Tahereh Ghaemi rad; Mohammad Karimi
Abstract
Forest fire is one of the most common ecological hazards whoseproper prediction of spreading is a vital issue in minimizing its destructive effects.This phenomenon depends on factors such as topography, vegetation and climate. Among the existing models, the definite empirical models presented in the ...
Read More
Forest fire is one of the most common ecological hazards whoseproper prediction of spreading is a vital issue in minimizing its destructive effects.This phenomenon depends on factors such as topography, vegetation and climate. Among the existing models, the definite empirical models presented in the form of raster including cellular automata are more populardue to their modeling simplicityand the ability to model complex systems. Different simulation systems have been developed to simulate and predict the spread of fire using cellular automata. The quality of the results obtained from these systems, in addition to the complexity of the model, depends on the accuracy and reliability of the input parameters, most of which have a degree of uncertainty. One of the constructive suggestions to overcome the uncertainty problem is the use of a two-stage simulation approach. In this approach, all of theexisting parameters in the model are first optimized by comparing the results derived from the simulation with the reality, then,the related simulation model will performthe simulation of the next step fire spread by considering the optimal values obtained for the parameters. One of the most important points in designing this system is the use ofdesirable optimization method. In this research, two optimization methods namely Particle Swarm Optimization (PSO) and Artificial Bee Colony (ABC) have been used to overcome the uncertainty problem and enhance the accuracy of forest fire spread modeling and implementation of two-stage simulation approach for a part of the forests of Gilan province. The results show that the Artificial Bee Colony (ABC) algorithm optimization method has abettercapability than the Particle Swarm Optimization (PSO) to produce optimal parameters of the desired model.
