While the dynamics of the atmosphere can be described by model equations which are derived from first principles, the climate is ruled by feedback loops many of which are known only on a phenomenological level. Irrespective of the foundations of the model, both weather and climate are chaotic: their evolution depends sensitively on initial conditions and on control parameters. Therefore, predictions by modeling are usually complemented by the analysis of empirical data and their extrapolation into the future. In this talk, we present some relevant aspects of model based and statistical forecasting of weather and climate. Although the quality of the weather forecast for most of us is just an issue of convenience, I will discuss the possibility for a strict limit to the number of days ahead for which weather forecast can be successful. More relevant, a clear prediction of how climate will change will be crucial for mitigation and adaptation strategies, which have to be implemented in due time. Climate change goes along with changes in the dynamics, visible through the change of frequency of different regional weather patterns and in the meandering of the jet stream. Predictions here suffer from large time scales of relaxation and of exploration of the phase space, as we will exemplify for the warming trend in Germany.
[mehr]