Delineating the Cataclysmic Impact of Flood’s Inundation Areas in Pakistan: A Multi-Temporal Analysis of Satellite Imagery

Abstract

Floods pose a global threat and are responsible for major risks inflicted on communities worldwide. Timely and accurate mapping of floods plays a crucial role in effective hazard management, assessing risks, and developing plans for hazard mitigation. This can be realized through the right application of integrated research that could fill key research gaps, such as the limitation of present techniques, lacking capacity, and mapping challenges of flooding in vast semiarid areas coupled with poor validation of Synthetic Aperture Radar (SAR)-based flood mapping methods. In response to these challenges, this research focused on the development of advanced flood mapping solutions that leverage the capabilities of remote sensing technologies, specifically multitemporal SAR and multi-spectral imagery, to improve the timeliness and accuracy of flood mapping. The analysis conducted between September 10, 2022, and September 30, 2022, using Sentinel-2 MS imagery with NDWI and Sentinel-1 SAR imagery revealed significant flood events in various areas of Pakistan. The flood mapping results from both datasets were compared, showing relatively consistent estimates of the flood extent and the number of exposed people across different provinces. However, there were noticeable differences in the estimated affected cropland, highlighting the variability in flood mapping results obtained from different satellite imagery sources. In the SAR imagery, an area of about 1,258,544 hectares and, according to MS-based NDWI analysis, 1,245,226 hectares, was detected to be flooded all over Pakistan. Much of the impact was on agricultural land, affecting about 303,306 hectares and 515,455 hectares of cropland for SAR and MS-based NDWI results, respectively. As the final steps, this study compared the kappa coefficient and overall accuracy statistics of the flood mapping results from the different thresholds of SAR and NDWI. The best accuracy was noted when the threshold value in SAR was 1.19, and for an NDWI threshold of 0.10, overall accuracy equalled 87.12% with a kappa-coefficient equaling 0.7424. The accuracy assessments are very important to ascertain the efficacy of the suggested flood mapping methodologies. These findings are very relevant to disaster response and mitigation strategies. The purpose is to reduce the vulnerability of various communities to future flood events in Pakistan.