Fire hazard modeling and mapping in Bursa Forest Directorate

Studies on the prediction of fires affecting forest ecosystems are becoming increasingly important. Different methods are used to estimate the forest fire hazard. In this study, modeling and mapping of fire hazard in Maxent program was performed by using fire data obtained from Bursa Regional Directorate of Forestry between 2013-2017. For this purpose, fuel, topographic and meteorological data were used. A total of 366 forest fire data were used to create a fire probability map. According to the results of Jackknife test, the most important variables affecting the fire probability estimation model were determined as bio1, bio15, bio5, bio2, bio10. In constructing the conformity model developed to estimate the probability of fire, marginal responder curves containing the intervals determined to provide the highest contribution for each variable contributing are given in graphs. As a result; air temperature, precipitation and relative humidity parameters were the most effective parameters in the model for the creation of fire probability map. The ROC value of the suitability model developed to estimate the probability of a fire = 0.629. Forest fire occurrence data and based on some effective variables, MaxEnt the sensitivity of Turkey's western forest fires in forest areas were found to have high accuracy in modeling performance. Key Words: Forest fire, Maxent, Fire hazard modeling, Bursa