Back to Search
Start Over
Landslide susceptibility zoning with five data models and performance comparison in Liangshan Prefecture, China.
- Source :
- Frontiers in Earth Science; 2024, p1-17, 17p
- Publication Year :
- 2024
-
Abstract
- Liangshan Prefecture, located at the northeastern edge of the Hengduan Mountain System and within the southern section of the Sichuan-Yunnan tectonic belt in Sichuan Province, China, a region prone to landslides, collapses and debris flows due to its active tectonics, complex topography and significant river erosion. By analysing a dataset of environment factors and geological hazard catalogue, the research uses the Relief algorithm to identify critical influencing factors for each hazard type, selecting 10, 9 and 9 factors for landslides, collapses and debris flows, respectively. Five models are used to assess the vulnerability of these hazards: the Information Value model, the Evidence Weight model, the Logistic Regression model, and both the Evidence Weight-Logistic Regression and the Information Value-Logistic Regression coupled models. The effectiveness of these models is confirmed by confusion matrix and ROC curve analyses, with the combined models showing particularly high accuracy in assessing susceptibility. High risk zones were identified in specific areas and along major fault zones in Liangshan Prefecture. The research provides significant insights into the susceptibility of geological hazards in mountainous and canyon regions, offering a comprehensive approach that goes beyond the limitations of single model applications. This methodology not only provides more accurate and comprehensive results, but also serves as a fundamental reference for geological hazard mitigation and management in Liangshan Prefecture, potentially benefiting similar regions worldwide. [ABSTRACT FROM AUTHOR]
Details
- Language :
- English
- ISSN :
- 22966463
- Database :
- Complementary Index
- Journal :
- Frontiers in Earth Science
- Publication Type :
- Academic Journal
- Accession number :
- 178604305
- Full Text :
- https://doi.org/10.3389/feart.2024.1417671