THE MERIDIONAL OVERTURNING CELLS OF A WORLD OCEAN MODEL IN NEUTRAL DENSITY COORDINATES

A comparison is made of the meridional overturning circulation in a co arse-resolution World Ocean model when the integration is performed al ong (i) level, (ii) potential density, and (iii) neutral density surfa ces. In the level-surface calculation, all the usual cells are evident , including the Atlantic ''conveyor,'' the Deacon cell, and the direct Antarctic cell. In the potential or neutral density calculations, all cells remain present; however, the Deacon cell is greatly reduced in strength (to just a few Sverdrups). An analysis of the thermodynamics underlying the dianeutral motion is conducted. Most dianeutral motion results from fluxes associated with the vertical diffusivity and the ( unphysical) horizontal diffusivity. Caballing is not important, despit e the inclusion of isopycnal diffusivity. The mechanism of the residua l Deacon cell involves densification near 40 degrees S resulting from fluxes associated with the horizontal diffusivity. Horizontal diffusiv ity results in substantial dianeutral motion in several other parts of the ocean. Most significant is motion toward lesser density in the fa r Southern Ocean, which integrates zonally and between 67 degrees S an d 57 degrees S to give a transport of about 25 Sv across density surfa ces. This transport dominates other dianeutral transports at high dens ity in the ocean interior and indicates serious distortion of the solu tion by the horizontal diffusivity. A second model run is conducted wh ere the horizontal diffusivity is reduced to near the (experimentally determined) limit for the numerical integrity of water properties on t he large scale. Dianeutral transports associated with horizontal diffu sivity generally decline modestly. In neutral density coordinates, the Deacon cell now vanishes almost completely. The Deacon cell of the le vel-surface integration results mainly from large-scale isopycnal moti ons occurring on sloping density surfaces, which superpose to yield a cell upon zonal integration at constant depth. Finally, it is apparent that the neutral density coordinate provides a clearer picture of the ocean circulation than do potential density coordinates, because of i nherent ambiguity in choosing the reference pressure of potential dens ity.