MAPEAMENTO E CLASSIFICAÇÃO DE RISCO À QUEIMADAS - LINHAS DE TRANSMISSÃO DE ENERGIA ELÉTRICA

The fire risk will be measured and mapped combining two temporal and two spatial scales, divided in three analysis periods, resulting eight fire risk models. Each one of these modeling will include meteorological data (averages, maximum and minimum of historical series). The modeling for different periods will allow an understanding of fire risk temporal behavior. The two spatial scales are different by the coverage area and risk mapping detail level (medium and high resolution scale). The first one refers to a 353 km stretch; the high resolution scale refers to a 30 km sub-range to the one there area acquired Remote Sensing data (airborne SAR, LiDAR and pairs of stereoscopic Pleiades). The ignition risk modeling will be performed through the logistic regression analysis application in order to estimate the fire occurrence probability (start), according to variables related to natural and anthropic causative agents and factors associated to fire behavior (Climate, topography and vegetation). The products of this modeling will be raster maps whose pixels will have ignition probability values. The logistic regression model does not assume data normality and it is proper to categorical variables, such as soil use and the fire occurrence, for example. From the explanatory variables coefficients it will be possible to evaluate their influence (weight) on the ignition probability.