Mapping forest fire in relation to land-cover and topographic characteristics

In this study, a forest fire occurring in an area within the boundary of Izmir province has same vegetational and climatic characteristics as of the Mediterranean Region was evaluated. Normalized Difference Vegetation Index (NDVI) and Normalized Burn Ratio (NBR) index data pertaining to pre- and post-fire periods were produced from Landsat 5 TM imagery. In addition, the datasets were compared by calculating the two differences of NDVI (dNDVI) and NBR (dNBR) values. The dNDVI data was filtered by using an adaptive filter. The filtered data was categorized as unburned, low, medium and high levels of fire effectiveness. The levels were evaluated to determine the size and distribution, corresponding to the landcover types, slope, aspect and elevation classes. For examining the changes within the burned area, difference and ratio images were used, too. For each band, the images were classified as to one standard deviation (SD) threshold value. As a result, the 4th and the 7th bands were designated the most appropriate ones for mapping of the burned area and dNDVI and dNBR data were found usable for determination fire effectivity. The adaptive filtering technique applied is advisable to be put into practice as it is a fast and simple method of image enhancement to get utilizable and necessary data to be used at fire effectiveness levels.