Evaluation of Machine Learning and Deep Learning Techniques For Extraction Of Flood Extents Using Synthetic Aperture Radar

Abstract

Flooding is a worldwide issue, and floods are getting increasingly frequent and severe around the world . Floods account for 40% of natural disasters in the world (“Weather-related disasters” 2021) . Flood inundation extent is crucial information for flood mitigation, readiness, planning, and response actions (“Flood mitigation measures | Department of Infrastructure, Planning and Logistics” 2018) . Hence it is important to model and anticipate floods. Flood inundation extent estimates that are accurate and updated in real time can help people comprehend flood risk and damage, as well as aid flood mitigation and planning. Using intelligent systems and unique communication platforms, flood inundation maps can be employed in flood risk communication. Climate change, urbanisation, and other human activities disturb the hydrological cycle across the world, resulting in water pollution, floods, and droughts (Li, Zhang, and Xu, n.d.). "An accumulation of water over places that are typically not submerged" is what a flood is defined as . It is a natural event that can result in fatalities, environmental damage, and economic growth in a town. The extent of the deluge's destruction is largely determined by the actions taken by local and global authorities. In this case, flood management is crucial, with a focus on flood mapping, monitoring, forecasting, warning, and floodplain management can be classified based on the speed of the water, the location of the flooding, or the source of the flooding.