Predictive modelling of fire occurrences from different fire spread patterns in Mediterranean landscapes.

Fire regimes are shifting worldwide because of global changes. The relative contribution of climate, topography and vegetation greatly determines spatial and temporal variations in fire regimes, but the interplay of these factors is not yet well understood. We introduce here a novel classification of fires according to dominant fire spread pattern, an approach considered in operational firefighting, to help understand regional-scale spatial variability in fire regimes. Here, we studied whether climate, topography and fuel variables allowed the prediction of occurrences from different fire spread patterns in Catalonia, NE Spain. We used a correlative modelling approach based on maximum entropy methods, and examined, through variation partitioning, the relative contribution of different factors on determining their occurrence. Our results accurately predicted the occurrence of different fire spread patterns, and the results were consistent when temporal validation was conducted. Although forest fuel factors made a higher contribution to the occurrence of convective fires, wind-driven fires were strongly related to topographic and climate factors. These findings may have a strong impact on investigations into how fire regimes may be projected into the future under forecast global change as they suggest that future environmental changes may affect different fire spread patterns in an idiosyncratic manner. [ABSTRACT FROM AUTHOR]