Modeling streambank erosion on composite streambanks on a watershed scale

Streambanks can be a significant source of sediment and phosphorus to aquatic ecosystems. Although the streambank-erosion routine in the Soil and Water Assessment Tool (SWAT) has improved in recent versions, the recently developed routine in SWAT 2012 has undergone limited testing, and the lack of site or watershed specific streambank data increases the uncertainty in the streambank-erosion predictions. There were two primary objectives of this research: (1) modify and test the 2012 SWAT streambank-erosion routine on composite streambanks, and (2) compare SWAT default and field-measured channel parameters and assess their influence on predicted streambank erosion. Three modifications were made to the SWAT 2012 streambank-erosion routine: (1) replacing the empirical effective shear stress equation with a process-based equation, (2) replacing bankfull width and depth measurements with top width and streambank height, and (3) incorporating an area-adjustment factor to account for non-trapezoidal cross-sections. The proposed streambank-erosion routine was tested on gravel-dominated streambanks on the Barren Fork Creek in northeastern Oklahoma. The study used data from 28 cross-sectional surveys, including streambank height and top width, side slope, thickness and texture of streambank layers, and an area-adjustment factor. Gravel d50 and kd- τ c relationships were used to estimate the critical shear stress ( τ c) and the erodibility coefficient (kd), respectively. Incorporating the process-based shear stress equation, areaadjustment factor, or the top width and streambank height increased predicted streambank erosion by 85%, 31%, and - 30%, respectively. Incorporating the process-based effective shear stress equation, sinuosity, radius of curvature, and measured bed slope improved the predicted versus observed streambank erosion Nash-Sutcliffe efficiency from -0.33 to 0.49 and the coefficient of determination (R2) from 0.02 to 0.65 at the ten study sites. Although the process-based effective shear stress equation was the most influential modification, incorporating the top width, streambank height, and area-adjustmentfactor more accurately represented the measured irregular cross-sections.