1. Temporal and spatial pattern analysis and susceptibility assessment of geological hazards in Hunan Province of China from 2015 to 2022.
- Author
-
Tian, Hao-Hua, Xiao, Ting, Shu, Biao, Peng, Zu-Wu, Meng, De-Bao, and Deng, Min
- Subjects
- *
MACHINE learning , *PROBABILITY density function , *HAZARD mitigation , *MASS-wasting (Geology) , *RADIOACTIVE waste management , *DEBRIS avalanches , *RISK assessment - Abstract
Exploring the interplay between the spatial and temporal distribution of geological hazards and the complex hazard-prone environment provides valuable insights for effective management of geological hazards. We analyzed the spatial and temporal distribution characteristics of historical geological hazards in Hunan Province from 2015 to 2022 using standard deviation ellipse, mean nearest neighbor, and kernel density estimation. We also employed three machine learning models to evaluate the susceptibility of geological hazards. Our research reveals that the frequency trends of landslide, debris flow, and collapse over time exhibit highly similar characteristics, and their high-density areas in spatial distribution significantly overlap. Spatial correlation analysis and kernel density estimation show that geological hazards tend to aggregate in certain areas. Central Hunan and its surrounding regions are high-density areas for geological hazards, with the most severe surface collapses occurring in the Loudi area. Geological hazards are more likely to occur at the intersection of various administrative regions, and landslides, in particular, manifest as multi-point distributions, forming a belt around cities—especially in the Hengyang area, surrounded by ridges and mountains. The locus of gravity shift for geological hazards is complex, yet the distance is small, and it primarily concentrates in and around Loudi. Lithology is the most crucial factor affecting geological hazards, followed by elevation and the topographic relief index. The extreme gradient boosting model achieved an AUC value of 0.786, outperforming the random forest and support vector machine models. The susceptibility assessment aligns closely with the kernel density estimation results. This study provides a solid foundation for understanding the spatiotemporal evolution and susceptibility of geological hazards on a local scale, thereby aiding in hazard risk prevention and control. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF