Background The "National Key Soil Erosion Control Area in the Lower Reaches of Jinsha River" represents a critical ecological functional zone in China. Investigating the spatiotemporal evolution of soil erosion and its driving mechanisms under the unique and fragile dry-hot valley environment is essential for ensuring both agricultural production and ecological sustainability. However, quantitative evidence on cropland erosion in this region remains limited, particularly regarding the extent to which topography and vegetation jointly regulate erosion processes. This study therefore aims to fill this knowledge gap by providing an integrated assessment of cropland erosion patterns in the Yuanmou dry-hot valley.

Methods This study focused on the Yuanmou dry-hot valley, employing satellite remote sensing imagery with 30m resolution combined with standard runoff plot monitoring data to calibrate and validate a Revised Universal Soil Loss Equation (RUSLE) model. The model was then applied to assess regional soil erosion intensity and cropland erosion status. Statistical regression analysis was employed to investigate the spatiotemporal patterns of soil erosion from 2018 to 2022, while simultaneously integrating the spatiotemporal variations of cropland erosion to identify the spatiotemporal distribution characteristics.

Results 1) The RUSLE model constructed with multi-source remote sensing data demonstrated strong predictive capability (R² = 0.871) compared with field measurements. From 2018 to 2022, the total soil erosion area ranged between 863.09–950.91 km² with erosion intensity varying from 2601.47 to 3369.47 t/(km²·a). Specifically for cropland, erosion areas and intensities fluctuated between 19.93–29.03 km² and 338.45-630.83 t/(km²·a) respectively, peaking in 2018 and reaching the minimum by 2022. 2) Spatial analysis revealed concentrated erosion hotspots in central, south-central, and northern parts of the study area. High erosion occurred on steep hillside slopes, while low-erosion areas were located in low-lying valley bottoms, highlighting the strong influence of topography on erosion spatial patterns in this dry-hot valley landscape. Furthermore, spatial autocorrelation analysis confirmed that cropland erosion exhibited a significant clustering effect. Low-erosion croplands were mainly concentrated in valley plains, whereas high-erosion croplands were scattered across valley–mountain transition zones, underscoring the close relationship between erosion intensity and topographic setting. 3) Regression analysis indicated that slope gradient was the dominant driver of cropland erosion (β = 0.411), followed by vegetation coverage (β = 0.204). This suggests that the combination of terrain characteristics and vegetation management jointly regulates the erosion process.

Conclusions The results confirm that the study effectively addressed the initial goal of characterizing soil erosion in the Yuanmou dry-hot valley. The reduced contribution of tillage-induced erosion over time suggests that current conservation measures and farming practices have increasingly adapted to local soil-water conservation requirements. These findings provide scientific support for optimizing soil erosion assessment and formulating sustainable land management strategies in the Yuanmou dry-hot valley ecosystem.