Soil erosion sensitivity and prediction for hilly areas of Hubei Province, China, using combined RUSLE and LSTM models.

Purpose: Hilly areas are highly susceptible to soil erosion. This study aims to discover the drivers of soil erosion, identify soil erosion–sensitive areas, and predict future soil erosion in typical hilly areas of Hubei Province, China, using combined RUSLE and LSTM models. Materials and methods: In this study, soil erosion in hilly areas of Hubei Province from 2000 to 2020 was quantitatively analyzed using the revised universal soil loss equation (RUSLE), and the soil erosion sensitivity evaluation system was constructed, a geographic detector was employed to identify the main drivers of soil erosion sensitivity, and using the long short-term memory neural network model (LSTM) to predict soil erosion in 2025. Results and discussions: The results showed that most areas were dominated by slight and moderate erosion. Slope and vegetation coverage were identified as the core elements influencing the space heterogeneity of soil erosion. Soil erosion sensitivity was mainly composed of moderate sensitivity, accounting for more than 70% of the total area. The strong and extreme sensitivity demonstrated a downward trend with the continued implementation of slope management and forest rehabilitation from slope agriculture, whereas the sensitivity was still higher in the northwest and southwest Hubei Province. Regions with severe soil erosion had high sensitivity, and the spatial distribution of the two is strongly coherent. Areas with surface relief > 300 m and vegetation cover < 30% had the highest sensitivity and should be highly valued. The percentage of moderate and higher soil erosion area in 2025 was 3.77% lower than in 2020, but severe erosion still exists in the northwest and southwest Hubei Province. Conclusions: Soil erosion sensitivity in the western part of the study area was the highest, followed by the southeast, and the overall erosion sensitivity was gradually decreasing during the studied period. In the future, soil erosion intensity will show a downward trend, whereas the deployment of soil and water conservation measures in soil erosion–sensitive areas should still be strengthened. The results are helpful for accurate soil erosion control and prediction in the hilly areas of Hubei Province, China. [ABSTRACT FROM AUTHOR]