Observations of interaction between the internal wavefield and low frequency flows in the North Atlantic

Abstract

A total of four moorings from POLYMODE Array I and II were analyzed in an investigation of internal wavefield-mean flow interactions. In particular, evidence for wave-mean flow interaction was sought by searching for time correlations between the wavefield vertically-acting Reynolds stress (estimated using the temperature and velocity records), and the mean shear. No significant stress-shear correlations were found at the less energetic moorings, indicating that the magnitude of the eddy viscosity was under 200 cm2/sec, with the sign of the energy transfer uncertain. This is considerably below the 0(4500 cm2/sec) predicted by Müller (1976). An extensive error analysis indicates that the large wave stress predicted by the theory should have been clearly observable under the conditions of measurement. Theoretical computations indicate that the wavefield "basic state" may not be independent of the mean flow as assumed by Müller, but can actually be modified by large-scale vertical shear and still remain in equilibrium. In that case, the wavefield does not exchange momentum with a large-scale vertical shear flow, and, excepting critical layer effects, a small vertical eddy viscosity is to be expected. Using the Garrett-Munk (1975) model internal wave spectrum, estimates were made of the maximum momentum flux (stress) expected to be lost to critical layer absorption. Stress was found to increase almost linearly with the velocity difference across the shear zone, corresponding to a vertical eddy viscosity of -100 cm2 s -1. Stresses indicative of this effect were not observed in the data. The only significantly non-zero stress correlations were found at the more energetic moorings. Associated with the 600 m mean velocity and the shear at the thermocline were a positively correlated stress at 600 m, and a negatively correlated stress at 1000 m. These stress correlations were most clearly observable in the frequency range corresponding to 1 to 8 hour wave periods. The internal wavefield kinetic and potential energy were modulated by the mean flow at both levels, increasing by a factor of two with a factor of ten in the mean flow. The observed stress correlations and energy level changes were found to be inconsistent with ideas of a strictly local eddy viscosity, in which the spectrum of waves is only slightly modified by the shear. When Doppler effects in the temperature equation used to estimate vertical velocity were considered, the observations of stress and energy changes were found to be consistent with generation of short (0.4 to 3 km) internal waves at the level of maximum shear, about 800 m. The intensity of the generated waves increases with the shear, resulting in an effective vertical eddy viscosity (based on the main thermocline shear) of about +100 cm2 s-1 The stresses were not observable at the 1500 m level, indicating that the waves were absorbed within 500 m of vertical travel. The tendency for internal wave currents to be horizontally anisotropic in the presence of a mean current was investigated. Using the Garrett- Munk (1975) model internal wave spectrum, it was found that critical layer absorption cannot induce anisotropies as large as observed. A mechanical noise problem was found to be the cause of large anisotropies measured with Geodyne 850 current meters. It could not be decided, however, whether or not the A.M.F. Vector Averaging Current Meter is able to satisfactorily remove the noise with its averaging scheme.