TOPOGRAPHIC SENSITIVITY STUDIES WITH A BRYAN-COX-TYPE OCEAN MODEL

This paper describes a series of four experiments, each run for 10 yea rs at 1 degrees x 1 degrees resolution on a North Atlantic domain, des igned to illuminate the sensitivity of a Bryan-Cox-type ocean model to changes in the representation of the ridges that restrict the Row of dense, deep water out of the Greenland-Iceland-Norway (GIN) basin. In reality, much of the outflow takes place through narrow sills, which a re subgrid-scale in the model, and small changes in the model topograp hy to reflect these sills have a large impact on the outflow and on th e compensating inflow of warm North Atlantic water. The circulation of the GIN basin is dramatically changed depending on the amount of this inflow; with no inflow, the basin cools and freshens, as would be exp ected, whereas with too much inflow, it becomes warm, salty, and homog eneous to great depths. Moreover, the small changes in topography have wider implications for the simulation. The presence or absence of den se overflows has a great impact on the mixed layer development in the subpolar gyre, with mixed layer depths differing by more than 500 m be tween two of the cases. This has implications for the formation of sub polar mode water, which is nearly shut off in the two cases with signi ficant overflow. The meridional overturning in the model in year 10 in creases by over 50% at its peak between the cases with no dense overfl ow and those with the greatest overflow, and this partly explains a ch ange in peak heat transport, which increases by around 50% in the case s with significant overflow. The results in this paper imply that care ful ''tuning'' of the model topography is necessary in ocean/climate m odels in order to get a reasonable simulation of the conveyor belt and of North Atlantic Deep Water formation.