| Many operational weather forecasting centers are developing or operating high-resolution convection-permitting numerical weather prediction models. For example, the German Weather Service is operating a 2.8km grid-spacing configuration of the COSMO model, named COSMO-DE. These new high-resolution models attempt to explicitly simulate weather phenomena on the mesoscale, such as small-scale orographic effects, severe convection, and heavy precipitation events. Due to large error growth on the convective scale, predictions of convective-scale models are expected to contain a substantial portion of patterns that are not deterministically predictable. There is urgent need to assist forecasters and other users in the interpretation of such predictions, for example by providing them with uncertainty estimates. Therefore, the German Weather Service is developing an ensemble prediction system based on COSMO-DE. The ensemble development comprises implementation of perturbations, ensemble verification and statistical postprocessing. There is lack of a principal understanding of error, especially with regard to error growth on the convective scale, so that current ensemble generation methods are mainly heuristic. As a forecast provider, the German Weather Service is also concerned with the integration of the additional uncertainty information into the existing forecasting routine. Therefore, the ensemble project also deals with ensemble visualization, the adaptation to subsequent forecasting tools and the coordination with forecasters˘ needs and comprehension. The talk will present the current status of the ensemble development project. |
|