Landslides occur frequently in Gudalur Taluk, Nilgiri District, Tamil Nadu with its steep terrains and high rainfall and weak geological features. The proposed research seeks to come up with the detailed landslide susceptibility map based on Shannon Entropy model combined with both Remote Sensing and GIS techniques. Soil type, soil depth, slope, lithology, rainfall, land use/land cover, NDVI, and lineament characteristics were some of the main causative factors analyzed. The values of entropy were obtained to show relative relevance of each factors and soil depth was found to be the most relevant followed by land use/land cover and soil type. A map of landslide vulnerability was developed and classified into low, moderate, and high-risk areas. The ROC curve was used to validate, with the prediction accuracy in terms of Area Under the Curve (AUC) values of 0.738 and 0.762 being obtained on the training and test datasets, respectively. The findings indicate that areas in the middle, south and southeast parts of the study area are prone to landslides with northern areas being less prone. The integrated approach will help offer a reliable tool to manage the disaster risk and help in land-use planning and mitigation measures in the areas with landslides.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Terrain-Based Modeling of Landslide Susceptibility

  • R. M. Yuvaraj

摘要

Landslides occur frequently in Gudalur Taluk, Nilgiri District, Tamil Nadu with its steep terrains and high rainfall and weak geological features. The proposed research seeks to come up with the detailed landslide susceptibility map based on Shannon Entropy model combined with both Remote Sensing and GIS techniques. Soil type, soil depth, slope, lithology, rainfall, land use/land cover, NDVI, and lineament characteristics were some of the main causative factors analyzed. The values of entropy were obtained to show relative relevance of each factors and soil depth was found to be the most relevant followed by land use/land cover and soil type. A map of landslide vulnerability was developed and classified into low, moderate, and high-risk areas. The ROC curve was used to validate, with the prediction accuracy in terms of Area Under the Curve (AUC) values of 0.738 and 0.762 being obtained on the training and test datasets, respectively. The findings indicate that areas in the middle, south and southeast parts of the study area are prone to landslides with northern areas being less prone. The integrated approach will help offer a reliable tool to manage the disaster risk and help in land-use planning and mitigation measures in the areas with landslides.