Space and time scales of mesoscale motion in the western North Atlantic

From moored data, primarily temperature, of the Mid‐Ocean Dynamics Experiment (Mode I) and its successor experiments we find a statistical description of the mesoscale variability. In the Mode I area itself the spectral characteristics of the thermocline and the deep water are different. The thermocline is conveniently described as being made up of three spectral bands: a ‘low‐frequency’ band dominated by zonal velocity fluctuations, an ‘eddy‐containing’ band in which the velocity field is nearly isotropic, and a ‘high‐frequency’ band consistent with models of geostrophic turbulence. In the deep water the zonal dominance at low frequencies is not apparent, and there is enhanced energy at periods of 20–50 days. Vertical structure scales with WKBJ approximation in the high‐frequency band but not in the lower frequencies, where low vertical modes dominate the motion. Linear models do not adequately describe the data in the Mode I region. Differences between rough and smooth topography regions are clearly seen only at 1500 m, where there is a loss of energy consistent with a reduced barotropic motion. Other differences, while apparently real, are small. It is found, consistent with the results of Schmitz (1976 a ), that the Mode I region is atypical of the midocean in that large changes of energy level are found elsewhere. A region due east of Mode I has slightly reduced kinetic energy levels in the main thermocline, but deep energy levels are much lower. Potential energy is less variable than kinetic; in the eastern region the frequency spectra change structure slightly. Linear models may be more adequate there. With more than 2 years of data, no statistically significant heat flux was found in the Mode I area, except for a weak zonal flux in the deep water. There is no direct evidence for baroclinic instability as a significant mechanism of eddy generation; the Gulf Stream is a possible, if unconfirmed, source.