Biases in Thorpe-scale estimates of turbulence dissipation. Part I : Assessments from large-scale overturns in oceanographic data

Abstract

Oceanic density overturns are commonly used to parameterize the dissipation rate of turbulent kinetic energy. This method assumes a linear scaling between the Thorpe length scale LT and the Ozmidov length scale LO. Historic evidence supporting LT ~ LO has been shown for relatively weak shear-driven turbulence of the thermocline; however, little support for the method exists in regions of turbulence driven by the convective collapse of topographically influenced overturns that are large by open-ocean standards. This study presents a direct comparison of LT and LO, using vertical profiles of temperature and microstructure shear collected in the Luzon Strait—a site characterized by topographically influenced overturns up to O(100) m in scale. The comparison is also done for open-ocean sites in the Brazil basin and North Atlantic where overturns are generally smaller and due to different processes. A key result is that LT/LO increases with overturn size in a fashion similar to that observed in numerical studies of Kelvin–Helmholtz (K–H) instabilities for all sites but is most clear in data from the Luzon Strait. Resultant bias in parameterized dissipation is mitigated by ensemble averaging; however, a positive bias appears when instantaneous observations are depth and time integrated. For a series of profiles taken during a spring tidal period in the Luzon Strait, the integrated value is nearly an order of magnitude larger than that based on the microstructure observations. Physical arguments supporting LT ~ LO are revisited, and conceptual regimes explaining the relationship between LT/LO and a nondimensional overturn size are proposed. In a companion paper, Scotti obtains similar conclusions from energetics arguments and simulations.