Application of a CA-based model to predict the fire front in Hyrcanian forests of Iran

Forest fire is one of the most important source of land degradation that lead to deforestation and desertification processes. Thus, prediction of forest fire front is necessary to control it. In this study, Alexandridis model based on Cellular Automata (CA) rules was applied to predict the fire front in a part of Hyrcanian forests of Iran. The data of effective factors on fire front in the model (including vegetation type and density, wind speed and direction, and ground elevation) were provided from Mazandaran Natural Resources Administration (MNRA), Mazandaran Meteorological Administration (MMA), and Digital Elevation Model (DEM) of ASTER sensor. The model was used to simulate the front of a wildfire that burned a part of District Three of Neka-Zalemroud forests (DTNZ) on December of 2010. The required data of actual fire for simulation of fire front (including actual fire map, fire area, fire start point, etc.) were provided from MNRA. All effective factors maps for actual fire confine were organized in a Geographic Information System (GIS). The simulation environment was provided based on the ASCII files of altitude, vegetation density, and vegetation type matrices, together with a matrix containing the burned area. The fire front model was programmed and it was implemented by uploading of all digital layers (coding ASCII matrices) of effective variables and considering of the certain wind speed and direction in fire confine. Fire front simulation was run by considering of fire start point coordination and the fire front simulation was depicted. Finally, the number of burned and unburned cells in fire confine matrix was obtained. Results of model implementation including fire front direction and shape were compared with the actual fire confine to evaluate the accuracy of the used model qualitatively. Thus, the fire front polygon was overlaid on the actual fire polygon and the high similarity was observed between them. In addition, total accuracy and Kappa index were used to evaluate the accuracy of the used model quantitatively. The total accuracy and Kappa index were obtained 0.88 and 0.74, respectively. These results can show the accuracy of CA-based model to predict the fire front in Hyrcanian forests of Iran in current research.