An important aspect of decadal climate prediction is the necessity of capturing and predicting internal modes of variability that evolve on decadal and longer times scales. The importance of internal variability sets decadal prediction apart from scenario calculations. The problem can only be solved by initialising a climate model with a realistic state description that contains information about the tendencies in heat and fresh water reservoirs.
Modini will develop a new method to initialise the ocean and sea ice components of a coupled climate model. The method is aimed to be simple and cost effective as well as flexible. The project will test the skill and the predictive potential with such initialization which represents a simple method of data assimilation into a coupled climate model while still avoiding artificial adjustment processes during the prediction period.
State estimation of ocean and sea ice using data assimilation could in principle be used to achieve a dynamically consistent state of these climate system components that is close to the observed state. However, alternative methods must be explored because of the computational expense and the huge logistical problems to include subsurface ocean data and sea ice drift and thickness data.
Ocean-sea ice models run in a forward mode without data assimilation are more constrained by observations than coupled climate models. The atmospheric forcing provided by actual atmospheric variability allows ocean-sea ice models to reproduce many observed features. Because of missing feedbacks between the evolving ocean conditions and the specified atmospheric forcing, these models are not suitable for coupled climate model initialization.
Modini will pursue a method to initialize coupled climate model decadal prediction runs by using a spin-up of the same comprehensive coupled model. The system is brought close to the observed state in the beginning of the prediction by using a time series of observed wind stress anomalies to force the ocean-sea ice subsystem. Heat and fresh water fluxes are computed interactively by the coupled model. The method can be applied as an initialisation tool for the global ice-ocean component of a coupled model. In addition to wind stress forcing, we will also explore the use of other forcings for the ocean-ice system (e.g. surface heat and freshwater fluxes).
Modini has developed a concept, whereby the ocean component of a coupled climate model is driven by the time series of anomalies of observed wind stress added to the wind stress climatology from the coupled model. The resulting output is used to initialise the climate model in order to produce predictions.
The concept has been implemented and tested in the Kiel Climate Model (KCM) and the MPI-ESM. The advantage of this concept is that one can almost completely avoid model drift at the beginning of the prediction. This method is an example of “partial coupling“ and it is able to reproduce the observed climate variability in the North Pacific (Ding et al., 2013a). It could further be shown that the interdecadal changes in the feedback between ENSO and the East Asian summer monsoon results from atmospheric noise. In a further paper (Ding et al. 2013b), it was shown that the partially coupled model system is able to successfully hindcast (historically forecast) the 1976/1977 and 1998/99 climate shifts in the Pacific (see Figure).
The implementation of this method into the MPI-ESM is finished and the results are currently being analysed. The concept is compatible with the MiKlip system.
Ding, H., Greatbatch, R. J., Latif, M., Park, W. and Gerdes, R. (2013a) Hindcast of the 1976/77 and 1998/99 climate shifts in the Pacific Journal of Climate, 26 . pp. 7650-7661. doi: 10.1175/JCLI-D-12-00626.1.
Ding, H., Greatbatch, R. J., Park, W., Latif, M., Semenov, V. and Sun, X. (2013b) The variability of the East Asian Summer Monsoon and its relationship to ENSO in a partially coupled climate model Climate Dynamics . doi: 10.1007/s00382-012-1642-3.
Alfred Wegener Institut for Polar and Marine Research (AWI - Bremerhaven)
Prof. Dr. Rüdiger Gerdes
Prof. Dr. Richard J. Greatbatch
Ding, H. | R.J. Greatbatch, M. Latif,W. Park, and R. Gerdes