In this diploma thesis the interannual variability and trends of the winter mean tropospheric circulation in the northern hemisphere (NH) extratropics in winters from 1960/61 to 2001/02 are investigated.
Output is analysed from a recent version of the atmospheric ECMWF model that has been used to perform various hindcast experiments, including experiments with selected regions of the atmosphere relaxed toward reanalysis data (ERA-40), i.e., the tropics and the stratosphere.
The results are compared with the reanalysis data to examine the forecasting skill of the single experiments in the NH extratropics.
It is found that the stratosphere is influential on the interannual variability of the North Atlantic Oscillation (NAO) related part of the winter tropospheric circulation variability in the North Atlantic sector (NAS), but less important for other modes of variability in the NAS or over the North Pacific sector (NPS).
The influence of the stratosphere on the NAO is thought to be caused by the downward propagation of circulation anomalies, for example, caused by sudden stratospheric warmings.
Relaxing the tropical atmosphere is influential for the general circulation variability in the NAS, although the influence on the NAO is somewhat smaller than relaxing the global stratosphere.
Both regions of the atmosphere therefore are important for a seasonal forecast in the NAS in winter.
In the case of tropical relaxation, adding prescribed observed sea surface temperatures and sea ice (SSTSI) from reanalysis data improves the representation of the NAO in the model and is even more influential in reproducing the observed 42 year trend to a more positive NAO index.
The stratosphere has a significant impact on the observed positive NAO trend only between 1964/65 and 1994/95, and in this period is comparable with the other forcings, the tropics and observed SSTSI.
However, our model experiments are not able to account for more than 25% of the interannual variance of the NAO and not for more than 40% of the observed trend of the NAO in the ensemble means.
In the NPS the tropical atmosphere clearly has a strong impact on the interannual variability, and hence the seasonal predictability, which is measured here by means of the Pacific North America (PNA) pattern index.
Single realisations of model experiments with tropical relaxation represent, on average, between 40% and 50% of the variance of the observed PNA pattern.
The ensemble mean with SSTSI from reanalysis data and no relaxation captures about 25% of the interannual variance of the observed PNA, but leads to a wrong trend in the PNA between 1960/61 and 2001/02 compared with the observations.
The observed trend of the PNA is well captured in terms of the trend pattern if the tropics are relaxed to reanalysis data, except that the magnitude of the trend is reduced in the ensemble mean.
It is remarkable that the observed PNA trend between 1960/61 and 2001/02 is also found to be within the range of trends of a control experiment, that sees the climatological mean cycles of SSTSI only.
The strong impact of the tropical atmosphere on the extratropical atmosphere in the NPS is associated to the strong link between El Nino related variability in the tropical Pacific and the PNA that was confirmed by a number of previous studies.