Forests and rangelands? wildfire risk zoning using GIS and AHP techniques

Abstract

Wildfire in forests and rangelands, apart from its initiating causes, is considered as an ecological disaster. Zoning natural areas according to their susceptibility to fire helps to put off operations and reduces catastrophic losses caused through a wise management plan. In this study, the zoning map of wildfire risk in forest and rangeland areas has been produced using GIS, Analytical Hierarchical Processing (AHP) and remote sensing techniques. The study area is about 196000 ha of Ilam Township, located in western Iran. The influencing factors in wildfire occurrence include current land use/cover, roads and rivers, as well as physiographic, climatic and anthropogenic themes. The locations of the wildfires have been registered by using a GPS from 2007 to 2009, to map wildfire occurring pattern in the study area. Then, using AHP techniques the influencing factors in occurrence and extension of the fires were compared in pairs and weighed. According to the weight calculated for each factor and its corresponding classes, the weighed maps of the factors were produced and employed to produce the final map of wildfire risk zoning. Finally, the zoning map of wildfire risk was produced including five classes of the risk from high to very low. Comparing the map of the wildfire risk potential to the actual fires that happened, it was found that 50 and 40 percents of the fires initiate form the areas, marked as high risk and risky zones on the map, respectively. The results indicate a high compliance of the map of wildfire risk zoning and the location of the fires in the study area. As so it predicts more than 90 percent of occurring forest and rangelands wildfires and would be helpful data for arranging a better wildfire fighting annual plan in national and regional forests and rangeland management headquarters. The model could turn to a more sophisticated one by adding extra influencing factors like, wind speed and its directions. The present model is a static one and to solve such a problem it should be promoted to a dynamic model.