Scholarlink

<p>Landslide disasters are increasingly endangering lives, infrastructure, and the environment, driven by rapid land use and land cover (LULC) changes. This study analyzes LULC changes from 2003 to 2023 in Rangamati Sadar, Bangladesh and projects future landscapes to 2033, utilizing Google Earth Engine (GEE) and machine learning (ML) algorithms. Analysis of multi-temporal Landsat images reveals critical changes, including reduced vegetation cover and expanded built-up and bare land areas, which correlate with rising landslide susceptibility. A random forest ML algorithm incorporating 12 environmental variables was used to generate landslide susceptibility maps (LSM). The field validation suggests that the accuracy of LSM varies from 94 to 98% based on Area Under Receiver Operating Characteristic Curve (AUROC). Our study results indicate an increase in moderate- to high-risk zones, particularly in elevated regions affected by deforestation, urbanization, and unplanned agricultural expansion. Vegetation loss and bare land significantly heighten vulnerability, with susceptibility expected to intensify by 2033. Our findings suggest enhancing slope stability, regulated land use planning, and restrictions on high-risk activities in vulnerable areas. We suggest integrating LULC data into disaster preparedness to strengthen regional resilience efforts and inform stakeholders on strategies that balance development with ecological conservation.</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

Integrated analysis of land use and land cover changes and landslide susceptibility: a machine learning approach in Rangamati Sadar, Bangladesh

  • 期刊论文

  • 发表日期：

    2025年8月18日

查看全文
  • Saddam Hossen,
  • M. Salim Uddin,
  • Yaqub Ali,
  • Parvez Rana

摘要

Landslide disasters are increasingly endangering lives, infrastructure, and the environment, driven by rapid land use and land cover (LULC) changes. This study analyzes LULC changes from 2003 to 2023 in Rangamati Sadar, Bangladesh and projects future landscapes to 2033, utilizing Google Earth Engine (GEE) and machine learning (ML) algorithms. Analysis of multi-temporal Landsat images reveals critical changes, including reduced vegetation cover and expanded built-up and bare land areas, which correlate with rising landslide susceptibility. A random forest ML algorithm incorporating 12 environmental variables was used to generate landslide susceptibility maps (LSM). The field validation suggests that the accuracy of LSM varies from 94 to 98% based on Area Under Receiver Operating Characteristic Curve (AUROC). Our study results indicate an increase in moderate- to high-risk zones, particularly in elevated regions affected by deforestation, urbanization, and unplanned agricultural expansion. Vegetation loss and bare land significantly heighten vulnerability, with susceptibility expected to intensify by 2033. Our findings suggest enhancing slope stability, regulated land use planning, and restrictions on high-risk activities in vulnerable areas. We suggest integrating LULC data into disaster preparedness to strengthen regional resilience efforts and inform stakeholders on strategies that balance development with ecological conservation.