A framework for quantifying soil erosion and evaluating the effectiveness of best management practices with process-based modeling in southcentral Kansas

Abstract

The Kansas Water Plan identified sedimentation in federal reservoirs in Kansas as one of its highest priorities, which requires improvement in stream water quality. Rainfall-driven sheet and rill erosion in agricultural landscapes in Kansas remains a major environmental concern for its contribution to stream water quality degradation and reduction of crop yield. Yet, accurately quantifying the spatial variation of sheet and rill erosion at the field scale remains challenging.

Empirical models to calculate sheet and rill erosion, such as USLE/RUSLE and SWAT, follow a lumped area approach and have a limited spatial resolution. This study presents a framework for quantifying soil loss for sheet and rill erosion on 6027 representative hillslopes in the Running Turkey Watershed, situated in southcentral Kansas, using a process-based model known as the Water Erosion Prediction Project (WEPP). The 6027 hillslopes were identified through a systematic selection procedure by applying specific thresholds for different topographic attributes such as grid order, length, slope of the hillslopes, land use, elongation ratio, and topographic wetness index.

Sediment from hillslopes can be managed by implementing various best management practices (BMPs). In this study, BMP effectiveness was evaluated by comparing no-till and cover crop scenarios with conventional tillage system as a baseline scenario for dominant crop rotations in Running Turkey Creek. Results showed that the no-tillage practices reduce average soil loss by around 60\% to 75\% compared to conventional tillage. For continuous soybeans and sorghum, cover crop tillage practice works even better than the no-tillage system, reducing the soil loss by approximately 82-83\%. The combination of crop rotation with a no-till system consistently outperformed either of the no-till systems or crop rotations alone. Considerable spatial variability in BMP effectiveness was observed across different hillslopes due to varying topographic and soil attributes.

A building block approach was developed to assess whether complex three-year crop rotations could be represented by simplified 1-year, 2-year, or a combination of 1-year and 2-year rotations in the WEPP model. The performance of the building block combinations varied by rotation. Summer-crop rotation, such as corn-soybean-soybean (CSS), showed systematic bias, with building block combinations producing either overestimation or underestimation of actual soil loss depending on the combination type. Three-year rotations with winter wheat, such as soybean-winter wheat-winter wheat (SWW) and soybean-sorghum-winter wheat (SGW), showed generally good representation, though with minor underestimation and overestimation tendencies depending on the building block combination. The double-cropping rotations like soybean-winter wheat- winter wheat/soybean (SWWS) and corn-soybean-winter wheat/soybean (CSWS) showed good representation in both of their representations. The systemic bias in simplified representation is due to the management schedule and the distribution of rainfall events to each crop system by the WEPP model. The introduction of linear adjustment factors significantly improved the accuracy of the building block approach. Incorporating the slope of the flowpath as an additional factor in a multilinear regression analysis further improved the performance of the approach, reflecting the strong influence of slope on soil loss. Finally, the Monte Carlo analysis showed that around 200 flowpaths per HUC-12 is adequate for building a robust dataset for erosion analysis. These findings provide a practical and computationally efficient framework for watershed-scale and regional-scale erosion modeling for improved conservation planning and decision-making.