Adjustment of the Ocean under Buoyancy Forces. Part II: The Role of Planetary Waves

Abstract

A numerical ocean general circulation model is used to investigate the early stages in the adjustment to equilibrium of an ocean initially at rest with imposed uniform meridional potential temperature gradients, which yield density gradients representative of those observed in the North Atlantic. The main feature of the adjustment during the early stages (the first year) is the formation and decay of a ?subtropical? (warm core) and a ?subpolar? (cold core) density gyre. The gyres are formed by the irreversible winding-up of the initially zonal isotherms by first baroclinic mode, coastally-trapped, dissipative Kelvin waves. This phase and the north?south asymmetry arising from the variation in viscous?diffusive Kelvin wave properties with latitude were discussed in Part I of this series. Within a month ?-effects become significant, especially in the evolution of the southern gyre, which develops a distinct east?west asymmetry through western intensification and long planetary wave propagation of boundary information from the east. In the north, penetration of boundary information into the interior is attributed to diffusive effects, because the maximum allowable planetary wave frequency is lower than that of the Kelvin wave signal. This phase in the adjustment is described in Part A of the present paper. A simple model based on the quasi-geostrophic potential vorticity equation for a single vertical mode is used to explain and quantitatively assess the effects of finite spatial resolution and viscous and diffusive processes on the Kelvin wave?planetary wave dynamics. The longshore temperature gradients at the middle depths of the ocean are considerably reduced by first baroclinic mode Kelvin wave propagation. The adjustment then enters another phase characterized by the destruction of the density gyres by either planetary wave or diffusive processes, there being no further significant Kelvin wave propagation. This phase is described and investigated in Part B, using the simple model developed in Part A with modified boundary conditions.