Cabbeling due to isopycnal mixing in isopycnic coordinate models

The cabbeling that arises as a consequence of isopycnal mixing in a North A tlantic model based on MICOM (the Miami Isopycnic Coordinate Model) is quan tified. Annually averaged over the model Atlantic, the diapycnal volume flu x associated with cabbeling reaches 1.5 Sv, with an associated net density flux of 2 x 10(6) kg s(-1) (equivalent to an annual-basin mean cooling of 0 .6 W m(-2)). Over the range of densities that incorporate the major water m asses of the model Atlantic, cabbeling effectively weakens the density flux due to parameterized diapycnal turbulent mixing by similar to 25%. The str ength of cabbeling varies in proportion to the isopycnal mixing of heat and salt, the local buoyancy frequency, and a "cabbeling parameter" (which is inversely proportional to temperature). As a consequence of these dependenc ies, cabbeling is highly localized and seasonal. In the model, strongest ca bbeling occurs during summer at the subpolar front in the northwest Atlanti c. Model cabbeling arises both physically (due to the independent mixing of he at and salt in isopycnic layers) and. to a lesser extent. nonphysically (du e to the advection of heat and salt). Fields of layer thickness changes due to model cabbeling compare reasonably well with changes predicted by "phys ical" cabbeling. Physical cabbeling is therefore predicted for a global mod el (QGIM) based on a more recent version of MICOM, which features salinity- only advection and mixing (and hence no cabbeling). In the circumpolar Sout hern Ocean of QGIM, intermediate water would be transformed (by cabbeling) to higher density at rates of up to 7 Sv, primarily due to end-of-winter fr eshwater forcing around Antarctica This suggests that the cabbeling associa ted with isopycnal mixing, although neglected in later versions of MICOM, m ay play a significant role in water mass transformation around the Southern Ocean. However. the layer temperature, salinity, and thickness fields used to initialize MICOM lead to unrealistically strong cabbeling around the Me diterranean outflow during the early stages of spinup, a problem which furt her highlights the unsuitability of sigma(0) as a layer variable for water masses below similar to 1000 m.