Modelling the phenological niche of large fires with remotely sensed NDVI profiles

Although the area burned by single wildfires may range over many orders of magnitude, only a handful papers has explored the factors that contribute to drive fire ignitions into large burns. While in southern Europe most fire ignitions are of human origin, fire propagation is mostly influenced by weather conditions, topography and fuel type. In this framework, the phenological status of vegetation represents the primary driver influencing fuel availability and moisture content. Therefore, any investigation on fire behavior requires the capability of capturing spatio-temporal changes in coarse-scale vegetation phenology that are descriptive of changes in fuel conditions. The aim of this study is thus to apply a habitat suitability modelling tool originally developed in wildlife science for reconstructing the phenological niche of the ignition points of wildfires from remotely-sensed multitemporal NDVI profiles in Sardinia (Italy) in the period of 2000–2008. Overall, our findings indicate that wildfires occur preferentially in remotely-sensed phenological conditions that considerably differ from the mean phenological conditions in Sardinia. When compared to the entire set of ignition points (irrespective of the resulting fire size), ignitions resulting in burned areas larger than 20ha are preferentially associated to a longer and later fire season showing also marked preference for higher fuel loads.