Variability of water masses and circulation in the subtropical North Atlantic
Observations of interannual variability in 18 Water (Talley and Raymer 1982) and Gulf Stream transport (Worthington 1977) motivate an ocean model sensitivity study. The North Atlantic circulation is simulated with a three-dimensional isopycniccoordinate GCM. Idealized anomalous buoyancy-forcing fields (associated with outbreaks of cold, dry continental air over the Gulf Stream/Sargasso Sea region) are constructed. In a series of sensitivity experiments, wintertime buoyancy loss over the Gulf Stream and Sargasso Sea is thus increased to varying degrees, with anomalous ocean-to-atmosphere buoyancy fluxes of up to double climatological values. Under excess buoyancy loss, winter mixed layer depths increase, and a greater volume of model 18 Water is formed. End-of-winter mixed layer density also increases, leading to the formation of a denser variety of 18 Water. The anomalous 18 Water recirculates around the Sargasso Sea as a signal of low potential vorticity, which spreads out and weakens on a decadal timescale. Strengthened horizontal pressure gradients in the vicinity of the anomalous 18 Water drive intensified baroclinic transports at the "immediate" end of winter (in March), after which a full-depth barotropic intensification of the Gulf Stream develops. Strongest intensification occurs in May, when the Gulf Stream barotropic transport is increased locally by up to 10 Sv. The anomalous transports which account for barotropic intensification are confined to deep and abyssal layers of the model. Where the associated anomalous bottom currents traverse isobaths, "extra" bottom pressure torque (BPT) is invoked. An anomalous BPT term in the barotropic vorticity balance may therefore account for the intensification. Computed from the model fields of density and sea surface height, such a term does appear to produce the extra negative vorticity associated with anticyclogenic intensification. It is concluded that wintertime excess buoyancy loss drives a springtime barotropic response of the subtropical gyre, through BPT, due to "JEBAR" (the Joint Effect of Baroclinicity And Relief). The Gulf Stream intensification decreases after May as lateral eddy mixing weakens anomalous cross-stream pressure gradients. This eddy mixing is parameterized in the model by a layer thickness diffusion velocity, ud, nominally chosen to be 1.0 cm s'1. Further experiments establish the sensitivity of intensification strength to the choice of ud. With ud = 0.1 cm s"1 (weak thickness diffusion), the intensification is increased by ~50%, while, for ud - 10 cm s"1 (strong thickness diffusion), the intensification is roughly halved. These further sensitivity experiments also reveal the varying degrees to which the model subtropical gyre can be dominated by diffusive eddy mixing (Rhines and Young 1982a, 1982b) or adiabatic (nondiffusive) ventilation of the thermocline (Luyten, Pedlosky and Stommel 1983). Recent (1980-97) interannual variability in the formation and recirculation of 18 Water, and other water masses, is deduced from observed surface heat and freshwater fluxes. Interannual variations in the strength of 18 Water renewal (thus deduced) and a wintertime index of the North Atlantic Oscillation (NAO) are found to be strongly anticorrelated (with a correlation coefficient of -0.70, statistically significant at a 99% confidence level). A further sensitivity experiment establishes that anomalous wind forcing, characteristic of a minimum phase in the NAO, does not intensify the Gulf Stream in the manner of excess cooling. It is concluded that 18 Water is more strongly renewed, with accompanying Gulf Stream intensification, under NAO-minimum buoyancy forcing.