Institute of Mathematics > Staff > Dr. Annette Möller > DFG project Ensemble Postprocessing

DFG project Ensemble Postprocessing

Network members

 

 

 

Network Project

Today, so-called numerical weather prediction (NWP) models are typically used for weather prediction. They consist of a system of differential equations describing the state of the atmosphere as accurate as possible, which are integrated in time to obtain predictions of future atmospheric states.  Typically, the NWP models are run multiple times, each time with different initial conditions and/or model formulations to represent the uncertainty in these quantities. This results in an ensemble of forecasts, as each model run yields a single deterministic forecast.

However, ensemble forecasts are often uncalibrated and require so-called statistical posprocessing. Here, statistical models are applied to the ensemble forecasts in conjunction with observations to improve the properties of the forecasts and to explicitly quantify forecast uncertainty.

Aim of the network cooperation is to develop new and improved statistical methods for ensemble postprocessing.

Specifically it is planned to investigate the following topics

  • Modification of existing models for normal distributed weather variables as temperature for other weather variables, such as skewed distributed wind speed or precipitation which is often modelled by a mixture distribution
  • Extension of existing models to the multivariate setting, that is incoropration of spatial, temporal or inter-variable dependencies into the models
  • Application of Machine Lerning methods in the context of ensemble postprocessing, specifically investigation of benefit of Machine Learning for discrete variables such as cloud cover
  • Implementation of the new models, combining new implementations with already existing ones

 

 

 

Previous publications of network members in the area of ensemble postprocessing

 

Current publications of network members

 

Sitemap  Contact  Data Privacy  Imprint
© TU Clausthal 2020