Revealing the mixed driving mechanism of hydrological disaster risk in the Dongting Lake Basin based on an improved risk index

Study region: Dongting Lake Basin, China Study focus: This study focuses on assessing basin-scale hydrological disaster risk and disentangling its underlying mixed driving mechanisms. Two core objectives are defined: (1) assessing basin-scale hydrological disaster risk through satellite-derived surface water observations, and (2) quantifying the individual and combined contributions of climate variability, economic development, and vegetation dynamics to hydrological disaster risk. We develop a hybrid framework for basin-scale hydrological disaster risk assessment, which integrates satellite imagery, spectral water indices (including NDWI, MNDWI, and SPWI), and machine learning algorithms. Further, we conducted multiple attribution analysis using multiple linear regression and SHAP interpretation analysis based on random forest to quantify the long-term driving mechanism of hydrological disaster risk. New hydrological insights for the region: This study found the hydrological disaster risk in the Dongting Lake Basin has shown inconsistent trends, with an overall decreasing trend. In the study area, approximately 79.56% of the basin experienced a reduction in risk, while 20.44% exhibited an increase. The increase in risk is related to the increase in precipitation within the basin and the transition of land use from forest areas to farmland. However, changes in GDP, land use changes, and vegetation greening offset the hydrological disaster risks caused by increased precipitation in most regions. These findings provide a scientific foundation for water resource management and hydrological risk mitigation strategies.