Landslide susceptibility assessment based on remote sensing interpretation and DBN-MLP model: a case study of Yiyuan County, China.
Saved in:
| Title: | Landslide susceptibility assessment based on remote sensing interpretation and DBN-MLP model: a case study of Yiyuan County, China. |
|---|---|
| Authors: | Li, Shufeng, Yin, Chao, Li, Jiaxu, Sun, Tianqi |
| Source: | Stochastic Environmental Research & Risk Assessment; Feb2025, Vol. 39 Issue 2, p493-508, 16p |
| Subject Terms: | LANDSLIDE hazard analysis, MACHINE learning, OBJECT-oriented methods (Computer science), REMOTE sensing, FIELD research |
| Abstract: | Landslide susceptibility assessment is significantly influenced by the comprehensiveness and accuracy of landslide data sets. Landslide data obtained through a single remote sensing interpretation method are often incomplete and prone to noise. Additionally, the degree of fit between mathematical models and landslide disaster-pregnant environment is not immediately apparent, and the optimal model can only be identified through extensive comparative studies. To address this issue, this study integrates SBAS-InSAR technology and object-oriented classification to provide foundational data for landslide susceptibility assessment. Eleven hazard-inducing factors are selected for five machine learning models: RF, SVM, CNN-RF, CNN-SVM, and DBN-MLP. The most effective model is determined based on TPR, AUC, and Precision values. Using Geodetector, the influence of individual factors and factor interactions on landslide susceptibility is analyzed. The results confirm 171 landslide potential sites in the study area, identified through the combined application of remote sensing interpretation and field surveys. The data obtained from these interpretations is used for model training and validation, with the DBN-MLP model yielding the highest TPR, AUC, and Precision values. Factor detector analysis reveals that elevation and land use type exhibit strong explanatory power for landslide susceptibility, followed by TWI, distance from rivers, and distance from roads. Other factors, such as NDVI, slope aspect, slope gradient, and engineering rock formation, exhibit weaker explanatory power, while SPI and plane curvature show no explanatory power. Interactive detector analysis indicates that the explanatory power of hazard-inducing factors follows two primary interaction patterns: NE or BE. These interactions not only increase the risk of landslides but also contribute to more complex disaster patterns. This study provides a critical foundation for scientific land-use planning and comprehensive landslide prevention and control, with applicability to other regions with similar environmental conditions. [ABSTRACT FROM AUTHOR] |
| Copyright of Stochastic Environmental Research & Risk Assessment is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Complementary Index |
Full Text Finder 
Nájsť tento článok vo Web of Science Be the first to leave a comment!