Statistcal modelling for evaluation and post-processing of climate predictions
- Datum
- 08.05.2015
- Zeit
- 10:30 - 11:30
- Sprecher
- Stefan Siegert
- Zugehörigkeit
- University of Exeter, UK
- Sprache
- en
- Hauptthema
- Physik
- Andere Themen
- Physik
- Host
- Holger Kantz
- Beschreibung
- Predicting future weather is inherently limited to a few days to weeks by the chaotic nature of the atmosphere, and by our incomplete knowledge and imperfect implementation of the governing equations. Certain large-scale features of the atmosphere are, however, predictable on time-scales of several months. One example is the North-Atlantic Oscillation (NAO) index, a climate index related to storm activity and average winter temperatures in Europe. I am going to use a recent data set of retrospective forecasts of winter NAO index to discuss current practice in statistical post-processing of climate forecasts. Post-processing is necessary to correct systematic biases of the climate model. I will then propose a simple statistical framework to describe climate models and their relation to the real world. The parameters of the statistical framework are estimated -- and their uncertainty quantified -- by Bayesian inference. The statistical framework is used to address important questions about the quality and application of the NAO forecasts: Can the numerical model output be used as a substitute for the real world? How certain can we be about the observed correlation between predicted and measured NAO index? How should the deterministic climate model output be transformed into a probabilistic prediction for future NAO? I will conclude by discussing a number of possible extensions of the statistical framework which I am intending to study during my one-month stay at MPIPKS. Main reference: S.Siegert et al (2015): A Bayesian framework for verification and recalibration of ensemble forecasts: How uncertain is NAO predictability? [submitted to Journal of Climate] Preprint: arxiv.org/abs/1504.01933
Letztmalig verändert: 08.05.2015, 09:42:11
Veranstaltungsort
Max-Planck-Institut für Physik komplexer Systeme (Room 1D1)Nöthnitzer Straße3801187Dresden
- Telefon
- + 49 (0)351 871 0
- MPI-PKS
- Homepage
- http://www.mpipks-dresden.mpg.de
Veranstalter
Max-Planck-Institut für Physik komplexer SystemeNöthnitzer Straße3801187Dresden
- Telefon
- + 49 (0)351 871 0
- MPI-PKS
- Homepage
- http://www.mpipks-dresden.mpg.de
Legende
- Ausgründung/Transfer
- Bauing., Architektur
- Biologie
- Chemie
- Elektro- u. Informationstechnik
- für Schüler:innen
- Gesellschaft, Philos., Erzieh.
- Informatik
- Jura
- Maschinenwesen
- Materialien
- Mathematik
- Medizin
- Physik
- Psychologie
- Sprache, Literatur und Kultur
- Umwelt
- Verkehr
- Weiterbildung
- Willkommen
- Wirtschaft