ResearchBib Share Your Research, Maximize Your Social Impacts
Sign for Notice Everyday Sign up >> Login

Forest Fire Risk Modeling Using Logistic Regression and Geographic Information Systems: A Case Study in Muğla - Milas

Journal: Dogal Afetler ve Cevre Dergisi (Vol.8, No. 1)

Publication Date:

Authors : ; ; ; ; ; ;

Page : 66-75

Keywords : Forest Fire; Fire Risk Mapping; Logistic Regression; Geographical Information Systems; Milas (Muğla);

Source : Downloadexternal Find it from : Google Scholarexternal

Abstract

Forest fires are an important environmental problem, they negatively affect the entire ecosystem and human and animal life in it. In Turkey 192.734 hectares of forest area has been damaged in 46.669 forest fires in the last 20 years. Negligence-accident is the primary cause of these fires. For this reason, in order to minimize the frequency of forest fires and prevent damages, areas with fire risk should be determined and it is necessary to be prepared for the precautions to be taken before, during and after the fire. In this study, Logistic Regression (LR) and Geographic Information Systems (GIS) were used to model the forest fire risk for the Milas province in Muğla. Considering the topographic features, stand data and cultural data, the relationship of these factors with the occurrence of fires was investigated. Accuracy analyzes of fire risk estimation with LR and fire risks of areas with different properties were examined by Receiver Operating Characteristic (ROC) and Hosmer-Lemeshow test. In line with the findings obtained by the LR method, a forest fire risk map was created in the GIS environment. Here, forest fire risk is evaluated at five levels, with “1” very low risk and “5” very high risk. In the resulting forest fire risk map, it was concluded that 16% of the total forest areas in the study area are in high and very high risk classes.

Last modified: 2022-02-01 18:30:03