability in retreat rates per year. With calibration focused on the erodibility parameters, the model was able to match both the observed total amount of retreat as well as the timing of retreat at both the protected and unprotected sites as derived from aerial imagery. During calibration BSTEM was not sensitive to the specific value of the soil cohesion or the additional soil cohesion added due to roots for the cohesive topsoil layer, suggesting that the proposed simplified techniques could be used to estimate root cohesion values. The BSTEM modeling also provided an advantageous assessment tool for evaluating retreat rates compared to in situ bank retreat measurements due to the magnitude and episodic nature of streambank erosion and failures. Process-based models, such as BSTEM, may be necessary to incrementally model bank retreat in order to quantify actual streambank retreat rates and understand mechanisms of failure for the design of stabilization projects. (C) 2015 Elsevier Ltd. All rights reserved.