Better observations are required in order to improve the representation of the physical processes determining the sea surface temperature and the depth of the ocean surface boundary layer in numerical models of the interaction between the ocean and the atmosphere.
Measurements of the turbulent kinetic energy dissipation rate, ε, are commonly used to derive mixing rates, with velocity microstructure profiling using a free-fall instrument being the stand- ard observation technique. The use of acoustic Doppler current profile (ADCP) instruments offers the possibility of collecting ε data using relatively standard instruments, suitable for long-term deployment and operating in conditions when profiling would be impossible. However, the ve- locity structure function method used with ADCP data is still relatively novel in marine applica- tions and is evolving.
Here we present observations from a site in the central Celtic Sea covering a period of sixteen months. The observations include water column temperature, salinity and velocity structure, to- gether with meteorological and wave observations from surface buoys. An additional mooring included three ADCP configured to collect data for turbulence calculations at nominal depths of 20 m, 35 m and 50 m in an overall water depth of 145 m. The site is subject to temperature driven seasonal stratification. Temperature gradients and velocity shear are concentrated at the upper and lower boundaries of the thermocline, which itself is subject to substantial vertical displace- ment at semi-diurnal and higher frequencies. Meteorological and near-surface observations are used to derive wind stress and surface buoyancy flux. Wave observations are used to calculate Stokes drift profiles using a range of methods.
The turbulence ADCP along-beam velocities include a contribution due to surface wave forced orbital motion even under low wave energy conditions. Analysis based on linear wave theory demonstrates that such motion will unavoidably contribute to the second-order structure func- tion. Synthesised velocities for relevant observation depths and a range of surface wave amp- litudes and wavelengths demonstrate that the surface wave forced bias is significant compared with the anticipated ε levels.
A modified method is described that exploits the different length scale dependency of the surface wave contribution and that associated with genuine turbulence. The modified approach proposes an additional term for the regression in order to separate the components. Using data from the present study under winter, well mixed conditions, the modified methodology is demonstrated to produce results that are consistent with standard wind stress and buoyancy forcing scaling and independent of the maximum range over which the structure function is evaluated, unlike the standard methodology.
Examination of the heading and tilt data from the ADCP indicated that the instruments oscillated about all three axes - heading, pitch and roll - over an observation burst. Whilst not directly contributing to the along-beam velocity measured by the ADCP, analysis demonstrated that such motion in the presence of a sheared current would result in a portion of the velocity difference between bins due to the current shear contributing to the structure function and leading to a residual ε bias. This has not previously been identified as a potential source of bias in structure function ε estimates.
Testing with synthetic velocity data demonstrated that the residual bias fraction was primarily determined by the angular range of the oscillation, with heading oscillation likely to be the dom- inant cause of residual bias. The structure function contribution arising from the ADCP motion in the presence of linear shear has the same length-scale dependency as that due to the surface wave forced orbital motion. Consequently the modified method also corrects for any residual bias due to the motion of the ADCP.