The utilization of multi-sensor remote sensing and cloud-computing platform for mapping burned areas

Forest fire is a recurring environmental problem in Indonesia. In 2019 there were extensive fires in Indonesia, affecting parts of Jambi and South Sumatra. Therefore, the government tries to continue making efforts to inventory the area of the fires using satellite remote sensing data. This study used the Landsat-8 and Sentinel-2 optical satellites and the Sentinel-1 Synthetic Aperture Radar (SAR) satellites to inventory the fire area. This study aims to identify the burned area using these three sensors with their respective advantages and disadvantages. The data used in this study are limited to May-July 2019 for the time before the fire incident, August-September 2019 for the time of the fire, and October-December 2019 for the time after the fire incident. From the existing timeframe, there are 14 scene data before the fire incident and 11 scene data after the fire incident for Landsat-8, 112 scene data before the fire incident and, 109 scene data after the fire incident for Sentinel-2, 14 scene data before the fire incident and 16 post-fire scene data for Sentinel-1. The method used to identify fires in optical sensors is the delta Normalized Burn Ratio (dNBR). Whereas for SAR, the changes in vegetation structure were seen using VH polarization between before and after the incident. The burned area can be analyzed using Google Earth Engine (GEE) in this study. This study indicates that the fire detected by Landsat-8 is 14201.12 hectares with an overall accuracy of 95.83, Sentinel-2 is 62540.57 hectares with an overall accuracy of 96.19, and Sentinel-1 is 79689.95 hectares with an overall accuracy of 83.33.