Coupling of a High-Resolution Atmospheric Model and an Ocean Model for the Baltic Sea

Abstract

The coupling between a high-resolution weather forecasting model and an ocean model is investigated. It is demonstrated by several case studies that improvements of short-range weather forecasting in the area of the Baltic Sea require an accurate description of the lower boundary condition over sea. The examples are taken from summer situations without sea ice as well as from winter situations with extreme sea ice conditions. It is shown that the sea state variables used in the model influence the weather forecast both directly on the local scale due to the local impact of surface fluxes of latent and sensible heat and on regional and larger scales. The convective snowbands during winters with cold airmass outbreaks over the open water surfaces of the Baltic Sea are extreme examples of the influence of sea state variables on a regional scale. It is furthermore demonstrated that the sea state conditions may change considerably within forecasting periods up to 48 h. This implies the necessary application of ocean models, two-way interactively coupled to the weather forecasting model. The coupling of an advanced 2.5-dimensional ice?ocean model to the operational Swedish Meteorological and Hydrological Institute (SMHI) weather forecasting model HIRLAM is described. The ice?ocean model includes two-dimensional, horizontally resolved ice and storm surge models and a one-dimensional, vertically resolved ocean model applied to 31 Baltic Sea regions. The coupled model system is applied operationally in a data assimilation system at the SMHI. No data assimilation is applied in the operational ocean component; manual modifications to the sea state variables are introduced a few times every winter season. The application of this operational coupled model data assimilation system to the mesoscale reanalysis for the Baltic Sea Experiment (BALTEX) shows that it is necessary to apply data assimilation for the sea state variables in order to avoid drift of the coupled model system toward less realistic model states. A successful application of a simple assimilation of SST observations is presented. The observed SSTs are first subject to a horizontal filter in order to minimize the effects of observational errors and to restrict the influence to a larger horizontal scale. Then the differences between these filtered temperature observations and the model SSTs are used to construct a modified sensible heat flux that is applied as a form of a ?nudging? term to the ocean model. It turns out that this ?nudging? is successful in avoiding the drift away from realistic sea state conditions. The described atmosphere and ocean data assimilation scheme has been applied in a rerun of the BALTEX mesoscale reanalysis for the cold winter 1986/87. The quality of this reanalysis was assessed through the successful simulation of the convective snowbands in January 1987.