Search

The observed and simulated low-frequency winter variability of the North Atlantic and European region is investigated based on the climate regimes paradigm. Empirical orthogonal functions and cluster analyses are used to describe the variability of monthly mean sea level pressure over the 1958–2001 period, both in the European Centre for Medium-Range Weather Forecasts (ECMWF) 40-yr reanalysis (ERA40) data set and the multi-model ensemble hindcasts from the DEMETER seasonal forecasting system.The clustering partition of ERA40 fields yields four climate regimes. The first two clusters capture the negative and positive phases of the North Atlantic Oscillation (NAO), respectively. The third and fourth clusters, respectively, display a strong anticyclonic ridge off western Europe almost covering the entire basin and a zonal pressure dipole between Greenland and Scandinavia, with a clear south-eastward extension of the low-pressure anomalies towards the Iberian peninsula.DEMETER seasonal forecasting models are able to reproduce the multimodal variability of the winter atmosphere with the same number of modes. For the ECMWF model, the pressure patterns of the regimes are very similar to those obtained for ERA40. For the six other models, the two NAO modes are well reproduced but the two other regimes are more different. In terms of forecasts, different scoring methods are used to evaluate the ability of the models to predict the correct regimes for a given date, but scores appear to be quite low.Then, the link between the pressure regimes and the corresponding temperature fields is investigated using different composite methods. All the ERA40 regimes are characterized by specific patterns of temperature. DEMETER models are also able to reproduce the temperature impacts of the different regimes. Thus, some predictability could arise from reliable seasonal predictions. Indeed, if models are able to forecast pressure fields at a monthly time-scale, it should allow us to forecast which regime will be excited and then to deduce the corresponding large-scale pattern of temperature anomalies. [ABSTRACT FROM AUTHOR]