Abstract
Object Based Image Analysis (OBIA) is a hierarchical classification approach which is applicable in solving numerous remote sensing research problems. Even though, the usage of OBIA for monitoring wildfire and burned area estimation are quite limited. This study focusses on comparing the potential of satellite-derived indices including Normalized Multi-Band Drought Index (NMDI), Normalized Burn Ratio (NBR), Burn Area Index Modified (BAIM), and Normalized Difference Vegetation Index (NDVI) for mapping the area of forest effected by forest fire. In this research, pre and post forest fire images of Landsat 8 (OLI) were processed using multi-resolution segmentation and assign class algorithm in eCognition software. Burned and non-burned areas were mapped using the combinations of selected indices (NMDI-NBR, BAIM-NBR and NDVI-NBR). The combination of BAIM-NBR index produced the highest accuracy of (80%). The study conclude that spectral indices, if use in combinations, assembled satisfactory outcomes in terms of classification accuracy.
