Atmospheric processes

Midlatitude atmospheric blocking events are important drivers of long-lasting extreme weather conditions at regional to continental scales. However, modern climate models consistently underestimate their frequency of occurrence compared to observations, casting doubt on future projections of climate extremes. Using the prominent and largely underestimated winter blocking events in Europe as a test case, this study first introduces a spatio-temporal approach to study blocking activity based on a clustering technique, allowing to assess models’ ability to simulate both realistic frequencies and locations of blocking events. A sensitivity analysis from an ensemble of 49 simulations from 24 coupled climate models shows that the presence of a mesoscale eddy-permitting ocean model increases the realism of simulated blocking events for almost all types of patterns clustered from observations. This finding is further explained and supported by concomitant reductions in well-documented biases in Gulf Stream and North Atlantic Current positions, as well as in the midlatitude jet stream variability.