The scientific part of the Module A Coordination project deals with methods of initialisation of coupled climate models.

Improvements in the forecast skill of coupled climate models has to come from improved initial condition and improved initialisation procedures, but also from reducing errors in the coupled model. To further improve decadal predictions therefore requires that we improve the way we determine the present day climate, specifically the ocean state, as initial conditions for a coupled climate model from the existing climate observing system. This implies that we improve the estimation procedure of the ocean state using modern data assimilation approaches, and that we, at the same time, improve the procedure by which we use those estimates of the present-day ocean state as initial conditions of a coupled climate model. That is, to provide an initialisation procedure that is optimal for a given coupled forecast system. The coordination proposal of the Module A Coordination will deal (1) with the determination of initial conditions for a coupled climate model for use in the prototypical system, and (2) with the improvement of the initialisation of forecasts of coupled model systems, trying to use existing initial conditions in a most dynamically-consistent way in the coupled system, thereby avoiding initialisation shocks. The Module-A Coordination will deal also with model improvements through parameter estimation, thereby contribute directly to the overall goal of MiKlip. In detail, the project address two central and unique aspects of MiKlip 1) by improving initial conditions while simultaneously learning how to estimate uncertain model parameters (objective A1, and 2) to improve initialisation procedures and thereby to understand the way to initialise, not only model anomalies, but also the full model state (objective A2).

Goals

• Improving Initial Conditions of Coupled Climate Models: Existing estimates initial conditions need to be improved: they need to be truly global, used improved uncertainty measures for data and models, and to include sea ice estimates as well.

• Improving Initialisation Procedures: much has to be learned to make better used of ocean observations in coupled model systems through initial conditions determined by ocean-only or ocean-sea ice coupled models.

• Improving Assimilation Approaches for the Ocean, Cryosphere: What is required here is to start using coupled systems that include ocean-ice and land-atmosphere subsystems, which can be used to produce global ocean/sea ice initial conditions and land initial conditions.

• Improving Coupled Models through Parameter Optimisation: Using an assimilation infrastructure we can also optimize model parameters such that the model fits best observations. Possible model parameters that need improvement include viscosity and diffusivity in the ocean, but include also the coupling between the ocean and atmosphere, ice parameters, cloud parameters.

• The use of new (satellite) observations during the initialisation procedure (e.g. SMOS, CRYOSAT-II, GOCE): In the near future many additional and important observations are expected, especially from satellites such as SMOS, CRYOSAT-II and GOCE. We have to learn how to incorporate respective information about the ocean, sea ice and land (soil moisture) to improve coupled models and their forecasts.