Indonesian savanna fire-related spectral analysis

Abstract

Peat fires in Indonesia frequently result in ecological degradation and loss. Public attention has shifted to savanna fires, which are usually overlooked but have increased in frequency in recent years, surpassing the area of controlled peat fires in Indonesia. Mapping the short-lived burns in Indonesia’s savanna is challenging. Only a few studies have examined fires in the Indonesian savanna, specifically the spectral responses associated with fires. Our study aims to analyse spectral changes caused by fires in Sumba Island, East Nusa Tenggara, Indonesia using field data and satellite images (Sentinel-2 and Landsat 8). We plotted the spectral values from blue to middle infrared bands for each burned and unburned site. In addition, we investigated the time series associated spectral indices for each site to determine how long spectral changes can be noticed immediately following fire incidents. Our research discovered that vegetation regrowth occurred rapidly (within two months), obscuring the detection of burned areas in the savanna. The red, near infrared, and middle infrared bands (>1,300 nm) were found to be the optimal spectral bands for discriminating between savanna vegetation regrowth and newly burned areas. Our findings will assist the Indonesian government in improving the mapping of burned areas for the purposes of national fire management.