Detection of Flash Flood Inundated Areas Using Relative Difference in NDVI from Sentinel-2 Images: A Case Study of the August 2020 Event in Charikar, Afghanistan

DOI Web Site 研究データあり 参考文献53件
  • Mujeeb Rahman Atefi
    Department of Architecture, Hiroshima University, Kagamiyama 1-4-1, Higashi-Hiroshima, Hiroshima 739-8527, Japan
  • Hiroyuki Miura
    Department of Advanced Science and Engineering, Hiroshima University, Kagamiyama 1-4-1, Higashi-Hiroshima, Hiroshima 739-8527, Japan

説明

<jats:p>On 26 August 2020, a devastating flash flood struck Charikar city, Parwan province, Afghanistan, causing building damage and killing hundreds of people. Rapid identification and frequent mapping of the flood-affected area are essential for post-disaster support and rapid response. In this study, we used Google Earth Engine to evaluate the performance of automatic detection of flood-inundated areas by using the spectral index technique based on the relative difference in the Normalized Difference Vegetation Index (rdNDVI) between pre- and post-event Sentinel-2 images. We found that rdNDVI was effective in detecting the land cover change from a flash flood event in a semi-arid region in Afghanistan and in providing a reasonable inundation map. The result of the rdNDVI-based flood detection was compared and assessed by visual interpretation of changes in the satellite images. The overall accuracy obtained from the confusion matrix was 88%, and the kappa coefficient was 0.75, indicating that the methodology is recommendable for rapid assessment and mapping of future flash flood events. We also evaluated the NDVIs’ changes over the course of two years after the event to monitor the recovery process of the affected area. Finally, we performed a digital elevation model-based flow simulation to discuss the applicability of the simulation in identifying hazardous areas for future flood events.</jats:p>

収録刊行物

参考文献 (53)*注記

もっと見る

関連プロジェクト

もっと見る

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ