Development and testing of a physically based model of streambank erosion for coupling with a basin-scale hydrologic model SWAT

Authors: NARASIMHAN, B - Indian Institute Of Technology; ALLEN, P - Baylor University; COFFMAN, S - Tarrant Regional Water District; Arnold, Jeffrey; SRINIVASAN, R - Texas A&M University

Submitted to: Journal of the American Water Resources Association
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/7/2016
Publication Date: 4/1/2017
Citation: Narasimhan, B., Allen, P.M., Coffman, S.V., Arnold, J.G., Srinivasan, R. 2017. Development and testing of a physically based model of streambank erosion for coupling with a basin-scale hydrologic model SWAT. Journal of the American Water Resources Association. 53(2):344-364. https://doi.org/10.1111/1752-1688.12505.
DOI: https://doi.org/10.1111/1752-1688.12505

Interpretive Summary: In many watersheds across the U.S., stream bank erosion exceeds erosion from agriculture, forest, range, and urban lands. How we manage our streams becomes as important as how we manage our agricultural lands. In this study, a comprehensive stream bank erosion model was developed and incorporated in the Soil and Water Assessment Tool (SWAT) hydrological model. The stream bank erosion model was tested in the Cedar Creek watershed in north-central Texas where stream bank erosion rates are high. A rapid geomorphic assessment was performed in Cedar Creek using techniques developed by NRCS. Based on the field assessment, erosion monitoring was established at seven locations with severely eroding stream banks. The observed erosion rates varied from 25 mm per year to 367 mm per year. The model was calibrated and validated with daily streamflow and was shown to realistically predict stream bank erosion. The model developed in this study provides a tool that incorporates the impact and management of channels into a comprehensive framework with the management of cultivated agriculture, pastures and urban areas. It is critical for national conservation assessments that we realistically estimate sediment and chemical contributions from the landscape and from streams.

Technical Abstract: A comprehensive stream bank erosion model based on excess shear stress has been developed and incorporated in the hydrological model Soil and Water Assessment Tool (SWAT). It takes into account processes such as weathering, vegetative cover, and channel meanders to adjust critical and effective stresses while estimating the bank erosion. The stream bank erosion model was tested for performance in the Cedar Creek watershed in north-central Texas where stream bank erosion rates are high. A Rapid Geomorphic (RAP-M) field assessment of the Cedar Creek watershed was done adopting techniques developed by the NRCS (Windhorn, 2002), and the stream segments were categorized into various severity classes. Based on the RAP-M field assessment, erosion pin sites were established at seven locations within the severely eroding stream banks of the watershed. To develop a reasonable estimate of bank full width and depth, a survey of channel cross sections was carried out at a few locations across the basin and a power function relationship was established to extrapolate the bank full width and depth for simulation across the watershed. The daily stream flow predictions compared well with the observed flows both in terms of volumetric flow rate (R2: 0.5; E: 0.31) and flow depth. A Monte Carlo simulation was carried out to assess the sensitivity of different parameters that control stream bank erosion such as critical shear stress, erodibility, weathering depth, and weathering duration. The weighted parameters used for channel cover, based on the channel condition and local land use, were adjusted, and the model was calibrated based on the bank erosion severity category identified by the RAP-M field assessment. This was later verified using the erosion observed at the erosion pin sites. The observed erosion rates were in the range of 25 mm/year to 367 mm/year. The SWAT model was able to reasonably predict the bank erosion rate within the range of variability observed in the field, with mean erosion rates well predicted with the model (R2= 0.90; E = 0.78).