Simulation of Ocean Temperature and Heat Storage from a GCM Coupled to a Variable Depth Upper Ocean

Abstract

The simulation of zonal mean ocean surface temperature and heat storage (time rate of change of heat content) obtained from a low resolution, atmospheric general circulation model (GCM) coupled to a variable depth upper-ocean formulation (A/VO model) is compared with observations and with a parallel simulation using a constant (50 m) depth upper-ocean component (A/CO model). Variable depths are used to specify the heat content in the seasonal mixed layer and thermocline; the depths are prescribed to vary latitudinally and seasonally but not longitudinally. The Northern Hemisphere depths are taken from Meehl; for the Southern Hemisphere, Meehl's depths are modified to account for differences in ocean thermal structure between the hemispheres. The annual variation of upper ocean depth produces important effects on seasonal simulation of ocean surface temperature in the extratropics. In this study, the extremes of ocean temperature in both hemispheres occur earlier by about 30 days in the A/VO model, compared to the A/CO model. This leads to a warmer ocean in spring/summer and a colder ocean in fall/winter for the A/VO model. In contrast to the A/CO model, an important asymmetry in the structure of the monthly departures of ocean temperature from the annual mean is produced by the A/VO model. In the southern extratropics, the A/VO model simulates a reduced annual cycle of ocean temperature (by about 25?30%) from that produced by the A/CO) model. Both models underestimate the observed seasonal amplitude of zonal mean heat storage; the underestimation is greatest for the A/VO model. The differences in simulations of heat storage are linked to differences in ocean surface temperature computation between the two models. Errors in zonal mean ocean surface temperature for the A/VO model are less than for the A/CO model in the extratropics. particularly the Southern Hemisphere. However, the errors in the coupled A/VO model simulation are larger in the northern extratropics than in the previously published uncoupled calculations using prescribed heat storage estimates. It is argued that significant improvement in ocean temperature simulation by the A/VO model can be achieved with better approximation of heat storage in conjunction with a small adjustment to the prescribed variable depths.