Abstract

Introduction This article explores the possibility of including multiple spatial scales into a mixed model for predicting the density of human-caused fire ignitions in Catalonia (North-East Spain). Methods The location of the ignitions of all human-caused forest fires recorded in Catalonia during the period 1994-2007, including 7,847 human-caused fires, was considered as modeling data. Three spatial scales were considered in the modeling. The first or small spatial scale was defined by tiling 3,003 hexagonal plots of 1,000 ha each. The two remaining spatial scales were defined by the inclusion of the plots within municipality and county (comarca) borders. Fire ignition density was estimated at plot level using proxy variables as predictors (elevation, presence of roads, density of pathways, and dominant land-use patterns). The administrative borders of municipality and county were used to allocate the spatial variation of ignition density not explained by the plot-level estimation. Conclusions The results of the study highlight that human-caused ignitions present clear spatial aggregation patterns at different spatial scales, and that an analysis at a single spatial scale ignores explanatory factors affecting other scales.