Mixing and dynamics of the Mediterranean outflow

Abstract

Hydrographic and expendable current profiler (XCP) data taken during the Gulf of Cadiz Expedition in September 1988 are analyzed to diagnose the mixing and dynamics of the Mediterranean outflow. The overall structure of the outflow is consistent with that described in the historical literature (Heezen and Johnson, 1969). This data shows that the overflow transport doubles from .85 Sv to 1.9 Sv, and that the velocity weighted salinity decreases from 37.8 pss to 36.7 pss in the first 60 km of the path. The core salinity of the neutrally buoyant outflow near Cape St. Vincent is 36.6 pss, which indicates that most of the mixing has taken place close to the Strait in the initial descent of the outflow. Cross stream variations in the overflow T/S properties increase as the flow spreads from 10 km to 90 km wide. The outflow begins with less than a 0.5°C across-stream variation in temperature in the Strait with the saltiest, coldest water to the south and slightly fresher and warmer outflow to the north. As the outflow spreads, the northern near-shelf flow follows a path higher in the water column and mixes with warmer North Atlantic water than does the deeper offshore flow. Within the first 100 km, the cross stream variation in temperature on an isopycnal becomes more than a 2°C. The flow eventually settles along two preferred isopycnals: 27.5 and 27.8 (Zenk 1975b). The spreading of the flow contains both a barotropic and baroclinic character. The average change in angle above and below the maximum velocity of the outflow is 8°while at the edges of the flow the average direction of the outflow diverges by as much as 50°. Gradient Richardson numbers less than 1/4 are found in the interface (up to 50 m thick) between westward flowing Mediterranean water and eastward flowing North Atlantic water, even though there is a strong stabilizing stratification present. Bulk Froude numbers greater than 1 are found near the Strait coincident with the vigorous mixing noted above. Lower bulk Froude numbers were observed in regions where less entrainment was taking place. The momentum balances are diagnosed using hydrographic and XCP data. Evaluation of the cross stream momentum balance shows the importance of advection as the flow makes a 90 degree inertial turn upon entering the Gulf of Cadiz. A form of the Bernoulli function can be evaluated to infer the total stress (entrainment and bottom drag) acting on the outflow. This stress is as large as 5 Pa within 20 km of the Strait, while further downstream the stress decreases to about 1/2 Pa. The entrainment stress estimated from the property fluxes reaches a maximum of about 0.8 Pa near section C, indicating that bottom stress is dominant. Near the Strait, advection, bottom drag and the Coriolis force are all critical to the dynamics of the outflow. Further downstream, the outflow becomes a damped geostrophic current. A simple geostrophic adjustment model is used to show that in the absence of frictional stresses, the outflow would very quickly become geostrophically balanced and descend only about 10 m down the continental slope. Thus, friction is critical for the outflow to cross isobaths. A simple numerical model that uses a Froude number dependent entrainment and a quadratic bottom friction law is used to simulate the outflow (Price and Baringer, 1993). Some of the properties of the outflow including localized entrainment, large stresses and high Rossby number of the flow (initially as high as 0.6), are simulated rather well, though the model overestimates the magnitude of the outflow current. We suspect that this is a consequence of assuming a passive ocean. Two different methods for specifying the broadening of the flow are compared: one using the highly parameterized concept of Ekman spreading, the other using the conservation of potential vorticity. The potential vorticity broadening more accurately reproduces the observed width of the flow near Cape St. Vincent where the width varies inversely with the bottom slope. However, both methods produce essentially the same equilibrium temperature, salinity and transport of the outflow which is a testament to the robustness of the model solution. with the formation process of NADW.