Spatial Analyses of Cyclone Amphan Induced Flood Inundation Mapping Using Sentinel-1A SAR Images Through GEE Cloud

Abstract

AbstractThe study focused on mapping flood inundation in the lower Indo-Gangetic plains in Purba Medinipur district, West Bengal, India, using synchronized C-band Sentinel-1A synthetic aperture radar satellite images with a cloud computing API on the GEE. The study showed that a considerable proportion of the district 3978.93 km2 was flood inundated during May 2020 due to heavy rainfall and severe cyclone Amphan. Impact of flood inundation was found in agricultural areas (35.93% of the total agricultural land), followed by built-up area (5.03% of the built-up area) that affected a large population (30.85% of the total population), in the study area. Validation assessment is carried out for final flood layer mapping with seven flood status parameters such as VV, VH backscatter of Sentinel-1A, precipitation, NDVI, NDWI, soil moisture, and elevation data acquired in the month of May 2020. The validation accuracy estimation of the flood inundation map through the AUROC method along with the four machine learning models showed that the Naïve Bayes (AUROC = 92.6%) outperformed the SVM (AUROC = 89.9%), RF (AUROC = 89.4%), and logistic regression (AUROC = 88.5%) models. Cloud-based GEE services provide quick flood inundation mapping with associated damage information that can be beneficial for disaster management and emergency response. This will support crucial policy makers and stakeholders in creating urgent decisions for flood investigation for sustainable environment disaster risk mitigation management. The study requires implementing a cohesive, multi-decision-making approach with importance on self-reliance of the community for nourishment with local practices and resources.KeywordsFlood inundationCyclone AmphanSentinel-1ASARGoogle earth engine