Abstract
Northern Hemisphere western boundary currents, like the Gulf Stream, are key regions for cyclogenesis affecting large-scale atmospheric circulation. Recent observations and model simulations with high-temporal and -spatial resolution have provided evidence that the associated ocean fronts locally affect troposphere dynamics. A coherent view of how this affects the mean climate and its variability is, however, lacking. In particular the separate role of resolved ocean and atmosphere dynamics in shaping the atmospheric circulation is still largely unknown. Here we demonstrate for the first time, by using coupled seasonal forecast experiments at different resolutions, that resolving meso-scale oceanic variability in the Gulf Stream region strongly affects mid-latitude interannual atmospheric variability, including the North Atlantic Oscillation. Its impact on climatology, however, is minor. Increasing atmosphere resolution to meso-scale, on the other hand, strongly affects mean climate but moderately its variability. We also find that regional predictability relies on adequately resolving small-scale atmospheric processes, while resolving small-scale oceanic processes acts as an unpredictable source of noise, except for the North Atlantic storm-track where the forcing of the atmosphere translates into skillful predictions.
Highlights
The climate model is EC-Earth v3.0.1, which is an update of an earlier (v2.3) version[18]
The climatology and variability of EC-Earth compares favourably with other GCMs35–39 illustrating the benefits of a climate model being derived from a weather model
The seasonal forecasts were performed at three different resolutions: SRes, IRes, and HRes
Summary
The analysis period is December-February (DJF) and the focus is the Northern Hemisphere mid-latitude climatology and variability; the analyses are restricted to 20N–90N and later-on to the Euro-Atlantic sector. For the different fields analysed, the model climatology is computed as the ensemble-mean averaged across all start dates, whereas the model interannual variability is computed as the year-to-year differences in DJF anomalies (i.e. standard deviation) after linear detrending. Potential predictability (pp) is the predictability in a perfect model environment It is defined here as pp = σens-mean/σtot, where σens-mean is the standard deviation of the ensemble-mean anomalies and σtot the standard deviation of all the members[43]. Statistical significance of differences in climatology (variability) is assessed with a two-tailed t-test (F-test) for equal means (variances) at 95% confidence level. Statistical significance of the prediction skill, with respect to ERA-Interim, is assessed with a one-tailed t-test for correlation at 95% confidence level, as only positive correlations indicate skill; note that negative correlations are masked ou
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
