Homepage

Project title:

Impact of fundamental assumptions of probabilistic data assimilation/ensemble forecasting: conditional mode vs. conditional mean

Researchers:

• Milija Zupanski - Principal Investigator
• R. Arif Albayrak

Project description:

The relative value of estimating the conditional mean vs. the conditional mode will be examined for the first time in realistic applications of ensemble data assimilation/forecasting, using an operational (e.g., National Centers for Environmental Prediction (NCEP) numerical weather prediction (NWP) model and observations. Potential computational savings, relevant to operational implementation of an ensemble data assimilation/forecasting system will also be evaluated. The proposed research includes the following tasks:

• Add the NCEP's global operational NWP model and data assimilation observation operator to the ensemble data assimilation algorithm
• Compare the performance of the conditional mean vs. conditional mode systems, with and without the estimate of model error
• Assess the forecast PDF skill
• Evaluate the dual-resolution ensemble data assimilation potential for operational applications.

The proposed collaborative research is a three-year project, addressing the Data Assimilation and Observing Strategies, and the Predictability, Dynamical Processes, and Forecast Procedures sub-programs of the TORPEX project. This research will directly address the issue of probabilistic forecasting skill of THORPEX, and the means for the skill improvement. As a result of this study, a better understanding of the ensemble (probabilistic) methodology will be achieved, eventually leading to improved operational probabilistic NWP system.