Different approaches to model future burnt area in the Iberian Peninsula

• We distinguish four regions with different fire regimes in the Iberian Peninsula.
• Statistical models (meteorology driven) are developed to reproduce burnt area series.
• We check performances of RCMs simulating present climate burnt area distributions.
• Different methods to correcting model biases are tested and applied to RCM outputs.
• Our models project about 2–3 more times mean burnt area in Iberia around 2075.

In this work we developed projections for future fire regimes in the Iberian Peninsula using outputs from Regional Climate Model (RCM) from the ENSEMBLES project. Wildfires are the cause of major ecological and economic impacts in this region, and the increasing evidence of climate change consequences in this region raises concerns on the future impacts of fires in the Iberian forests ecosystems. Our results confirm that the inter-annual variability of total burnt area is mainly controlled by meteorological conditions, in spite of the current efforts for fire control and suppression. We also show that this meteorology dominance over fire activity is not only true during the fire season itself, but also that certain specific meteorological backgrounds (such as prolonged droughts) may enhance the risk for severe wildfire episodes in some areas. Based on a previous classification of the Iberian Peninsula into four distinct pyro-regions, we developed statistical models which reproduce about two thirds of the inter-annual variability of the burnt area, using meteorological variables as predictors (calibrated with data from the ERA-Interim reanalysis). Specific models were developed for each sub-domain, testing their robustness for extrapolation under climate-change conditions. Using an ensemble of state-of-the-art RCM future climate scenarios, we present future BA projections considering two alternative techniques of statistical correction of model data often used in climate change impact studies: (1) unbiasing method; (2) delta change method. Our results clearly project large increases in mean burnt areas for all the considered pyro-regions, despite some fluctuations regarding each considered technique. By 2075, mean burnt areas could be about two to three times larger than in the present, taking into account current climate projections for the next century, and non-significant changes in other external factors, such as human activity, fire suppression or land use.