In the Cianjur and Sukabumi regions of West Java, frequent landslides over the past few decades have caused considerable loss of life and property. To aid in effective mitigation planning, predicting landslide-prone area is essential. This study utilizes a logistic regression algorithm to analyze landslide susceptibility, based on occurrence data from 2009 to 2022 provided by the Center for Vulcanology and Geological Hazard Mitigation (PVMBG). The analysis incorporates factors such as land cover, slope, lithology, rainfall, geological structures, proximity to rivers, and the Normalized Difference Vegetation Index (NDVI) derived from satellite imagery. The resulting susceptibility map was validated with field data collected in 2024, achieving an overall accuracy of 65% based on the ROC AUC graph. The map highlights the most susceptible areas in the north and east of the study site, which is dominated by hills and steep slopes are prevalent. The model’s performance was influenced by the uneven distribution of reported landslide events, suggesting that further refinement is needed.

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

Landslide Susceptibility Analysis in Cianjur and Sukabumi, West Java, Using Logistic Regression

  • Reza Oceania,
  • Sutan Vasya Assydiqi,
  • Mohamad Roviansah,
  • Rio Priandri Nugroho,
  • Misbahudin Misbahudin

摘要

In the Cianjur and Sukabumi regions of West Java, frequent landslides over the past few decades have caused considerable loss of life and property. To aid in effective mitigation planning, predicting landslide-prone area is essential. This study utilizes a logistic regression algorithm to analyze landslide susceptibility, based on occurrence data from 2009 to 2022 provided by the Center for Vulcanology and Geological Hazard Mitigation (PVMBG). The analysis incorporates factors such as land cover, slope, lithology, rainfall, geological structures, proximity to rivers, and the Normalized Difference Vegetation Index (NDVI) derived from satellite imagery. The resulting susceptibility map was validated with field data collected in 2024, achieving an overall accuracy of 65% based on the ROC AUC graph. The map highlights the most susceptible areas in the north and east of the study site, which is dominated by hills and steep slopes are prevalent. The model’s performance was influenced by the uneven distribution of reported landslide events, suggesting that further refinement is needed.