Automated Mapping of Flood Inundation Areas using Cloud Based Processing Tools of Google Earth Engine for Efficient Disaster Management – A Case Study of Kerala Floods

Abstract

Availability of reliable information on extent and impact of flood becomes critical for faster and efficient disaster management planning. Access to near realtime satellite data from optical and microwave sensors on Google Earth Engine (GEE) platform helps in rapid flood inundation mapping. Also, integration of Optical and Synthetic Aperture Radar (SAR) data from Sentinel series of satellites helps in detection of landuse/landcover features like Built-up, Agriculture Lands and Water bodies under flooding. The GEE platform provides extensive library tools for simultaneous pre-processing of SAR images and cloud free optical images from multiple satellites. In this research, Differential Thresholding and Differential Smoothening algorithms developed in GEE have been applied on Pre and Post SAR images for automatic identification of flood inundation areas. The inundated areas thus delineated were validated with inundation extents provided by the National Database for Emergency Management (NDEM). The methodology of automation developed in this study is applied to severe floods occurred in Kerala, India during August 2018. The sensitivity of polarization on mapping accuracy is estimated using Vertical- Horizontal (VH) and Vertical-Vertical (VV) modes available with Sentinel-1 data. The result indicates that VV mode of polarization provides better accuracy of 96% than VH mode. The coding is implemented in GEE environment with automation to provide extent of flood inundation for efficient management and mitigation planning during flood disasters.