Abstract: |
A study was conducted in 2002 with the aim of defining and analysing, both spatially and temporally, meteorological, topographic, sociodemographic and vegetation conditions which might be linked to the occurrence of forest fires caused by humans in Catalonia in Spain, by integrating them into a GIS. Based these analyses, various daily prediction models were developed for human caused fires. All of them predicted occurrences in binary values (yes/no) in the 10 km2 grid generally employed by various government bodies in Spain with responsibility for fire prevention and suppression. The best of these models, a statistically based logit model, was evaluated in an operative manner during the summer of 2003, once the process for generating a daily prediction map before 8:00 am had been automated. This paper presents the modelling process, the testing of the models and the advantages and disadvantages of applying the model to the daily prediction of forest fires. |
