15 - Modeling the Spatial Variability of Forest Fire Susceptibility Using Geographical Information Systems and the Analytical Hierarchy Process

Abstract

The purpose of this research was forest fire susceptibility mapping in the western part of the Republic of Serbia. A forest fire inventory map was developed using a serial historical forest fire database, Worldview-2, Landsat 8 OLI, MODIS, and field recording. The proposed methodology is based on the combined application of geographical information system and multicriteria decision analysis. The methodology considers 13 conditioning factors for forest fire susceptibility mapping. These are altitude, slope degree, aspect, plan curvature, topographic wetness index, average annual rainfall, wind power, average annual temperature, distance from river, soil type, normalized difference vegetation index, distance from road, and distance from urban areas. The expert evaluation takes into account the nature and severity of observed criteria, and it is tested using three modalities of the analytical hierarchy process (AHP). The first of these uses the interval rough numbers with the analytical hierarchy process (IR’AHP). The second uses the fuzzy technique for the exploitation of uncertainty with the AHP method [fuzzy analytical hierarchy process (FAHP)], and the third scenario uses the traditional (crisp) AHP method. The final maps of forest fire by different scenarios were prepared using the weighted linear combination algorithm. For accuracy of results, a receiver operating characteristics curve and the kappa coefficient (kappa) were applied. According to the validation of results using the area under the curve and kappa coefficient, the final map obtained from the IR’AHP model was the most accurate compared to the FAHP and AHP models. In addition, based on the obtained accuracy assessment results, all models had most satisfactory performance and could be utilized for forest fire susceptibility mapping in the study area.