Abstract
Accurate assessment of marine currents is critical for meaningful planning of tidal stream energy deployments. Methods for assessing mean flow speed are well-established, but the issue of assessing phenomena that vary over much shorter time scales than the semidiurnal tide (principally turbulence and waves) is not as settled. This is in part due to the limitations of the standard instruments used for surveying currents at potential tidal stream sites i.e., acoustic Doppler current profilers (ADCPs). Because ADCPs use multiple acoustic beams to sample single components of velocity at widely-separated spatial locations, a reliable analysis of the mean flow can be obtained by a suitable average over measurements from all beams: this approach works well for determining mean flow properties, but is less well-suited for the smaller and faster variations associated with waves and turbulence. In particular, conventional analysis approaches such as the variance method do not meaningfully distinguish between wave-driven and turbulence-driven velocity variations.
An important consequence of this is that ADCP estimates of turbulent kinetic energy (TKE) are biased high because the TKE calculation method includes wave-driven velocity fluctuations as well as those caused by turbulence. Previous work by the authors has successfully applied a mixed spectral-statistical filter to ADCP estimates of TKE from a tidal stream test site; this filter showed good performance in distinguishing between true TKE and pseudo-TKE due to wave action. However, for field data it is only possible to assess the filter’s performance with respect to wave properties: this is because independent measurements of wave properties can be obtained from a surface buoy, but there is no alternative instrumentation that can yield independent measurements of turbulence.
The study presented here addresses this shortcoming by generating an artificial flowfield with known turbulence and wave properties, sampling with a virtual ADCP, and applying the spectral-statistical filter to the results. In this way it is shown that the combined filter improves the separation of waves and turbulence over the individual filters by an even greater degree when assessed against turbulence data rather than against independent wave measurements. The study discusses the effects of different artificial turbulence generation schemes, specifically contrasting the spectral or Sandia method with the synthetic eddy method; it also discusses the tuning of spectral filter parameters for optimum separation of wave and turbulence effects.