Daily fire risk (DFR) is a forecasting index defined on the basis of two meteorological parameters. Such parameters are associated with the local atmospheric column: dry stability e in 850700-hPa layer and saturation deficit D in 850-hPa level. In an earlier study, and from data collected over 10 years, a categorization of four type days based on DFR was established. In this way, from evaluation of e and D at 0000 UTC for each particular day, the associated type day was deduced. Consequently, it is possible to know whether that day had either very high, high, low, or very low fire activity. With this technique it is not possible to forecast a numerical value for the number of fires, however.
In this paper a model for estimating the outbreak of fires is presented. On the basis of an autoregressive process, AR(2), it is possible to obtain the predicted number of fires (PNF) during a day d as PNF(d) = F[TD(d), RNF(D − 1), RNF(d − 2)], where TD(d) is the type day according to the categorization established on the basis of e and D (deduced from rawinsoundings at 0000 UTC) and RNF(d − 1) and RNF(d − 2) are the numbers of fires registered over the area during two previous days.
In contrast to other papers in the literature, all fires are considered. No limitations are placed on the burned area or other measures of fire activity. Several statistical computations confirm the validity of this model.