The live fuel moisture content is an important factor in estimating the risk of forest fires and their rate of spread. However, due to a lack of research, the FMC values in the Mediterranean region of Andalusia, Spain, must be obtained by sample collection. This study is therefore the first to provide tools for estimating the moisture content of the most widespread plant species in Andalusia. First, samples were collected to estimate the moisture content of the plants; these data were collected from May 2007 to the present. Each species has its own range of moistures that depend on the time of year and the physiological state in which they are found. Secondly, an extensive database was obtained for each day of sample collection from the nearest weather station with free access. The statistics are performed at 12 solar hours on the day of sample collection and 24 h before collection, and then at 7 days, 14 days, 1 month, 3 months and 6 months before the day of collection. Finally, this database was statistically analyzed in two ways: Multiple linear regressions and random forest for each species. The predictive capacity of random forest is superior (R2 > 0.89) to that obtained in linear regression (R2 < 0.86). The highest root mean square error obtained in the case of the random forest is 0.74479 while in the linear regressions it was 1.29184. Consequently, uncertainty regarding fire behavior in the case of forest fires is reduced.
Keywords: Fuel moisture content; Linear regression; Meteorological stations; Random forest; Southern Spain.
Copyright © 2022 The Authors. Published by Elsevier Ltd.. All rights reserved.