Scholarlink

<p>Indian Himalayas is a landslide hotspot in the world where numerous landslides occur every year. Mandakini Catchment in the Indian Himalayas experienced the largest flood and landslide disaster of the last millennium in 2013. Identification of new landslides following a disaster is required for appropriate decision-making on the infrastructure development, stable local economy, reducing landslide risk, and disaster preparedness. In this study, the landslide boundary expansion and new landslides were detected in the Mandakini Catchment at one-year interval from 2014 to 2023 by modifying the Semi-Automatic Landslide Detection - Change Detection (SALaD-CD) framework. The SALaD-CD framework is an Object-Based Image Analysis (OBIA) technique, that includes Mean-Shift segmentation, change detection methods, and Random Forest classification. The framework was modified by including change vector analysis, image ratio, soil adjusted vegetation index, and spectral angle to the change detection components. Two new modules, training signature separability check and removal of false positives, were added in the modified framework. The modified SALaD-CD framework resulted in an F1 score of 0.876 leading to the identification of expanded landslide boundaries and new landslides across the catchment since 2014. Results indicate higher frequency of new landslides on gentle slopes (14°-43°) and within a buffer of 50–150&#xa0;m from the previous year landslides. In this work, we have produced the most updated landslide inventory for the Mandakini Catchment up to year 2023, which is useful for management of landslide disaster in this region. Further, the proposed modified SALaD-CD framework can be implemented in other landslide-prone areas to generate rapid landslide inventory.</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

Spatiotemporal landslide evolution in upper Indian Himalayas post the great 2013 disaster using modified SALaD-CD framework

  • 期刊论文

  • 发表日期：

    2025年9月6日

查看全文
  • Arnab Chowdhury,
  • Alok Bhardwaj

摘要

Indian Himalayas is a landslide hotspot in the world where numerous landslides occur every year. Mandakini Catchment in the Indian Himalayas experienced the largest flood and landslide disaster of the last millennium in 2013. Identification of new landslides following a disaster is required for appropriate decision-making on the infrastructure development, stable local economy, reducing landslide risk, and disaster preparedness. In this study, the landslide boundary expansion and new landslides were detected in the Mandakini Catchment at one-year interval from 2014 to 2023 by modifying the Semi-Automatic Landslide Detection - Change Detection (SALaD-CD) framework. The SALaD-CD framework is an Object-Based Image Analysis (OBIA) technique, that includes Mean-Shift segmentation, change detection methods, and Random Forest classification. The framework was modified by including change vector analysis, image ratio, soil adjusted vegetation index, and spectral angle to the change detection components. Two new modules, training signature separability check and removal of false positives, were added in the modified framework. The modified SALaD-CD framework resulted in an F1 score of 0.876 leading to the identification of expanded landslide boundaries and new landslides across the catchment since 2014. Results indicate higher frequency of new landslides on gentle slopes (14°-43°) and within a buffer of 50–150 m from the previous year landslides. In this work, we have produced the most updated landslide inventory for the Mandakini Catchment up to year 2023, which is useful for management of landslide disaster in this region. Further, the proposed modified SALaD-CD framework can be implemented in other landslide-prone areas to generate rapid landslide inventory.