This study analyses the suitability of remote sensing data from different sources (Landsat 7 ETMþ, MODIS
and Meteosat) in evaluating the effect of fuel conditions on fire severity, using a megafire (11 891 ha) that occurred in a
Mediterranean pine forest ecosystem (NW Spain) between 19 and 22 August 2012. Fire severity was measured via the
delta Normalized Burn Ratio index. Fuel conditions were evaluated through biophysical variables of: (i) the Visible
Atmospherically Resistant Index and mean actual evapotranspiration, as proxies of potential live fuel amount; and (ii)
Land Surface Temperature and water deficit, as proxies of fuel moisture content. Relationships between fuel conditions
and fire severity were evaluated using Random Forest models. Biophysical variables explained 40% of the variance. The
Visible Atmospherically Resistant Index was the most important predictor, being positively associated with fire severity.
Evapotranspiration also positively influenced severity, although its importance was conditioned by the data source. Live
fuel amount, rather than fuel moisture content, primarily affected fire severity. Nevertheless, an increase in water deficit
and land surface temperature was generally associated with greater fire severity. This study highlights that fuel conditions
largely determine fire severity, providing useful information for defining pre-fire actions aimed at reducing fire effects.
