This study focused on how to identify a disaster stricken area hit by localized heavy rains using the remote sensing data from trees grown there. It has been known that the condition of a plant reflects the soil water-content. There is a good possibility, therefore, that trees in the areas caught in a heavy downpour should show some observable change in the remote sensing data.
The spectral responses of broad-leaf trees, Sudaji (Castanopsis cuspidate (Thunb) Schottky, var. sieboldii Nakai, sp. sudaji), were measured by a handheld spectroradiometer (400 nm to 1,000 nm at 3 nm intervals, IFOV with 3 degrees) at various water-stress levels. From this fundamental experiment, a couple of indices were found to be candidates; normalized differential vegetation index, NDVI and the ratio of reflectances in green and red regions, RIGR, which could represent the water-content. These results were confirmed using aerial photographs as well as video camera images. Images of a video camera with several band pass filters were analyzed of which two band pass filters, namely 660 nm and 750 nm, were usable. Indices were subsequently derived using the following equations, ND750, 660 = (R750 - R660)/(R750 + R660).
The concept of water-content fluctuation, WCF, representing the changes of water-content was introduced. WCF was derived by the following equation: WCF = [NDA - NDB]/NDB, where NDA and NDB represent the indices of the water-content of leaves. As the areas with lower WBF values collapsed after several weeks, it was confirmed this concept would be effective in estimating collapsed areas.
