Utilizing an 11-year runoff plot dataset to evaluate the regulation of six land management practices on runoff and sediment on Mollisols slopes and the applicability of the WEPP model

Abstract

Mollisols in Northeast China have gradually shifted from a highly productive agroecosystem to a degraded landscape due to severe soil erosion. However, there is a notable lack of long-term monitoring studies on runoff and soil loss across various land management practices, as well as insufficient model calibration and validation efforts aimed at mitigating land degradation. This study conducted in situ observations of runoff and soil loss under natural rainfall from 2013 to 2023 using 12 runoff plots (20 m × 4.5 m, 8.7 % slope) in Northeast China, evaluating six land management practices: bare land (BL), natural vegetation restoration (NVR), no-tillage (NT; 3 replicates), reduced tillage (RT; 3 replicates), conventional tillage (CVT; 3 replicates), and contour tillage (CTT). The Water Erosion Prediction Project (WEPP) model was subsequently employed to evaluate its applicability for modeling runoff and soil loss at event scale. Results indicated that (i) land management practices significantly affected runoff and soil loss, with the annual average losses ranked as follows: BL > RT > CVT > CTT > NT > NVR. Implementing NT or CTT was recommended to mitigate soil erosion. (ii) The spatial flux of runoff and sediment was significantly altered by the vegetation and the ridges-furrows system. Highly erosive rainfall was the dominant type of induced soil loss, highlighting its importance in ridge-furrow design. (iii) Due to differences in soil erosion types, soil loss predictions for the BL, RT, and CVT practices were sensitive to rill erodibility, while NVR, NT, and CTT practices were sensitive to interrill erodibility. (iv) High-resolution rainfall breakpoint data were used to minimize climate-induced errors in the WEPP model. Overall, the calibrated WEPP model effectively predicted both runoff depth and soil loss (NSE > 0.5), and predicted soil loss better than runoff depth. However, predictions for CTT were less satisfactory (NSE < 0.5), suggesting further model internal adjustments. This study is a foundational step toward optimizing land management and extending soil erosion predictions.