Exploring the spatial patterns of fire density in Southern Europe using Geographically Weighted Regression

• We explored the spatial patterns of fire density in two regions of Southern Europe.
• Geographically Weighted Regression was applied to investigate main drivers of fire.
• A strong spatial variability of the explanatory power of the variables was found.
• Precipitation, livestock and shrubland were significant factors in both regions.
• Fire prevention strategies can be adjusted to particular fire conditions in an area.

The spatial patterns of fire occurrence were analyzed in two regions of Southern Europe, focusing on the long-term factors that influence fire distribution. The relationship between fire occurrence and the physical and anthropogenic variables collected was investigated with Geographically Weighted Regression (GWR) and the results were compared with Ordinary Least Squares (OLS). Local patterns of the significant variables were explored and a strong spatial variability of their explanatory power was revealed. Climate (precipitation), livestock and land cover (shrubland) were found to be significant in both regions, although in particular areas and to different extents. Regarding model performance, GWR showed an improvement over OLS in both regions.The investigation of the spatial variation in the importance of the main drivers over a broad study area, gives a valuable contribution to the improvement of fire management and prevention strategies, adjusted to the particular conditions of different areas.