Effects of combination of influencing factors on earthquake-induced landslide susceptibility assessments

• Published in: Environmental earth sciences Vol. 84; no. 7; p. 189
• Main Authors: Zhang, Yingbin, Zeng, Ying, Xu, Peiyi, Liu, Jing, Li, Zixin, Sun, Yu, Feng, Zhenhai
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.04.2025; Springer Nature B.V
• Subjects: Accuracy; Biogeosciences; Correlation analysis; Curvature; Distance; Earth and Environmental Science; Earth Sciences; Earthquake prediction; Earthquakes; Environmental Science and Engineering; Geochemistry; Geological structures; Geology; Hazard assessment; Human engineering; humans; Hydrology; Hydrology/Water Resources; Land cover; Landslides; Landslides & mudslides; Lithology; Original Article; prediction; Predictions; regression analysis; rivers; Seismic activity; Seismic response; Susceptibility; Terrestrial Pollution; topography; Landslide susceptibility assessment; Logistic regression model; Combination of influencing factors; Multiple combination strategy
• ISSN: 1866-6280; 1866-6299
• DOI: 10.1007/s12665-025-12203-6

This study aims to generate a multiple combination strategy and explore a combination of influencing factors on earthquake-induced landslides susceptibility assessments. Landslides caused by two earthquakes with similar magnitudes that both have surface rupture zones are taken as the study subjects. Data on twenty landslide influence factors including earthquakes, topography, geological structure, hydrology, stratum lithology, and human engineering activities were acquired. Based on frequency ratio analysis, normalization and correlation analysis, a multiple combination strategy is applied to develop different combinations of these factors. The logistic regression (LR) method is used to establish an evaluation model for all possible combinations. After constructing 2047 combinations of influence factors, the model with the greatest prediction accuracy was extracted. Results show that the combination of aspect, slope, distance from fault, elevation, lithology, plan curvature, profile curvature, NDVI, land cover type, PGA, and distance from river has better applicability in evaluation of earthquake-induced landslides susceptibility in the study area. The prediction accuracy of the model did not continue to improve as the number of influence factors increased, and varied depending on the type of influencing factors. We expect the multiple combination strategy to advance the prediction accuracy of modeling in the landslide susceptibility assessment process.