Task 1.2 Model-consistent Initialisation

Atmosphere and ocean initialisation is essential for seasonal to decadal climate forecasts (e.g., Polkova et al., 2019). The experiences from the initialisation of weather forecasts are valuable, but only to a limited extent transferable to decadal predictions (e.g., Brune and Baehr, 2020). While the ICON-Consortium possess tremendous experience in the initialisation of weather forecasts, Coming Decade can contribute to the development of the ICON-XPP prediction system its experience in the initialisation of decadal predictions, which is the focus of task 1.2. For ICON, DWD and UHH are currently working in the DWD-funded projects DAKLIM and GCFS3.0 to advance the ocean initialization in ICON-XPP to an ensemble Kalman filter (based on the MiKlip system, Pohlmann
et al., 2022).

To complement these efforts, the task proposed here will adapt the data assimilation scheme currently operational at DWD for the weather forecast time scale in the ICON atmosphere model, also based on an ensemble Kalman Filter approach, to decadal requirements. We will pay particular attention to the question of which data are most useful for initializing climate forecasts whereas the DWD projects focus mostly on the technical implementation of the data assimilation workflow. Although Coming Decade will initially refine the initialization of the atmospheric component, the focus of the project will be on a joint initialization of the atmospheric and ocean components, which Coming Decade will ensure to be incorporated into ICON-XPP , including a consistent use of the technical framework. An early version will be part of the initial operational ICON-XPP system, and a revised version will be part of the refined ICON-XPP system delivered to DWD at the end of Coming Decade.

Contributors

University of Hamburg
Prof. Dr. Johanna Baehr
Dr. Sebastian Brune

Max Planck Institute for Meteorology
Dr. Holger Pohlmann

Deutscher Wetterdienst
Dr. Kristina Fröhlich
Dr. Trang van Pham