A key process in the maintenance of primary production in the
summer stratified regions of continental shelf seas is vertical mixing in the
thermocline. Existing knowledge of the processes involved is limited and the
current generation of turbulence closure models fails to accurately simulate
mixing in stratified regimes.
The aim of this study was to seek better understanding of the
processes involved in internal mixing and to improve on current
parameterisations of turbulence. A series of new measurements of the vertical
structure of density, velocity, and the dissipation rate of turbulent kinetic energy, c:, were made at a site in the interior of the Celtic Sea in the summer of 2003. Measurements of c: reveal a thermocline clearly separated from bottom boundary layer turbulence by a low energy mid-water region, O[10-6Wm3]. In the thermocline dissipation rates are weakly enhanced over this mid-water minimum and punctuated by sporadic maxima which elevate c: by an order of magnitude. The average thermocline dissipation rates and diapycnal diffusivity during this study were 6.0 x 10-5 (±0.19) wm-3 and 0.46 (±0.25) cm2
s-1 respectively. These levels of mixing were found to be sufficient to play a critical role in the vertical structure of shelf seas and when combined with local measurements of the vertical nitrate gradient appeared sufficient to account for up to 40% of the annual new production of carbon in the region.
Two potential sources of energy have been identified; internal wave
energy and inertial oscillations. By direct measurement of the baroclinic energy
flux the internal wave field was found to be generally weak except during the
passage of an internal tide propagating towards the west with an associated
energy flux of 45 wm-1. Inertial oscillations were present to varying degrees
throughout the observations and contributed significantly to a strong and
persistent thermocline shear layer.
The transmission of energy from the candidate mixing mechanisms
to turbulence is not clear. Shear instability (Rig< ¼) is measured in the
thermocline for only 2.2% of the time; however, the majority of the thermocline
to be only marginally stable, held within a narrow band of low Richardson
numbers with 75% of observations falling within 0.25 :S Rig :S 2.
Using measurements of e and high resolution thermocline shear three
proposed parameterisations of turbulence are tested. The results surprisingly
reveal averaged dissipation rates to have no Richardson number dependence
instead scaling with increasing N2 and S2. Measurements of e are poorly
represented by traditional turbulence models based on local stability but conform reasonably well to an empirical scaling suggested by MacKinnon and Gregg (2003). Including this scaling in a dynamical model results in significant
improvements in the simulation of thermocline turbulence at the site.