Filtered future projections in the Western Mediterranean: atmospheric circulation patterns and climate extremes

The Western Mediterranean (WMed) has been pointed out as a hotspot region for both warming and drying signals. However, there is still large uncertainty in future projections due to model uncertainty and misrepresentation of specific processes. Thus, the need for a better understanding of the future climate of the WMed becomes evident. Improved climate indicators for decision-making can benefit from a deeper insight on future climate extremes and their related atmospheric circulation, taking into account the spread in model performances over the WMed region. 

The present work is based on the analysis of future projections of rainfall and temperature extremes from a set of CMIP6 global climate models (GCMs) during 2070-2100, according to their representation of the dominant synoptic circulation patterns (CPs). CPs are defined using daily mean sea level pressure (SLP) using hierarchical clustering and data reduction through empirical orthogonal functions. The ERA5 reanalysis during 1950-2022 was considered as the reference to evaluate the historical GCMs simulations, constructing the CPs with SLP and analyzing their link to surface variables including precipitation, maximum and minimum temperatures. To assess the future synoptic circulation, the clustering algorithm is replicated and future CPs are compared to the historical ones in terms of frequency, shape and intensity changes in the CPs.

Based on the historical CPs, a model ranking is generated using a combination of spatial and temporal reproduction metrics for the SLP patterns and the associated surface conditions. GCMs manage to reproduce the annual cycle of the CPs frequency, with a dominant summer CP enhancing warm and dry conditions. However, the correct timing of this pattern and the transitional CPs (that is, during the autumn and spring seasons) still need to be more accurate. The analysis of the associated surface patterns shows good model performance, better for temperature than for rainfall, particularly in the transition seasons, for which the GCMs spread in their skill score increases. This process-based evaluation leads to a model ranking that is used to construct multiple model ensembles, considering different weighting strategies based on model performance, spread and independence. 

In terms of climate extremes, the uncertainty in future projections of the indices ─including the expected increases in the frequency of warm days and dry spells─ can be reduced by selecting specific subsets of GCMs, according to the process-based ranking. In particular, the warming and drying signals over areas such as the northeastern Iberian Peninsula are clearer in the best-performing GCMs ensemble. This constraining procedure shows more clear results in summer than in winter, when natural variability has a larger role in modulating the WMed changing climate.

These changes in temperature and rainfall extremes were related to the changing frequency of the CPs driving the specific extremes. CPs present some differences in their seasonal distribution for the late 21st century compared to their historical records, while the centroids of the CPs often present changes, evidencing modifications in the intra-pattern variability. Altogether, the projected future extremes can be associated with differences in future climate variability, given a context of global warming.