Potential seasonal predictability of the observed Euro-Atlantic atmospheric variability using SST forced ECHAM4-T42 simulations
Résumé
To estimate the potential seasonal predictability of the Euro-Atlantic atmospheric variability, canonical correlation analysis is used, and a comparison is made between the NCEP reanalysis with ensemble simulations of ECHAM4-T42 forced with observed sea surface temperature (SST) and sea-ice boundaries for 1951–94 and changing CO2 concentration. The method identifies those atmospheric modes of variability that have similar temporal evolution in the observations and the ensemble mean of the simulations. Signals due to long-term changes in the forcing were first reduced by removing a third-order polynomial from all data.
Significant covariability in the 500 hPa geopotential height over the Euro-Atlantic region is found from autumn to spring. The best agreement between the patterns is seen in late winter, where a mixed Pacific–North America (PNA) and tropical/northern-hemisphere teleconnection pattern is the dominant signal. Although it tends to modulate the North Atlantic Oscillation (NAO), the direct influence over Europe is limited. In all cases, the covariability seems to be due to remote forcing by tropical Pacific SST. However, the model response to the El Niño Southern Oscillation (ENSO) always shows a PNA pattern, while the related observed signal undergoes large seasonal changes.
The sea level pressure (SLP) over the Euro-Atlantic region seems to be much less sensitive to remote ENSO forcing, although traces of its influence can be detected in late winter. On the other hand, a highly significant covariability is found between modelled and observed SLP anomalies in autumn, reflecting the influence of the North Atlantic SST on the NAO. However, the model does not reproduce the observed SLP structure. It is also shown that the reproducibility of the NAO in the undetrended ECHAM4 ensemble simulations in winter that was found by Latif et al. in the same simulations is related to long-term trends in tropical Pacific and Indian Ocean SST variability. However, its origin cannot be determined as the CO2 concentration was specified from the observations in the simulations.