Many seasonal forecast systems show skill at monthly to seasonal timescales in predicting the winter North Atlantic Oscillation (NAO), the primary mode of variability in surface pressure over the North Atlantic and European sector. This skill has practical benefit for prediction of winter conditions over Northern Europe, and arises from the representation of remote teleconnections within the prediction system, such as from the stratosphere or the tropical troposphere. Despite skill in the NAO, most prediction systems have little skill in other patterns of North Atlantic winter circulation variability, such as East Atlantic Pattern (EAP – the second mode of regional winter surface pressure variability). This is despite the clear contribution that patterns such as the EAP make to European winter climate variability and their demonstrated role in the generation of extreme winter conditions.
We examine the role of the tropical troposphere and extra-tropical stratosphere in driving North Atlantic and European winter circulation patterns, with a focus on teleconnections to the EAP. We use relaxation experiments, in which a set of seasonal-length hindcasts are run with the atmospheric conditions within the relaxation region constrained to be similar to reanalysis. These are then compared with an initialised, but otherwise freely evolving, hindcast set, and with reanalysis, in regions outside the relaxation region. The aim is to assess how better prediction of the relaxation regions would influence the skill in prediction of winter atmospheric circulation in the North Atlantic-European sector.
We find that both regions play a role in influencing regional circulation. Tropical tropospheric relaxation in particular increases the reproduction of winter surface pressure anomalies. A key part of this improvement is in the EAP, which is very well reproduced. It is shown that forcing of the EAP occurs via propagating Rossby waves linked to convective anomalies in the tropical Atlantic. In addition, we find that teleconnections from either the tropics or stratosphere lead to reproduction of observed large-scale surface pressure patterns in most winters. In contrast, the diagnosed response to tropical forcing is rarely matched in the hindcast without relaxation, despite a similar rate of matches with the response to stratospheric forcing. This suggests that while winter stratospheric influences are well represented in the prediction system, tropical influences are under-represented. The results suggest that the improvement of tropical Atlantic predictability could lead to improvements for European winter predictability, and should be an important focus for future work.