Identifying Burnt Areas in Forests and Land Fire Using Multitemporal Normalized Burn Ratio (NBR) Index on Sentinel-2 Satellite Imagery

Abstract

Forest and land fires are global disasters that pose a serious threat to society, particularly during the dry season. Sumatra Island, located in Indonesia, is one of the areas prone to forest and land fires. It causes thick smoke every year which has implications for health and flight activities. It also threatens the 18,931 endemic plant species on the Sumatra Island so this needs to be monitored continuously for mitigation and recovery. This study aims to identify areas of past forest and land fires to support it. In this study, burnt areas were identified by utilizing Normalized Burn Ration (NBR) index on Sentinel-2 Satellite imagery using thresholding method. It is calculated as the ratio between the near-infrared (NIR) and short-wave infrared (SWIR) wavelengths. Both are very sensitive to vegetation and bare land so NBR is very potential for identification of burnt areas using multitemporal techniques. The results of this study indicated that the highest accuracy was achieved using the µ-2σ threshold model of dNBR, with the highest accuracy rate of 81.56% for the February 2019 fires. The results of this study suggest that the thresholding method using the NBR index can be utilized to identify burnt areas effectively.