Scholarlink

<p>The Sundarbans, a UNESCO World Heritage site, faces dynamic changes as its shorelines shift under the influence of erosion and accretion. Using machine learning, we analyzed decades of satellite imagery (1988–2023) to uncover how regional climate factors and local conditions shape these patterns. Despite the complexity of shoreline dynamics, linear models remain widely used for predictions. This study aims to improve the accuracy of shoreline change models by incorporating novel techniques like island-based filtering and shoreline extraction. Findings reveal southern islands eroding faster, while northern and western regions gain land through accretion. Machine learning tools, including random forest and gradient boosting models, highlighted that regional factor, like sea proximity, are key drivers of erosion, while accretion follows nonlinear trends. The study also established optimal thresholds for when linear models work best. These insights pave the way for better coastal management strategies, helping protect vulnerable ecosystems and communities from the growing threats of climate change and shoreline retreat.</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

Unraveling Sundarbans’ erosion: how machine learning maps climate change impacts

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月17日

查看全文
  • Jyotirmoy Biswas,
  • Sabyasachi Maiti

摘要

The Sundarbans, a UNESCO World Heritage site, faces dynamic changes as its shorelines shift under the influence of erosion and accretion. Using machine learning, we analyzed decades of satellite imagery (1988–2023) to uncover how regional climate factors and local conditions shape these patterns. Despite the complexity of shoreline dynamics, linear models remain widely used for predictions. This study aims to improve the accuracy of shoreline change models by incorporating novel techniques like island-based filtering and shoreline extraction. Findings reveal southern islands eroding faster, while northern and western regions gain land through accretion. Machine learning tools, including random forest and gradient boosting models, highlighted that regional factor, like sea proximity, are key drivers of erosion, while accretion follows nonlinear trends. The study also established optimal thresholds for when linear models work best. These insights pave the way for better coastal management strategies, helping protect vulnerable ecosystems and communities from the growing threats of climate change and shoreline retreat.