Quantifying the impact of raindrop dynamics on soil erosion in grass–shrub slopes: theoretical, experimental, and modeling perspectives

Abstract

Raindrop impact and overland flow scouring are primary drivers of slope soil erosion, while vegetation cover plays a crucial role in mitigating these erosive forces. However, limited understanding of the dynamic interactions between vegetation and rainfall processes hampers the advancement of process-based erosion models. To systematically investigate the combined effects of vegetation composition and rainfall characteristics on erosion dynamics, a total of 495 controlled rainfall simulations were conducted using runoff plots with 33 distinct grass–shrub cover ratios (shrub cover: 0–70 %; grass cover: 0–70 %), across five rainfall intensities (I) (60–120 mm h−1) and three slope gradients (θ) (5°–15°). The results demonstrated that mixed grass–shrub communities significantly enhanced the regulation of runoff and erosion compared to single-species covers. Specifically, under mild slope and rainfall conditions (θ = 5°, I = 60 mm h−1), the grass–shrub combinations reduced the runoff rate by 22–26 % and the erosion rate by 26–40 %, thereby demonstrating their practical significance for soil conservation. To further elucidate the mechanistic basis of erosion processes, a theoretical model was developed to quantify raindrop impact force (FR), revealing a power-law relationship with the erosion rate (ER) (adjusted R2 = 0.627–0.995). Building on this, an erosion prediction model was formulated that integrates vegetation cover, raindrop impact (FR), and stream power (ω). The model was subjected to error and sensitivity analyses, and the evaluation metrics demonstrated strong performance across datasets (adjusted R2 > 0.80; Nash–Sutcliffe efficiency, NSE > 0.65), surpassing the predictive capability of the widely applied WEPP model (adjusted R2 = 0.276; NSE =  −0.153). While the model shows high accuracy and generalizability within experimental conditions, parameter recalibration is recommended for applications beyond the tested domain. These findings contribute to a deeper mechanistic understanding of vegetation–rainfall–erosion interactions and offer a novel framework for improving the fidelity of process-based soil erosion modeling.