Uncovering spatiotemporal patterns of floods using multi-sensor optical and synthetic aperture radar imagery in the major agriculture zone of Sindh, Pakistan
摘要
Floods are becoming more frequent and severe due to climate change impacts. Near real-time flood extent mapping and damage assessment are crucial for flood management, relief operations, and risk mitigation. Mapping floods and their impacts using high-resolution optical and synthetic-aperture radar (SAR) satellite data is a fast and cost-effective choice. In this study, we used multi-sensor optical data from MODIS and Landsat satellite and SAR data from Senetinel-1 satellite to map flood extents in Sindh, Pakistan which is affected by increasingly devastating seasonal flooding. We collected and analyzed the dataset of large flood events since 2003 for extent mapping and assessing the resultant crop damage. Normalized Difference Water Index (NDWI) and thresholding and classification methods were used to map flood extents using optical and SAR images, respectively. Moreover, post-flood crop damages were calculated using the Normalized Difference Vegetation Index (NDVI) loss method. Spatiotemporal patterns of crop damage were also analyzed for a better understanding of their relationship with flood events. Our analysis showed that a significant area of Sindh Province was hit by major floods since 2003, and 1.06 million ha and 1.38 million ha of cropland were severely damaged during the 2010 and 2022 floods, respectively. Furthermore, satellite-based soil moisture analysis using Soil Moisture Active Passive (SMAP) L-4 data showed that median soil moisture was quite high throughout the planting window of the upcoming Rabi crops following the 2022 floods. The current study showcased the capabilities of satellite data, emphasizing their potential to reveal spatiotemporal patterns of floods using optical and SAR imagery in major agriculture zones.