Atmospheric and Oceanic Data Assimilation Plus ENsembles Generation

To enable climate models for decadal climate predictions is the central aim of MiKlip. The respective climate model needs to resemble the observed climate state and its tendency at the initialisation of the prediction. This is only possible through the incorporation of observations into oceanic and/or atmospheric components of the climate model.

The initialisation of weather prediction systems has a long tradition and is based on sophisticated techniques. The current development in atmospheric initialization is focused on variational methods and better representation of error covariances and their development in time, which is partly covered in 4D variational (4DVar) by singular vectors, but increasingly more based on ensemble initialisation techniques like Ensemble Transform and different types of Ensemble Kalman Filter (EnKF).

The central aim of the project AODA-PENG is threefold. First, we will incorporate state-of-the-art ensemble techniques into the recent but still very simple medium range climate prediction system using the experience gained during the development of a weather prediction system. The consortium formed by MPI for Meteorology (MPI-M), Meteorological Institute University Bonn (MIUB), and the German National Weather Service (DWD) has already successfully developed an ensemble forecasting scheme for the global weather forecast model GME of DWD within the DFG funded priority program PP1167-PQP “Quantitative Precipitation Forecast”. Second, we will apply data assimilation techniques such as 3D variational (3DVar) and EnKF schemes, and latest available observations jointly with ensemble generation by breeding to provide a state-of-the-art initialisation for the planned decadal prediction system. This will be done systematically for the atmospheric and oceanic components. We consider this combination of coupled data assimilation and ensemble generation techniques as the particular strength of our approach. Third, we will analyse the aerosol forcing being an important but still very uncertain external forcing factor on the long time scales of climate projections with respect to its relevance for medium range climate predictions.

Goals

1. Transfer the current operational assimilation methods at DWD from weather forecasting to the climate model components of ECHAM6/MPIOM and the medium range climate forecasting.
2. Apply modern stochastic assimilation methods like ensemble Kalman filtering to the ocean model and its dynamics.
3. Develop an ensemble generation and initialisation strategy based on the ensemble transform breeding approach. Include a process orientated statistical analysis to identify and study the physical processes being of relevance for the growth of uncertainty in decadal predictions in the coupled atmosphere-ocean system.
4. Perform a feasibility study and system analysis of the coupled assimilation / ensemble system to assess the results from (1) – (3), and integrate them into the proposed MiKlip prediction system from.
5. Assess the aerosol physics and chemistry of the ECHAM6/HAM system in close cooperation with (3) to determine uncertainty measures and the relevance of aerosol forcing for medium range climate predictions.