Identifying Secondary Disasters Caused by Local Heavy Rain Using Remote Sensing Data

Abstract

This study focused on identifying secondary disasters caused by the local heavy rains using remote sensing data. It was assumed that the trees, grown on the areas affected by a landslide were subjected to water-stress, since the roots of these trees had been damaged. The spectral response of a Japanese cedar canopy measured by a handheld spectroradiometer varied with water-stress levels. From this fundamental experiment, we developed indices, Normalized Differential Vegetation Index: NDVI and the ratio of green and red regions RVI, which could represent the level of water-stress. These results were confirmed using aerial photographs taken before and after the disaster and video camera images taken after the disaster. For the aerial photographs, the concept of activity, which represents the changes of water-stress, was introduced. The activity index, Act, was calculated with the following equation, Act=[RVIB–RVIA]/RVIB, where RVIB and RVIA represent the RVI indices of before and after the disasters, respectively. These activity values represent the degrees of danger in a location. Images of a video camera with band pass filters were analyzed. Among the several band pass filters, it was demonstrated that three band pass filters, 550nm, 660nm and 770nm, could be used. Indices were, then, calculated using the following equations, NDVIBP=(R770–R660)/(R770+R660), RVIBP=R550/R660, respectively. This study shows the identifying possibility of locating the secondary disaster prone areas using aerial photographs taken before and after disasters. Some issues on photo taking conditions are also discussed. Moreover, the effectiveness of using video camera images is demonstrated for situation in which the satellite data or aerial photographs are not available.