Forest fire models are generally used in different aspects of fire management and are helpful in understanding and prediction of fire behavior. Forest fires cause a significant damage for public property by destroying a large tract of forest. This helps fire fighters to focus on an area with greater risk and to develop better substructure for fire fighter training and ultimately to plan fire-fighting policies to minimize damage and stay safe. In the same way simulation modeling also provides an adequate tool to estimate risk when actual risk data are limited or unavailable. Ultimately there is a need to model forest fire in ground, crown, and surface fuel. Forest fire risk assessment, which based on an integrated index, becomes an important tool for forest fires management. The integrated index includes the information about fuel, topography and weather condition which constitute potential fire environment together. The fuel and weather condition are essential for forest fire occurrence, so the main potential fire environment parameters in the process of the forest fire risk assessment are temperature, fuel moisture content and vegetation status. The environment parameters data for traditional forest fire risk assessment were always obtained from the weather station. In present study forest fire risk was estimated as the proportion of simulation runs that burned a particular point and was accumulated over the entire study area. Study used satellite remote sensing datasets in conjunction with topographic, vegetation and climate datasets to infer the causative factors of fires. Spatial data on all these parameters have been aggregated and organized in a GIS (Geographic Information System) framework.
In this research, the relation between the most effective environmental elements (vegetation index (VI), Land surface temperature (LST), slope, aspect, wind speed and direction) and human factors (vicinity of roads and residential areas) has been investigated as a mathematical model with the occurrence and release of fire in the forest protected area of Arasbaran. In order to validate the results, the data from previous fire burns has been used.To this end, LDCM satellite imagery, digital elevation model, wind speed and direction, and other parameters were used in synthesis remote sensing and geographic information systems. At first, a combination of environmental factors, fire hazard maps and map of areas with a 50% fire risk was produced. Then to simulate its extension, Alexandriachr('39')s semi-experimental models and cellular automation algorithms were used and the genetic algorithm is used to optimize the model parameters. The obtained results of normalized correlation coefficients of environmental parameters showed that VI, LST, slope and aspect were 29.20%, 29.11%, 21.93% and 19.75%, have the greatest correlation with the risk of fire map, respectively. In addition, about 17% of the study area have a high fire risk potential and more than 50% of the area is in a high fire hazard. In addition to environmental elements, the study of the relation between human factors and fire risk showed that the proximity to the road had the highest share in the incidence of fire. Also, the simulation results of synthesis of the Alexandros semi-experimental and cellular automation models showed that expansion of fire in the first region of the test have an overall accuracy 95.56% and kappa 91.41% and an overall accuracy 62.69% and Kappa of 13.13% compared to the reference data in the second region of the test. These results were in good agreement with the results of the simulation studies in firefighting development. Therefore, the simulation process can be used to protect the forest effectively. Results from the current study were quite significant in identifying potential active-spots of fire risk, where forest fire protection measures can be taken in advance.