A synoptic circulation patterns evaluation framework for CMIP6 GCMs over the Euro-Mediterranean region

In a global warming scenario, there is a growing need to understand why the Euro-Mediterranean region is a hotspot for both warming and drying signals and the seasonal and regional details, thus providing improved climate information required by decision makers. This work focuses on the design of an evaluation framework for climate simulations based on a classification of synoptic circulation patterns (CPs). A set of 30 CMIP6 global climate models (GCMs) is evaluated in terms of how well they reproduce the spatio-temporal variability of a CPs classification within the region. CPs are constructed through a hierarchical clustering procedure, using daily mean sea level pressure (SLP) during 1950-2014 against the ERA5 reanalysis. The link with surface variables -including precipitation, minimum and maximum temperatures- is also studied. Model performance is quantified based on different metrics for the spatial and temporal representation of the SLP patterns and the associated surface conditions, allowing a ranking of the best-performing GCMs.

GCMs adequately reproduce the annual cycle of the CPs frequency, with a dominant synoptic structure during summertime enhancing warm and dry conditions. Best-performing models in this regard include MPI-ESM1-2-LR, EC-Earth3-CC and MRI-ESM2-0. However, the correct timing of this CP and the transitional patterns are often misrepresented, such as in GFDL-ESM4 and NorESM2-LM. The analysis of the surface patterns associated with each CP show good model skills, better for the extreme temperatures than for rainfall and particularly during the transition seasons, for which the GCMs spread also increases. In this sense, the models EC-Earth3-Veg, EC-Earth3-CC and GFDL-CM4 present the best scores, whereas INM-CM5-0, KIOT-ESM and NorESM2-LM show the lower skills. By blending both the spatial and temporal features of the CPs, the EC-Earth3-CC, EC-Earth3-Veg, GFDL-CM4 and MRI-ESM2-0 arise as the best-performing GCMs over the Euro-Mediterranean region.

Overall, it is highlighted that not all models perform best in all the aspects considered, emphasizing the need of a complete process-based model evaluation. This is a way to constrain the future projections in order to reduce uncertainty and come up with coherent climate information at a regional scale.