Abstract The Sundarbans, a globally significant mangrove forest and UNESCO World Heritage Site, faces increasing threats from tropical cyclones, with Severe Cyclonic Storm Remal in May 2024 causing widespread devastation. This study aimed to develop a remote-sensing method using integrated Sentinel-1 Synthetic Aperture Radar (SAR) and Sentinel-2 optical data via Google Earth Engine (GEE) to map cyclone-induced inundation and vegetation damage in the Sundarbans, concurrently quantifying associated economic losses through Ecosystem Service Valuation (ESV). Our multi-sensor approach involved Sentinel-1 backscatter change detection with Otsu’s thresholding for precise flood delineation and Sentinel-2-derived vegetation indices (NDVI, mVCI, DVDI) for assessing mangrove damage, employing pre- and post-cyclone imagery. Economic losses were calculated using inflation-adjusted ESV coefficients, with findings rigorously validated through field-based FGDs and KIIs. Results revealed a substantial inundation of 154,401.49 hectares (34.65% of the study area), with 70.85% of the mangroves experiencing moderate to exceptional damage, including 33.25% with "Exceptional" or "Extreme" damage, and an estimated highest economic loss of 1069.51 million US$ for Severe Damage. The robustness of these ESV estimates is supported by low coefficient of sensitivity (CS) values ranging from 0.14 to 0.19, indicating high reliability. Importantly, mapped inundation and damage showed high agreement with local perceptions and expert feedback from KIIs and FGDs. Using 400 randomly sampled field points, a ROC analysis showed DVDI accurately discriminated inundated from non-inundated sites (AUC = 0.95). The findings offer critical insights for management and adaptive strategies, including mangrove restoration, improved embankments, and strengthened disaster management plans, to bolster the Sundarbans' resilience against future cyclonic threats. Research highlights <UnorderedList Mark="Bullet"> <ItemContent> Developed an integrated Sentinel-1 SAR change-detection and Sentinel-2 vegetation index workflow for rapid, near–real-time mapping of cyclone-induced flooding and mangrove damage. Also, applied SAR backscatter differencing with Otsu thresholding to delineate cyclone-driven inundation under heavy cloud conditions. </ItemContent> <ItemContent> Used NDVI, mVCI, and DVDI to detect vegetation stress and classify mangrove damage severity across the landscape. </ItemContent> <ItemContent> Spatial analysis revealed clear patterns of inundation aligned with tidal rivers and widespread vegetation degradation, with high-impact zones concentrated in central and eastern Sundarbans. </ItemContent> <ItemContent> GWR analysis showed a strong spatially variable relationship between flooding and vegetation damage, uncovering hotspots of ecological vulnerability. </ItemContent> <ItemContent> Field surveys (400 points, FGDs, KIIs) strongly validated remotely sensed inundation and vegetation damage patterns, confirming high agreement between satellite-based assessments and community observations. </ItemContent> </UnorderedList>

Near real-time assessment of cyclone impacts on the Sundarban mangrove forest using multi-sensor satellite data and field observation approach: a case study of severe cyclonic storm Remal

期刊论文
发表日期：

2026年5月23日

查看全文

Israt Jahan,
Sajib Sarker,
Tanveer Ahmed

摘要

Abstract

The Sundarbans, a globally significant mangrove forest and UNESCO World Heritage Site, faces increasing threats from tropical cyclones, with Severe Cyclonic Storm Remal in May 2024 causing widespread devastation. This study aimed to develop a remote-sensing method using integrated Sentinel-1 Synthetic Aperture Radar (SAR) and Sentinel-2 optical data via Google Earth Engine (GEE) to map cyclone-induced inundation and vegetation damage in the Sundarbans, concurrently quantifying associated economic losses through Ecosystem Service Valuation (ESV). Our multi-sensor approach involved Sentinel-1 backscatter change detection with Otsu’s thresholding for precise flood delineation and Sentinel-2-derived vegetation indices (NDVI, mVCI, DVDI) for assessing mangrove damage, employing pre- and post-cyclone imagery. Economic losses were calculated using inflation-adjusted ESV coefficients, with findings rigorously validated through field-based FGDs and KIIs. Results revealed a substantial inundation of 154,401.49 hectares (34.65% of the study area), with 70.85% of the mangroves experiencing moderate to exceptional damage, including 33.25% with "Exceptional" or "Extreme" damage, and an estimated highest economic loss of 1069.51 million US$ for Severe Damage. The robustness of these ESV estimates is supported by low coefficient of sensitivity (CS) values ranging from 0.14 to 0.19, indicating high reliability. Importantly, mapped inundation and damage showed high agreement with local perceptions and expert feedback from KIIs and FGDs. Using 400 randomly sampled field points, a ROC analysis showed DVDI accurately discriminated inundated from non-inundated sites (AUC = 0.95). The findings offer critical insights for management and adaptive strategies, including mangrove restoration, improved embankments, and strengthened disaster management plans, to bolster the Sundarbans' resilience against future cyclonic threats.

Research highlights

Developed an integrated Sentinel-1 SAR change-detection and Sentinel-2 vegetation index workflow for rapid, near–real-time mapping of cyclone-induced flooding and mangrove damage. Also, applied SAR backscatter differencing with Otsu thresholding to delineate cyclone-driven inundation under heavy cloud conditions.

Used NDVI, mVCI, and DVDI to detect vegetation stress and classify mangrove damage severity across the landscape.

Spatial analysis revealed clear patterns of inundation aligned with tidal rivers and widespread vegetation degradation, with high-impact zones concentrated in central and eastern Sundarbans.

GWR analysis showed a strong spatially variable relationship between flooding and vegetation damage, uncovering hotspots of ecological vulnerability.

Field surveys (400 points, FGDs, KIIs) strongly validated remotely sensed inundation and vegetation damage patterns, confirming high agreement between satellite-based assessments and community observations.