Although there is a growing awareness of climate change, several analyses of regional changes are still controversial and not fully clarified. Utilising accurate observed datasets, and a new generation of high-resolution climate models (ENSEMBLES RCMs), this dissertation seeks to investigate the past and future evolution of extreme precipitation in Catalonia, and to estimate the impacts of climate change on forest fires. The analysis of the regional scenarios has been extended to Peninsular Spain.
Results show that no general trends in indices of daily precipitation extremes at a regional scale are observed. Only the consecutive dry days index (CDD) at annual scale shows a locally coherent spatial trend pattern. We have found that, although there are still significant uncertainties and despite the complexity of the task, most of the RCMs are able to reproduce observed features of present climate. In order to further downscale and calibrate the RCM outputs, we have developed a new Model Output Statistics (MOS) downscaling method that improves the RCMs regardless of the region and the model reliability. The future RCM and MOS scenarios are quite consistent over Spain: the enhanced greenhouse forcing result in a decrease in the mean precipitation fields and an increasing for the drought-related indices.
Finally, we have found a strong interaction between concurrent and antecedent climate conditions and fire variability. This highlights the importance of climate not only in regulating fuel flammability, but also fuel amount. On the basis of these results, we have developed a regression model (MLR) that produces reliable out-of-sample predictions of the impact of climate variability on summer forest fires. The application of the RCM-MLR model chain provides evidence that a transition toward warmer and (slightly) drier conditions has already started to occur and it is possible that they continue by mid-century (under the A1B scenario). These changes promote more fires, but less extended (i.e. a decreasing Burned Area). A carefully estimation of ”the cascade of uncertainty” has been addressed via Monte Carlo simulations with several climate scenarios and regression model specifications.