Developing Threshold Classification Model of Burned Area Severity using Sentinel-2 Medium Resolution Satellite Images

One of the most frequent yearly natural disasters in tropical nations, like Indonesia, is a forest and land fire. Creating effective post-fire policy strategies primarily requires accurate mapping of the burned area and its severity. This study aims to identify burned areas in Rokan Hilir Regency, Riau Province, Indonesia using the medium-resolution Sentinel-2 optical satellite which can detect small-scale fires. The burned area was generated using the threshold method with three composite indices including the Normalized Difference Vegetation Index (NDVI), Normalized Burn Ratio (NBR), and Normalized Burn Ratio 2 (NBR2). Based on the identification results, the best NBR model can detect $1,684.73$ hectares of burned area with a promising accuracy of $82.45 \%$. Our burn severity map resulting from the threshold classification model has produced comparable precision to the global dataset mapping on burned areas such as MCD64A1 and FIRECCI5.