Finally, model testing was conducted to demonstrate the reliability of the HEC-RAS/BSTEM algorithms. Several test scenarios were constructed and modeled with HEC-RAS, the standalone version of BSTEM 5.4 and the standalone FORTRAN version of BSTEM used in the integration (which was subjected to rigorous independent validation against BSTEM 5.4 (Simon et al., 2010)). The before and after cross sections for a bank failure event are displayed in the figure below. The FORTRAN version of the algorithms in HEC-RAS replicates BSTEM 5.4 very closely. Small divergence can be explained by a couple algorithm differences between the FORTRAN version and BSTEM 5.4.
Gibson et al. (2015) also applied the model to Goodwin Creek, following the work of Langendoen and Simon (2008). Goodwin creek is a highly instrumented reach with substantial bank migration, carefully measured over a decade with dozens of repeated cross sections, making it an ideal site for evaluating a bank process model. Gibson et al. (2015) used the parameters from of Langendoen and Simon (2008), to test the model against a known calibration. The integrated HEC-RAS/USDA-ARS BSTEM model performed well compared to prototype data (bottom figure) and the CONCEPTS runs in Langendoen and Simon (2008).
