Scholarlink

<p>This study introduces an integrated method to assess flood inundation and agricultural impact in the flood-prone Ghatal Subdivision, West Bengal, India, from 2016 to 2024. Combining Sentinel-1 (S1) SAR GRD imagery, advanced U-Net deep learning for precise flood mapping, Light Gradient Boosting Machine for flood depth estimation, and farm household survey, the research quantifies flood dynamics and crop losses. Results show that 30–40% of Ghatal was flooded at least three times in nine years, with certain villages experiencing inundation five times. Notably, 2021, 2023, and 2024 saw the majority of the villages in the subdivision submerged, with floodwaters reaching “second-floor levels” in low-lying towns. A critical finding revealed consecutive major floods in 2023 and 2024 in the same areas, leaving minimal recovery time for affected communities. The study identifies two distinct flood regimes: widespread, frequent, shallower floods and less frequent, but extremely deep, episodic events. Furthermore, precise crop vulnerability was established: jute and paddy suffered most in deeper flood zones (jute median depth 2.12&#xa0;m), while potatoes were primarily affected in shallower areas (rarely exceeding 2.5&#xa0;m). This methodology offers a robust framework for flood risk management, early warning systems, and targeted agricultural planning, directly supporting sustainable development goals in vulnerable regions.</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

Application of U-Net Deep Learning on Sentinel-1 SAR Data for Flood Inundation and Crop Damage Assessment in Ghatal Subdivision, West Bengal (2016–2024)

  • 期刊论文

  • 发表日期：

    2026年5月13日

查看全文
  • Nasir Ahammad,
  • E. Venkatesham,
  • Balamurugan Guru

摘要

This study introduces an integrated method to assess flood inundation and agricultural impact in the flood-prone Ghatal Subdivision, West Bengal, India, from 2016 to 2024. Combining Sentinel-1 (S1) SAR GRD imagery, advanced U-Net deep learning for precise flood mapping, Light Gradient Boosting Machine for flood depth estimation, and farm household survey, the research quantifies flood dynamics and crop losses. Results show that 30–40% of Ghatal was flooded at least three times in nine years, with certain villages experiencing inundation five times. Notably, 2021, 2023, and 2024 saw the majority of the villages in the subdivision submerged, with floodwaters reaching “second-floor levels” in low-lying towns. A critical finding revealed consecutive major floods in 2023 and 2024 in the same areas, leaving minimal recovery time for affected communities. The study identifies two distinct flood regimes: widespread, frequent, shallower floods and less frequent, but extremely deep, episodic events. Furthermore, precise crop vulnerability was established: jute and paddy suffered most in deeper flood zones (jute median depth 2.12 m), while potatoes were primarily affected in shallower areas (rarely exceeding 2.5 m). This methodology offers a robust framework for flood risk management, early warning systems, and targeted agricultural planning, directly supporting sustainable development goals in vulnerable regions.