Institut für Meteorologie und Klimaforschung

Karlsruher Meteorologisches Kolloquium

Dozenten: Prof. Dr. T. Leisner, Prof. Dr. P. Braesicke, Prof. Dr. A. Fink, PD Dr. M. Höpfner, Prof. Dr. C. Hoose, Prof. Dr. P. Knippertz, PD Dr. M. Kunz, Prof. Dr. J. Pinto 

Veranstaltungskalender

 
Vortrag

Decadal regional climate simulations 1961-2010: A Bayesian recalibration approach

Dienstag, 30. April 2013, 16:30-17:30
KIT Karlsruhe, Campus Süd,
Institut für Meteorologie und Klimaforschung
Geb. 30.23, Physikhochhaus, 13. Stock
Wolfgang-Gaede-Str. 1
Within the German BMBF funded MiKlip (Mittelfristige Klimavorhersage) programme a subproject explores the feasibility, the added value and the uncertainty range of decadal regional climate forecasts as well as the spatial and temporal variation of the predictive potential. For that purpose, predictive decadal hindcasts (1-10 years) for Europe are generated in high resolution with a regional climate model (here CCLM) using global model predictions (ECHAM6) for the past decades (1961-2010). In this study we focus on winter precipitation in Germany with the aim to improve the forecast by statistical post-processing or recalibration. The idea of recalibration is to extract statistical information about the model performance from hindcasts. Accordingly these are applied to the forecast under the assumption of “similar” statistical behavior of the model in the past and future. Since there are very few experiences in the novel field of decadal predictions, it is up to now not clear, if recalibration is an option to improve the predictability. Further, several recalibration methods have been developed under the framework of numerical weather prediction and seasonal forecasts and are applied to time scales of days to months. Thus, it is not clear if these methods could be useful for decadal simulations. In our approach we have developed a Bayesian Model Averaging (BMA) algorithm, which has been modified to handle decadal predictions, hence the name Decadal BMA (DBMA). First results on artificial test data and a real model ensemble reveal that standard recalibration methods from the field of seasonal forecasts are probably not able to improve decadal predictions, whereas the DBMA method could be useful for improving decadal forecasts.
Diese Veranstaltung ist Teil der Reihe Karlsruher Meteorologisches Kolloquium
Referent/in
Dr. Sebastian Mieruch

Karlsruher Institut für Technologie (KIT)
Institut für Meteorologie und Klimaforschung
Veranstalter
Institut für Meteorologie und Klimaforschung
Karlsruher Institut für Technologie
Kaiserstraße 12
76131 Karlsruhe
E-Mail: sekr does-not-exist.imk-tro kit edu
https://www.imk-tro.kit.edu/
Servicemenü

Hinweise

"CS" - KIT-Campus Süd (Universität), Gebäude 30.23 (Physikhochhaus), Seminarraum 13/2

"CN" - KIT-Campus Nord (Forschungszentrum), Gebäude 435 (IMK), Raum 2.05

(Besucher bitte Personalausweis mitbringen!)