Section Forest Ecosystems THE PLAUSIBLE WILDFIRE MODEL IN GEOINFORMATION DECISION SUPPORT SYSTEM FOR WILDFIRE RESPONSE Assoc. Prof. PhD. Maryna Zharikova 1 Prof. Dr. Vladimir Sherstjuk 1 Assoc. Prof. PhD. Nikolay Baranovskiy 2 1 Kherson National Technical University, Ukraine 2 Tomsk polytechnic University, Russia ABSTRACT The main objective of the paper is to describe a plausible formal model of wildfire, which should be useful for decision making for protection against wildfire. The model is based on the rough sets. The system of target objects, which need the protection against wildfire, and their values are described in the paper. The probabilistic criterion of fire danger estimation, which takes into account such factors as thunderstorm activity, anthropogenic load and meteorological characteristics, is proposed in the paper. Keywords: wildfire, wildfire perimeter, decision support system, fire danger, territorial system, rough set. INTRODUCTION Decision making for protection against wildfires is quite complicated because wildfires are affected by the simultaneous impact of a considerable number of stochastic factors, and the evolving processes are transient, non-linear and non-stationary [1, 2, 8, 9, 13, 14]. Furthermore, the input information for decision making is ambiguous, imprecise, incomplete and inconsistent, and the related events are territorially distributed. On top of it, the decision makers operate under the high responsibility conditions in a lack of time; hence, this stimulates the development of DSS for protection against wildfires. The analysis of literature shows that the majority of wildfire research is based on statistical processing of retrospective databases. Some of wildfire dynamic models use detailed mathematical descriptions of physical and chemical processes, which allow calculating the wildfire perimeter at predetermined times with a great accuracy [6, 12]. However, their high computational complexity prevents the achievement of required DSS performance. In other models the wildfire spreading processes are greatly simplified. This allows speeding up the calculations but reduces their accuracy. Therefore, the credibility of situation assessment is decreased, especially under the influence of an environmental state dynamics. The most reasonable decision could be obtained using the methods of intelligent analysis of accumulated data, which allow taking into account inconsistency and uncertainty of data. We suppose that the required DSS performance can be reached using a plausible wildfire model that is to be approximate. This implies that the wildfire perimeter could be vague. Thus, the requirements to the accuracy of the wildfire 575