Burned Area Identification using Landsat 8

Forest and land fire is a catastrophic event that always happens every year in Indonesia. In 2015, forest and land fires occured in many islands of Indonesia especially in Sumatera, Kalimantan, Sulawesi and Papua island. Therefore, it is necessary to monitor forest fires using satellite data to obtain the latest information of burned area in a large scale area. This research aims to develop a method for burned area mapping that happened between two
Landsat 8 data recording on August 13rd and September 14th 2015. The information of hotspot distribution during the period August - September 2015 was used to help visual
identification of burned area on the Landsat image and to verify the burned area resulted using the method. Samples were taken at several land covers to determine the spectral pattern
differences among burned area, bare area and other land covers, and then perform the analysis to determine the suitable spectral bands or indexs and threshold values that will be used in the model. Landsat recorded on August 13rd 2016 was extracted for an bare area, while Landsat recorded on September 14th was extracted for burned area. Multi-temporal analysis was done to get the burned area occurring during the period August 13rd to September 14th 2015. The results showed that the clouds could be separated by using a combination of ocean blue and cirrus band, the burned area by using a combination of NIR and SWIR band, while bare area
by using ratio SWIR / NIR. Burned area Obtained in this study had a high correlation with the hotspot density and visual appearance of burned area in the Landsat images.