Scholarlink

<p>Landslides present a significant risk in mountainous regions, leading to fatalities, damage to property, and long-term environmental consequences. These events can be induced by heavy rainfall, earthquakes, snowmelt, and human activities, with steep and unstable terrain being particularly prone to such hazards. This research focused on evaluating landslide susceptibility within the Khudi watershed in Lamjung District, employing two analytical techniques: Frequency Ratio (FR) and Logistic Regression (LR). A landslide inventory for the period 2010–2021 was compiled using satellite imagery and field verification. Thirteen factors were analyzed, including slope, aspect, curvature, topographic wetness index (TWI), stream power index (SPI), geology, soil type, land cover, normalized difference vegetation index (NDVI), proximity to roads, distance to rivers, elevation, and mean annual rainfall. The dataset was divided into 70% for model training and 30% for testing, adhering to standard practices in landslide susceptibility mapping. The resulting susceptibility maps were categorized into five classes as very low, low, moderate, high, and very high using the natural breaks classification method. According to the FR method, the areas in each category were 16.07%, 38.70%, 19.22%, 21.02%, and 4.97%, respectively, whereas the LR method predicted 24.15%, 38.78%, 30.77%, 21.34%, and 12.25%. Model validation revealed that LR outperformed FR, with an AUC of 0.816 compared to FR’s AUC of 0.767. The susceptibility map produced by the LR model showed a strong correlation with historical landslide data, suggesting that this map is an effective tool for informed decision-making in managing landslide hazards.</p>

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

Assessing landslide susceptibility using frequency ratio and logistic regression models in the Khudi Watershed

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月9日

查看全文
  • Shudarshan Hamal,
  • Susmita Dhakal,
  • Binita Shahi,
  • Padam Bahadur Budha,
  • Wafa Saleh Alkhuraiji,
  • Mohamed Zhran

摘要

Landslides present a significant risk in mountainous regions, leading to fatalities, damage to property, and long-term environmental consequences. These events can be induced by heavy rainfall, earthquakes, snowmelt, and human activities, with steep and unstable terrain being particularly prone to such hazards. This research focused on evaluating landslide susceptibility within the Khudi watershed in Lamjung District, employing two analytical techniques: Frequency Ratio (FR) and Logistic Regression (LR). A landslide inventory for the period 2010–2021 was compiled using satellite imagery and field verification. Thirteen factors were analyzed, including slope, aspect, curvature, topographic wetness index (TWI), stream power index (SPI), geology, soil type, land cover, normalized difference vegetation index (NDVI), proximity to roads, distance to rivers, elevation, and mean annual rainfall. The dataset was divided into 70% for model training and 30% for testing, adhering to standard practices in landslide susceptibility mapping. The resulting susceptibility maps were categorized into five classes as very low, low, moderate, high, and very high using the natural breaks classification method. According to the FR method, the areas in each category were 16.07%, 38.70%, 19.22%, 21.02%, and 4.97%, respectively, whereas the LR method predicted 24.15%, 38.78%, 30.77%, 21.34%, and 12.25%. Model validation revealed that LR outperformed FR, with an AUC of 0.816 compared to FR’s AUC of 0.767. The susceptibility map produced by the LR model showed a strong correlation with historical landslide data, suggesting that this map is an effective tool for informed decision-making in managing landslide hazards.