Modelling fire occurrence at regional scale: Does vegetation phenology matter?

Through its influence on biomass production, climate controls fuel availability affecting at the same time fuel moisture and flammability, which are the main determinants for fire ignition and propagation. Knowing the role of fuel phenology on fire ignition patterns is hence a key issue for fire prevention, detection, and development of mitigation strategies. The objective of this study is to quantify, at coarse scale, the role of the vegetation seasonal dynamics on fire ignition patterns of the National Park of Cilento, Vallo di Diano and Alburni (southern Italy) during 2000-2013. We applied a habitat suitability model to compare the multitemporal NDVI profiles at the locations of fire occurrence (the used habitat) with the NDVI profiles of the entire study area (the available habitat). Results demonstrated that, from May to October, wildfires occur preferentially at sites where the remotely-sensed NDVI observations have on average lower values than the available habitat. On the other hand, in the period November-April, wildfires tend to occur at sites where the corresponding NDVI observations have higher values than the available habitat. From a practical viewpoint, the proposed method can be implemented using many different ecogeographical variables simultaneously, thus integrating remotely sensed imagery with socioeconomic data, land cover, physiography or any landscape features that are thought to influence fire occurrence in the study area.