New DFG-funded collaborative research center to investigate errors in weather forecasts and the limits of predictability

Waves to Weather research center focuses on deficiencies and challenges of weather prediction with the aim to improve forecast models


Mid-latitude weather can be forecast with a lead time of approximately seven to ten days. This saves the lives of numerous people every year and has an economic value of several billion euros. Nevertheless, weather forecasts are sometimes rather poor, like for instance at Christmas 1999, when storm "Lothar" was not forecast well by any of the leading weather centers. Usually the reasons for such forecast busts are deficient initial data or forecast methods. However, since the atmosphere is a chaotic system, there is also a fundamental limit to predictability beyond which a forecast cannot be extended by any practical means. It is important to recognize this limit and, at the same time, produce the best possible forecast within this limit. This challenge is taken up by the new Transregional Collaborative Research Center "Waves to Weather" (TRR 165), which has recently received funding approval by the German Research Foundation (DFG).

"This collaborative project is unique in Germany. It will help us to distinguish situations in which the fundamental limits of predictability have been reached from those situations in which improved methods still result in improved forecasts," explained Professor Volkmar Wirth from the Institute of Atmospheric Physics at the Johannes Gutenberg University Mainz (JGU). In addition to the JGU, the consortium, which is being funded by the DFG for an initial four years, includes the Ludwig Maximilian University Munich as the coordinating university, the Karlsruhe Institute of Technology, Heidelberg University, Technical University Munich, and the German National Aeronoautics and Space Research Center (DLR).

Among other things, the scientists will analyze so-called ensemble forecasts in order to investigate the physical processes and interactions that occur, for instance, during the formation of storms or heat waves. Ensembles are generated through several dozens of forecasts starting from slightly different initial conditions; a measure of uncertainty is then obtained by the extent to which the individual forecasts deviate from each other. "Ensemble prediction allows us to determine the quality of weather forecasts and provides us with a tool to study predictability," added Wirth.

The scientists involved in the project are experts in atmospheric dynamics, cloud physics, statistics, numerical modeling, and visualization. The partners based in Mainz will primarily contribute their expertise in dynamical meteorology and cloud physics, but they also entrain colleagues from applied mathematics and computer sciences. "We expect to obtain new insight that allows us to better understand the deficiencies of today's forecast tools. This is how we aim to develop new models and methods in order to extend the forecast range, eventually out to the fundamental limit of predictability," said Professor Volkmar Wirth.