UDC 517.6 Forest Wildfire Modelling and Prediction in Russia Eu. <...> Shchetinin Chair of Applied Mathematics Moscow State Technology University “STANKIN” 3a, Vadkovsky lane, Moscow, Russia, 119136 The wildfire (forest fire) is a natural disaster that causes great economical losses in many regions of Russia. <...> In the present work the joint sample of daily values of the number of forest fire seats and the Nesterov meteorological index in Irkutsk region, seasons 1969–1988, are investigated. <...> The prediction of the future numbers of fire seats can be performed using special computer algorithm, which is shown to produce accurate and reliable estimates up to 2 days ahead. <...> Key words and phrases: forest fires, heteroscedasticity, vector multiplicative seasonal autoregressive coefficient Spearman, forecast. 1. <...> A reliable and accurate forecast allows to wisely allocate limited resources and concentrate on the optimal measures. <...> Common requirements to forest fire monitoring and forecasting are regulated by The phenomenon of forest fire is one of the most devastating natural disasters NI = ∑ i=1 n Russian state standard GOST R 22.1.09–99 “Safety in emergencies. <...> According to this document, the severity of forest fire danger is defined by the complex meteorological index of V. G. Nesterov: T(T −Td), where T is the air temperature (in Celsius degree), Td is the dew point (in Celsius degree), n is the number of days since last precipitation (precipitation values < 2.5mm are ignored) [2]. 2. <...> MAR(1)S Model to the meteorological factors, the joint sample of daily values of the number of forest fire seats and the Nesterov index in Irkutsk region, seasons 1969–1988, has been investigated: To assess statistical properties of the forest fire evolution process and its relation Xtk = (FStk NItk ) , k = p · kp, where p = 20 is the number of seasons (periods), kp = 214 is the number of observations per season (from April 1 to October 31, fig. 1). <...> Raw data – bivariate skew-normal [4] random component etk xtk = exp(ytk +s(τtk)), , . ytk = A· yt−1 <...>