Development a method to detect damaged forests by strong winds using vegetation index ratio before and after the disaster

Abstract

Estimating damage to forests by strong winds is presented. In this study, a method to detect disaster-stricken forests by using the normalized difference vegetation index (NDVI) ratio before and after a disaster was developed. Chiba prefecture was selected as the test area. Many trees in Chiba prefecture were fallen due to strong winds of the Typhoon 15 (T1915) in 2019. One Sentinel-2 MSI dataset before T1915 and three after it were used for detecting wind fallen trees (WFTs). NDVI ratios in pure pixels of WFTs (WFT100%) were examined. NDVI ratios in pure pixels of no WFT (WFT0%) were also examined. Three linear interpolation expressions between WFT100% and WFT0% for MSI dataset after T1915 were developed to estimate the area ratio of WFTs in each mixel. 78% of WFTs detected by visible interpretation of aerial photographs were located in pixels with the area ratio of WFTs greater than 0%. In contrast, 93.4% of pixels with the area ratio of WFTs greater than 0% actually included WFTs. This method is useful to estimate the distribution of WFTs in the wide area.