The present thesis documents the design and development of a three-dimensional prognostic ocean/sea-ice model of the Arctic Mediterranean and the North Atlantic. The model has been set up on the basis of the z-coordinate ocean model 'MOM 2' coupled to a dynamic/thermodynamic sea-ice model with viscous-plastic rheology. To implement the inflow of river water and other freshwater sources, a surface boundary condition is employed that accounts for the input of volume ('open surface'). Adding arctic freshwater sources step by step leads to a progressive improvement of the coupled model, and allows to analyse the sensitivity of the ocean/sea-ice system with respect to freshwater forcing.

The results reveal that freshwater plays a major role in Arctic Ocean dynamics. In particular, the path of the Transpolar Drift (TPD) and the strength of the East Greenland Current (EGC) in the western Fram Strait are strongly influenced by the input of freshwater. Thus, freshwater favours the exchange of water masses between the Nordic Seas and the Arctic Ocean. Moreover, freshwater input controls vertical oceanic heat fluxes into the ice by forming a stable density stratification. The model requires a total freshwater input of approx. 6800 km^3/yr to the Arctic Ocean in order to maintain a realistic hydrography. More than 40% of this freshwater leaves the Arctic Ocean as sea-ice through Fram Strait. The sum of liquid freshwater exports through Fram Strait and the Canadian Arctic Archipelago is of similar magnitude (ca. 1500 km^3/yr through each passage).

Introducing passive tracers into the model allows to detect the pathways of arctic river and Pacific waters. The bulk of river water is transported by the TPD from the Siberian shelf seas to the Nares and Fram Straits where it leaves the Arctic Ocean. A river-water storage of 46 10^3 km^3 along with a river-water influx of 3159 km^3/yr yields a mean residence time of 14-15 yr for river water in the upper 300 m of the Arctic Ocean.

Pacific water dominates the upper layers in the western Arctic (Chukchi Sea, East Siberian Sea, Canadian Basin) with concentrations exceeding 50%. The volume flux of about 0.8 10^6 m^3/s through Bering Strait is balanced by an increased outflow through Fram Strait. The inflow of relatively warm, low-saline Pacific water into the Arctic Ocean leads to a considerable reduction in Chukchi and East Siberian sea-ice covers, and results in a 20% weakening of the Atlantic meridional overturning circulation.

Taking the volume input of surface freshwater fluxes into account by applying an 'open surface', the model presented here is superior to other models of the Arctic Mediterranean, which are driven by 'virtual salinity fluxes'. Experiments with different salinity/freshwater flux boundary conditions reveal the shortcomings of salinity-flux formulations. It is concluded that other prognostic models of the Arctic Ocean can be improved substantially by implementing an 'open surface'.