STUDY ON DETECTION OF DAMAGED PART OF ROAD DURING A HEAVY RAINFALL USING SAR IMAGES

Abstract

 In recent years, heavy rain disasters have caused severe damages in many parts of Japan, and it is necessary to grasp the damage situation over a wide area for the sake of rapid restoration and rescue activities after the disaster. Therefore, in this study, the damaged part of the road is extracted by using the SAR images unaffected by weather conditions in order to grasp quickly the place where the rescue operation and the pump car arrangement are carried out preferentially in the disaster. As the result, though it is possible to grasp the flooded situation in the place where the road width is large, it is proven that the extraction is difficult on the road of small width around residential areas and paddy field. In the future, it is needed to extract the submerged part and the part with the earth and sand by machine learning, etc. in order to facilitate the deciphering of the change region.