Exchanges - Scientific Contributions - Exchanges - Scientific Contributions - * Contribution to Exchanges No. 28, December 2003 SPEEDO: A flexible coupled model for climate studies * W. Hazeleger * , F. Molteni # , C. Severijns * , R. Haarsma * , A. Bracco # , and F. Kucharski # * Royal Netherlands Meteorological Institute (KNMI), De Bilt,The Netherlands # The Abdus Salam International Centre for Theo- retical Physics (ICTP),Trieste, Italy Introduction In recent years considerable progress in under- standing the climate system has been achieved by using models of intermediate complexity. Those models range from idealized one or two-dimensional representations of the ocean and atmosphere dynamics, to more com- plex 3-dimensional coupled models (see Stocker and Knutti 2003). The advantage of intermediate complex- ity models lies in their computational efficiency. Large ensembles, long runs, twin-experiments and parameter sensitivity studies are easier done than with state-of-the- art GCMs. Intermediate complexity models proved to be valuable in detecting mechanisms of climate variabil- ity, they can help to interpret results from more com- plex coupled integrations, and are a precious tool in the assessment of the statistical significance of predictabil- ity studies (Molteni et al., 2003). Here, we report on the development of a coupled model of intermediate com- plexity that is closer to the state-of-the-art GCMs than previously developed simplified models. The atmos- pheric component is faster than state-of-the-art GCMs by an order of magnitude. A modular setup easily al- lows configuring integrations with different model com- ponents. This makes the model very attractive to study, for instance, the role of oceanic or land processes in cli- mate. In the following, technical aspects of the model are first discussed. The advantage of the modular setup and the hierarchy of ocean models implemented are il- lustrated with results from a study after South Atlantic coupled variability. Technical Aspects The atmospheric module, nicknamed SPEEDY (Simplified Parameterizations primitivE Equation DYnamics, see Molteni 2003 for a description), uses a set of parameterization schemes based on the same prin- ciples adopted in state-of-the-art AGCMs. It is configured with 7 vertical layers and with spectral trun- cation at wavenumber 30 and is computationally very efficient. Despite its relative coarse resolution, SPEEDY has been shown to reproduce reasonably well the ob- served variability of the atmospheric circulation in the 20 th century (Bracco et al., 2003). The land component consists of a land bucket model with interactive tempera- ture, soil moisture, soil ice, snow depth and run off. The ocean component consists of a hierarchy of models that facilitates studying the mechanisms of climate variabil- ity and the role of the oceans therein. Such a hierarchy includes a slab ocean model, a linear ocean model for tropical oceans (Burgers et al. 2001) and a primitive equa- tion isopycnic ocean model (MICOM, Bleck et al. 1992). The slab ocean model can be used in different ways. It can be run in a “qflux” configuration, in which heat fluxes are diagnosed from a run with prescibed SST and then specified, so that only heat transport by the ocean is rep- resented, or including also other processes, such as anomalous Ekman transports, anomalous wind-driven turbulent mixing and anomalous barotropic transport. Regional configurations of these ocean models can be overlayed on the global ocean and land models easily (e.g. it is possible to use MICOM in the Atlantic and a slab ocean in other basins). The coupled model is called SPEEDO (SPEEDy-Ocean). Except for the atmospheric model, all components have been set up in a Generic Model Framework (GMF). In GMF each module has an initialization phase and a time stepping loop. The latter includes time stepping of the model physics, collection and storing of the output data in a history file, writing of a restart file and prepara- tion for the next time step. Generic functions and sub- routines have been developed to implement such a struc- ture in all model components. Each module has a pa- rameter file in which critical parameters such as type of run, calendar, and physical parameters are set. Output and input data are in NetCDF format. SPEEDO uses a distributed coupler implemented as a library that is linked to each component of the coupled system. The modular set up facilitates identifying the role of the dif- ferent components of the coupled system in determin- ing climate variability and the physical mechanisms in- volved. More details can be found in a technical report on the model (Hazeleger et al. 2003) together with a vali- dation of the atmospheric component, which is also avail- able in Molteni et al. (2003) and Bracco et al. (2003).