Multi-annual predictability of summer temperatures over Central Europe verified

Scientists around Dr. Wolfgang Müller at the Max Planck Institute for Meteorology (MPI-M) have studied the skill of predictions for seasonal surface temperatures over Europe. Multi-annually averaged summer temperatures are predictable for central and eastern Europe; multi-annually averaged winter temperatures are predictable for northern Europe. Further improvements of the forecasting system are necessary to represent the observed correlations between temperatures of the North Atlantic and over Europe even more reliably.

Climate predictions of up to 10 years are being performed worldwide. For the first time, the Fifth Assessment Report of the IPCC (Intergovernmental Panel on Climate Change) will assess such decadal predictions. In contrast to the well established climate projections – where the future climate is simulated dependent on the assumed concentrations of greenhouse gases – decadal climate predictions provide prognoses not only for anthropogenic climate change but also for natural climate variability. To include natural variability, predictions have to be started (“initialized”) from the observed climate state.

Decadal climate predictions show high predictive skill especially for surface temperatures of the North Atlantic (Smith et al. 2007, Pohlmann et al. 2009, Krüger et al. 2012, Matei et al. 2012a). There is, in addition, one successful example for the prediction of the Atlantic Meridional Overturning Circulation (AMOC), which is closely linked to the heat transport to the middle latitudes (Matei et al. 2012b). Although significant predictive skill has thereby be established for the North Atlantic, it is considerably more difficult to establish corresponding predictive skill over the European continent.

Wolfgang Müller and his team base their research on the Earth System Model of the Max Planck Institute for Meteorology (MPI-ESM) and have performed an extensive set of decadal climate predictions. In contrast to earlier work, the predictive quality for surface temperature is not only analyzed for straight multi-annual means but also for the multi-annual means of seasonal temperatures. The predictive skill for multi-annual predictions over the European continent clearly depends on the season (see figures).

Predictive skill for surface temperature averaged over two to six years for the mean winter temperatures and for the mean summer temperatures

Figures: Predictive skill for surface temperature averaged over two to six years for the mean winter temperatures (DJF, above) and for the mean summer temperatures (JJA, below). Red fields display that initialized predictions have smaller errors than simulations that do not use the observed climate state as their starting point. The measure for predictive quality is based on the root-mean-square error.

While the mean winter temperatures show enhanced prediction skill for northern Europe, the mean summer temperatures show enhanced prediction skill especially for central and eastern Europe. A direct comparison with observations indicates that summer temperatures in central and eastern Europe are directly linked to North Atlantic surface temperatures, a region for which high predictive skill is well established.

Despite the prediction success shown here, there are clear indications that the summer predictions could be improved further. The predictions show that the atmospheric remote response – and thereby also the temperatures in central Europe – correlate with the observed surface temperatures of the North Atlantic and can be reproduced at least in the initial phase of the predictions. The predictive skill of the summer temperatures, however, decreases significantly in the further course of the predictions (for example averaged over the prediction years 2-6, see figure). This is mainly due to the model’s incomplete representation of the fundamental atmosphere-ocean coupling mechanisms over the North Atlantic.

Overall, the prediction experiments show that multi-annual climate predictions are feasible for central and eastern Europe. However, they can and have to be improved substantially, by improving the model’s representation of the remote atmospheric response to the North Atlantic.

Mueller, W. A., Baehr, J., Haak, H., Jungclaus, J. H., Kröger, J., Matei, D., Notz, D., Pohlmann, H., von Storch, J. S., & Marotzke, J. (2012). Forecast skill of multi-year seasonal means in the decadal prediction system of the Max Planck Institute for Meteorology. Geophysical Research Letters, 39: L22707. doi:10.1029/2012GL05332

Kröger, J., W. A. Müller and J. S. von Storch (2012): Impact of different ocean reanalyses on decadal climate prediction, Climate Dynamics, 39, 795-810.

Matei, D., J. Baehr, J. H. Jungclaus, H. Haak, W. A. Müller, and J. Marotzke, 2012a: Multiyear prediction of monthly mean Atlantic Meridional Overturning Circulation at 26.5ºN. Science, 335, 76-79.

Matei, D., H. Pohlmann, J. H. Jungclaus, W. A. Müller, H. Haak, and J. Marotzke, 2012b: Two tales of initializing decadal climate prediction experiments with the ECHAM5/MPI-OM model, J. Climate, in press, doi:10.1175/JCLI-D-11-00633.1.

Pohlmann, H., J. H. Jungclaus, A. Koehl, D. Stammer, and J. Marotzke, 2009: Initializing decadal climate predictions with the GECCO oceanic synthesis: Effects on the North Atlantic. J. Climate, 22, 3926-3938.

Smith, D.M., S. Cusack, A.W. Colman, C.K. Folland, G.R. Harris, and J.M. Murphy, 2007: Improved surface temperature prediction for the coming decade from a global climate model. Science, 317, 796-799.


Dr. Wolfgang Müller
Max Planck Institute for Meteorology
Phone: 0049 40 41173 370
Email: Opens window for sending emailwolfgang.mueller@we dont want