Determination of forest fire area by using Landsat-8 and Sentinel-2 satellite images: Muğla case

The areas which include spreading trees, plants and bushes are called forests. Forests are among the most important natural resources with their benefits. In addition, they have an important place in the ecosystem in terms of social and environmental balance. Prediction of the formation and behavioural characteristics of fires is extremely important in terms of combating forest fires. For this purpose, using satellite images obtained by means of the developing technology provides great convenience in determining the size of a fire area and the intensity of a fire. In the present study, a forest fire occurred in Zeytinköy region of Muğla province is analysed. The areas affected by the fire are determined through creating fire intensity maps by applying dNDVI, dGNDVI, dNBR, RdNBR, RBR and dNBRn difference indexes to Sentinel-2 and Landsat-8 satellite images. Forest fire areas are also determined by object-based classification technique which is claimed to give higher accuracy results in medium and high resolution satellite images. For this purpose, when the fire data sheet obtained from the Denizova Forest Sub-District Directorate was examined, it was found that a total area of 425 ha, including 225 ha forest area and 200 ha agricultural land, was destroyed by fire. At the end of the study, the size of the area affected by the fire is found to be 494.21 ha by dNDVI index, 553.22 ha by dGNDVI index, 525.60 ha by dNBR index, 524.32 ha by RdNBR index, 547.71 ha by RBR index applied to Landsat-8 satellite image, and 513.15 ha by object-based classification method. Also, the size of the area destroyed by the fire is found to be 419.40 ha by dNDVI index, 229.16 ha by dGNDVI index, 490.16 ha by dNBR index, 481.65 ha by RdNBR index, 536.44 ha by RBR index applied to Sentinel-2 satellite image, and 431.57 ha by object-based classification method. When the results obtained are compared with the data obtained from the Forest Sub-District Directorate, it is determined that the index with the highest accuracy is dNDVI for both satellite images. When the data obtained from the difference indexes applied to Sentinel-2 satellite image and the object-based classification method are compared with the actual values, these satellite images are found to be more suitable than Landsat-8 satellite images for the studies to determine forest fire areas.