FLOOD DETECTION IN BUILT-UP AREA USING INTERFEROMETRIC SAR DATA BY PALSAR-2

Rapid and all-weather detection of flood areas is needed for monitoring and mitigating flood disasters. In this paper, we investigated flood area detection in built-up area using PALSAR-2 data acquired during the 2015 heavy rainfall disaster in the Kanto and Tohoku areas, Japan. We performed an interferometric processing using pre- and co-event PALSAR-2 data and compared proposed phase-only statistics with traditional coherence-based flood detection. The result indicated that the flood detection accuracy by thresholding of phase standard deviation (PSD) image is more accurate than that of coherence and the accuracy was a 0.443 kappa coefficient in flood/non-flood discrimination. This result demonstrated the feasibility of using PALSAR-2 data for rapid flood monitoring in built-up area without being affected by strong backscattering from the buildings.