Estimation of Flooded Area Using Satellite Imagery and DSM Terrain Data

Abstract

This study used satellite image and topographical data simultaneously to identify and verify the areas flooded in Yeongdeok-gun, Gyeongsangbuk-do, Korea, as a result of Typhoon Kong-Rey. The satellite image data were obtained from ESA Sentinel-1 Synthetic Aperture Radar (SAR) imagery and PlanetLab's PlanetScope optical imagery. The topographical data used to identify the depth and extent of the flooding was Digital Surface Model (DSM) data obtained through drone surveying, with resolutions of 0.02 m, 0.1 m, 0.5 m, 1 m, 3 m, 5 m, and 10 m. The area with a large color difference before and after the flooding was designated the primary flooded area. The area that was identified as lower than the primary flooded area, according to the topography data, was determined as the final flooded area. The accuracy was calculated by the intersection versus union method. It increased as the resolution of the topographical data increased and ranged between 62 and 97 percent, depending on the analysis area. In the case of flat areas of land, SAR images showed high accuracy and in the case of areas with densely-placed buildings, optical satellite images showed high accuracy. Keywords: Satellite, Inundation Trace Map, Sentinel-1, PlanetScope, DSM