Scholarlink

<p>Over the last four decades, the United States has faced severe weather and climate disasters with damages exceeding a billion dollars. Hurricane Harvey (2017) stands out as the costliest hurricane, causing catastrophic flooding in Texas. This study presents an automated flood damage assessment methodology leveraging before-and-after satellite images and machine learning to rapidly analyze disaster impacts. The proposed method employs image processing techniques to identify flood-affected areas and categorize damage into low, medium, and high-intensity levels. Using a dataset of 72,000 labeled images generated from pre- and post-event imagery, an eXtreme Gradient Boosting (XGBoost) classifier was trained, achieving an accuracy of 94.4% in predicting flood damage intensity. The results demonstrate that the proposed approach is faster, less computationally intensive, and more practical than previous methods. The methodology enables accelerated damage assessment and provides critical insights for disaster management, such as creating evacuation routes and enhancing response strategies. This study contributes to the field of spatial information science by offering a scalable and efficient solution for disaster assessment, aiding planners and authorities in mitigating the impacts of extreme weather events.</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

Rapid damage assessment using satellite images after hurricane harvey with gradient boosting models

  • 期刊论文

  • 发表日期：

    2025年6月22日

查看全文
  • Saadet Toker Beeson,
  • Kasim A. Korkmaz,
  • Munther Abualkibash,
  • Celal Cakiroglu

摘要

Over the last four decades, the United States has faced severe weather and climate disasters with damages exceeding a billion dollars. Hurricane Harvey (2017) stands out as the costliest hurricane, causing catastrophic flooding in Texas. This study presents an automated flood damage assessment methodology leveraging before-and-after satellite images and machine learning to rapidly analyze disaster impacts. The proposed method employs image processing techniques to identify flood-affected areas and categorize damage into low, medium, and high-intensity levels. Using a dataset of 72,000 labeled images generated from pre- and post-event imagery, an eXtreme Gradient Boosting (XGBoost) classifier was trained, achieving an accuracy of 94.4% in predicting flood damage intensity. The results demonstrate that the proposed approach is faster, less computationally intensive, and more practical than previous methods. The methodology enables accelerated damage assessment and provides critical insights for disaster management, such as creating evacuation routes and enhancing response strategies. This study contributes to the field of spatial information science by offering a scalable and efficient solution for disaster assessment, aiding planners and authorities in mitigating the impacts of extreme weather events.