Identifying large fire weather typologies in the Iberian Peninsula

Rodrigues M., Trigo, R.M., Vega-García C., Cardil A.
Agricultural and Forest Meteorology, 280, 107789, doi:

Download PDF


The catastrophic events occurred in the Mediterranean basin in the last two decades have made large fires an increasingly prominent feature in the characterization of fire regimes. Large fires have turned forest and fire management focus to landscape fuels and extreme meteorological conditions at different spatial and temporal scales, since understanding fire-weather relations is essential to protect lives and assets from severe fires. For instance, synoptic conditions leading to anomalous and sudden warm episodes quickly increase fine fuel dryness, antecedent and persistent drought events decrease the moisture of coarse fuels, whereas strong winds increase fire spread by transferring heat to new burnable fuels more rapidly, or allowing spotting. ‘Aggregate’ indexes based on local meteorological variables like the Canadian Fire Weather Index (FWI) have proved useful in estimating and summarizing fire danger into straightforward numerical representations, with the caveat that a same result may be produced by different combinations of weather variables. Analyzing fire-weather and danger components separately may help to understand the relative importance of each factor, or reveal specific interactions between them that trigger specific fire typologies (i.e. wind-driven). In this work we analyzed the influence of two FWI's components and their input weather variables in large fire incidence across the entire Iberian Peninsula. We explored several spatial (four regions) and temporal (three levels of aggregation, i.e., daily, weekly and monthly) aggregations to account for potential dissimilarities on fire-weather associations in space and time. Statistical analyses involved a multi-group PCA analysis to identify large fire-weather typologies (LFWT), later submitted to optimized hierarchical clustering to reveal underlying associations among LFWT. Results revealed four distinctive LFWT, labelled: ‘heat-driven’, ‘heat wave’, ‘seasonal drought’ and ‘wind-driven’, and six cluster associations with noticeable spatial differences. The bulk of fires started in the vicinity of the ‘Sierra de Estrela’ (Portugal), under average conditions of the four typologies, but leaning towards ‘seasonal drought’ conditions. Fires in the Mediterranean side, the largest within the IP, were associated to hot and dry spells (‘heat wave’) without remarkable drought events. Most relevant combinations of LWTs included temperature, wind speed and DC at daily and monthly scale, making them the main fire-weather factors driving large fires in the Iberian Peninsula.