Objective The purple soil hilly region, located in the upper reaches of the Yangtze River, is an important agricultural production area. However, it faces problems such as uneven rainfall distribution, frequent seasonal droughts, and prominent soil and water loss, which seriously affect the efficient utilization of water resources and the sustainable development of agriculture. Accurate simulation of rainfall-runoff is a prerequisite for the efficient utilization of rainfall resources on farmland. This study aims to optimize the SCS-CN model to improve its applicability to purple soil regions, enhance the accuracy of runoff simulation, and provide scientific support for the efficient utilization of rainfall resources and soil and water conservation.

Methods To improve the runoff simulation accuracy of the SCS-CN model on purple soil sloping farmland, this study analyzed infiltration and rainfall-runoff characteristics based on monitoring data from standard runoff plots. A slope correction factor was introduced to modify the curve number (CN), and the initial abstraction coefficient (λ) was adjusted according to rainfall characteristics, thereby constructing an improved SCS-CN model suitable for purple soil sloping farmland. The model performance was evaluated using three indicators: Nash-Sutcliffe efficiency coefficient (NSE), root mean square error (RMSE), and percent bias (PBIAS).

Results 1) The soil infiltration rate on purple soil sloping farmland decreased with increasing slope. The infiltration rate on the 5° slope was generally higher than those on the 10° and 15° slopes. Compared with the 5° slope, the initial infiltration rates on the 10° and 15° slopes decreased by 13.75% and 50.09%, the steady infiltration rates decreased by 21.58% and 57.58%, and the average infiltration rates decreased by 23.03% and 58.99%, respectively. 2) Individual rainfall events were classified into three types based on rainfall depth: Type I (high rainfall amount, low frequency), Type II (moderate rainfall amount, moderate frequency), and Type III (low rainfall amount, high frequency). The runoff depth followed the order: Type I > Type II > Type III. 3) After applying the slope correction formula, the calibrated CN_II values for the 5°, 10°, and 15° slopes were 78.14, 78.45, and 78.77, respectively. Based on rainfall type characteristics, the optimal λ value for Type I events with P_r > 70 mm was 0.35, and for Type III events it was 0.14. 4) The simulation results of the validation set were consistent with those of the calibration set. Compared with the standard model, the slope-rainfall type correction model improved the NSE by 31%, achieved a PBIAS of 1.12%, decreased the RMSE by 33%, and yielded an R² of 0.98.

Conclusions The SCS-CN model incorporating both slope and rainfall type modifications exhibits the highest accuracy in simulating runoff from individual rainfall events on purple soil sloping farmland. The findings provide a scientific basis for runoff regulation and the utilization of rainfall resources during seasonal droughts in the purple soil hilly region.