Mediated and direct effects of the North Atlantic Ocean on winter temperatures in northwest Europe

This study has used a multiple regression model to quantify the importance of wintertime mean North Atlantic sea-surface temperatures (SSTs) for explaining (simultaneous) variations in wintertime mean temperatures in northwestern Europe. Although wintertime temperature variations are primarily determined by atmospheric flow patterns, it has been speculated that North Atlantic SSTs might also provide some additional information. To test this hypothesis, we have attempted to explain 1900-93 variations in wintertime mean central England temperature (CET) by using multiple regression with contemporaneous winter mean North Atlantic sea-level pressures (SLPs) and SSTs as explanatory variables. With no SST information, the leading SLP patterns (including the North Atlantic oscillation) explain 63% of the total variance in winter mean CET; however, SSTs alone are capable of explaining only 16% of the variance in winter mean CET. Much of the SST effect is 'indirect' in that it supplies no more significant information than already contained in the mean SLP; e.g. both SLP and SST together can only explain 68% of the variance. However, there is a small (5% variance) direct effect due to SST that is not mediated by mean SLP, which has a spatial pattern resembling the Newfoundland SST pattern identified by Ratcliffe and Murray (1970. Quarterly Journal of the Royal Meteorological Society 96: 226-246). In predictive mode, however, using explanatory variables from preceding seasons, SSTs contain more information than SLP factors. On longer time scales, the variance explained by contemporaneous SST increases, but the SLP explanatory variables still provide a better model than the SST variables. Copyright (C) 2003 Royal Meteorological Society.