Ricerc@Sapienza

Fire ignitions tend to be aggregated in time and space creating a clustered spatio-temporal pattern that is mainly
driven by climatic factors and the availability of ignition sources. The aim of this work is to identify the spatiotemporal
distribution of wildfires hotspots in Sardinia (Italy) during 2000–2013 and to relate their dynamics
with remotely-sensed NDVI-based fuel phenology patterns. We considered eleven bi-weekly time frames (TFs)
and used kernel density (KD) estimation to spatialize the corresponding fire ignitions. Then, to identify zones of
fire occurrence concentration, we performed a quartile classification of the KD values for each TF considered.
Finally, we analyzed the spatio-temporal association between the ignitions hotspots and the remotely-sensed fuel
phenology patterns by means of a correspondence analysis and a selectivity ratio. We found that wildfires
hotspots are strictly related to anthropogenic pressure and to the spatio-temporal variation of fuel conditions in
terms of both load and moisture: areas with less fires concentration proved to be mainly associated to coarse
fuels with low seasonal NDVI variability; to the contrary, the fire hotspots resulted strictly correlated to fine fuels
with high seasonal NDVI variability. Understanding the association between the seasonal distribution of wildfires
hotspots and fuel phenology may allow the projection of fire ignition patterns to future, especially under
changing climatic scenarios.