Aplinkos tyrimai, inžinerija ir vadyba, 2007.Nr.1(39), P. 3-7 ISSN 1392-1649 Environmental Research, Engineering and Management, 2007.No.1(39), P. 3-7 Operational Three Dimensional Ocean Modelling for the Baltic Sea Jun She, Per Berg, Jesper Larsen and Jacob Woge Nielsen Centre for Ocean and Ice, Danish Meteorological Institute, Copenhagen, Denmark (received in June, 2006; accepted in March, 2007) This paper discusses the major issues in operational three dimensional (3D) ocean modelling for the Baltic Sea, based on the experiences in DMI Centre for Ocean and Ice in recent years. Both scientific and technical aspects of the operational ocean modelling are discussed. Recommendations are given to the development of the Baltic Sea operational 3D ocean models. Key words: the Baltic Sea, 3D Ocean modelling 1. Introduction The issues concerned in operational modelling for the Baltic Sea cover both scientific and technical aspects. The purpose of the scientific treatment of an operational model is to ensure a correct theoretical and numerical framework, including two parts: physical processes and a numerical scheme. The model has to resolve major physical processes (e.g., 3D mixing and diffusion, advection and sub-grid processes, etc) correctly. In addition to this, the model has to be discretized correctly (i.e. with the least possible numerical error). The purpose of the technical treatment in the operational modelling is to generate an operational product with the highest possible quality and the shortest possible integration time. The major concerns in the technical aspect are:  Prescribe the forcing data (e.g. weather, river runoff, bathymetry and lateral boundary conditions, etc.) with the highest possible quality,  Select a proper resolution taking into account the following factors: − Reduction in a model error, − Resolving new physical processes (e.g., eddy/front dynamics), − Increase in computing cost and − End user needs.  Improve the computing efficiency through: − Optimizing the model code in terms of handling memory, compiling and code redundancy, − Parallelizing the model, − Choose a proper computing platform for the model.  Effectively use the observations in model validation and data assimilation. Many 3D ocean models have been applied for the Baltic Sea, e.g., POM, BSHcmod, HIROMB, MIKE3, GETOM, MIT GCM, etc. However, only some of them are used in operational forecast, e.g. BSHcmod, HIROMB and MIT GCM. It has been found that most of the operational models can simulate the transient and barotropic signals reasonably well but has a problem in modelling the stratification. In the following Sections, the above scientific and technical issues are discussed based on mainly the operational modelling experiences in DMI Centre for Ocean and Ice. Due to the limited time of writing this paper, the authors do not intend to form it as an in-depth scientific publication rather than to highlight some major issues in operational ocean modelling and stimulate the discussion in this field. 2. Scientific aspects The Baltic Sea is driven by the momentum and heat fluxes from the atmosphere, seasonal ice, river runoff, bottom topography, coastline and boundary flows through Danish Straits. The external forcing is transferred