Developing an improved, shock-capturing watershed model for simulating spatially variable runoff and soil erosion processes at the hillslope scale

<p>The overarching objective of this study was the development, validation and testing of an improved watershed modeling framework that accounts for the effects of spatial heterogeneity on overland flow and erosion processes and it is computationally sound for shallow, overland flows with shock waves. Most of the existing soil erosion models determine fluxes of water and sediment with the assumption of a homogeneous hill. In these models the physical and biogeochemical properties of the heterogeneous hill are spatially averaged, without considering roughness and longitudinal curvature effects as well as differences in the land use/land cover -LU/LC- and soil properties along the hill. This issue was addressed by improving the Water Erosion Prediction Project (WEPP-Original version 2010.1) soil erosion model at the hillslope scale to account for the physics in terms of spatial heterogeneity in flow using a well-established shock-capturing numerical scheme.</p><p>The improved WEPP model, referred to as "WEPP-Improved" model was (i) validated via detailed field experiments within an experimental plot and (ii) tested via generic simulations at the hillslope scale covering a variety of scenarios in terms of topography, LU/LC and soil type. Results showed that the WEPP-Improved model could effectively simulate the unsteadiness of the flow as well as the required time (lag) for the flow rate to reach equilibrium conditions. However, the model provided only a steady-state sediment transport rate and could capture only the equilibrium conditions. Further, the WEPP-Improved model reflected the effects of curvature, LU/LC and soil type on flow, as the model did not treat the hillslope as a homogeneous unit. Based on the generic simulations, landscape variability resulted to differences in the predicted peak runoff rate, Qpeak, between the WEPP-Improved vs. WEPP-Original models ranging ~ 3 - 62 % (avg. 19 %) due to curvature effects only, ~ 17 - 170 % (avg. ~ 66 %) due to added effects of LU/LC variability and ~ 5 % - 200 % (avg. ~ 52 %) due to added effects of soil type variability. The highest reported differences on the predicted Qpeak between the two models were attributed to the formation of the shock waves; these differences were dominant for the low in magnitude storm event and attenuated for the high event.</p><p>It is believed that if the physical processes are represented accurately at the hillslope scale using the suggested modeling framework, then by utilizing an appropriate routing scheme of the flow and sediment within the stream network, it will be possible to scale-up the flow/sediment routing from the hillslope to the watershed scale without losing the degree of heterogeneity encapsulated from different hillslopes within the drainage network.</p>