Global Flood Mapper: A Novel Google Earth Engine Application for Rapid Flood Mapping Using Sentinel-1 SAR

Pratyush Tripathy, Teja Malladi | 2022

Abstract

Timely and accurate information about the extent of floodwater is critical for emergency planning and disaster management efforts. Despite the advances in computational resources and in the field of remote sensing, there is a clear gap that restricts the disaster community from leveraging the technological resources in real time, which impedes on-ground response efforts. To bridge this gap, this paper makes two contributions. First, the paper presents a new web application, the Global Flood Mapper (GFM) that allows the user to generate flood maps quickly and without getting into technical intricacies. To derive the flood extent from Sentinel-1 satellite data, the pre-flood collection is considered as base and anomaly cells in the during-flood image(s) are identified using Z-Score values. Second, it advances an existing flood mapping method to (a) Map the peak of the floods by combining ascending and descending scenes when necessary, and (b) Check for false positives in hilly terrains by adding slope and elevation mask parameters. By comparing our results with Sentinel-2 MSI derived flood maps and field photographs, we show that the GFM can generate flood maps with precision. The GFM can be used to map and download the extent of multiple flood events of an area as vector data (.kml format), which can be a critical input for flood modeling and risk and impact assessments. The GFM shall enable first responders and practitioners across the globe to overcome technical barriers and lack of computational resources to map the extent of inundation during and after floods.