Research on mudslide sensitive area identification method based on multi-source data

Mudslide disaster identification and early warning is a key link in disaster prevention and mitigation in mountainous areas. This paper proposes a new method of mudslide sensitivity evaluation based on multi-source data fusion, innovatively combining remote sensing images, high-precision topographic data and meteorological monitoring data, and constructing an improved gradient boosting decision tree model. The study adopts deep cross-modal representation learning to achieve efficient fusion of heterogeneous data, which overcomes the problem of insufficient adaptability of traditional evaluation methods to complex terrain conditions. The results show that the multi-source data fusion method significantly improves the evaluation accuracy of debris flow sensitivity and the ability of spatial differentiation expression, which provides scientific basis and technical support for the risk management of debris flow disasters in mountainous areas.