Observations of turbulence in a partially stratified estuary

The authors present a field study of estuarine turbulence in which profiles of Reynolds stresses were directly measured using an ADCP throughout a 25- h tidal day. The dataset that is discussed quantifies turbulent mixing for a water column in northern San Francisco Bay that experiences a sequence of states that includes a weak ebb and flood that are stratified, followed by a strong, and eventually unstratified, ebb and flood. These measurements s how that energetic turbulence is confined to a bottom mixed layer by the ov erlying stratification. Examination of individual Reynolds stress profiles along with profiles of Richardson number and turbulent Froude number shows that the water column can be divided into regions based on the relative imp ortance of buoyancy effects. Using the measured turbulence production rate P, the dissipation rate epsil on is estimated. The observed turbulence had values of epsilon/vN(2) > 20 a ll of the time and epsilon/vN(2) > 200 most of the time, suggesting that th e observed motions were buoyancy affected turbulence rather than internal w aves. However. at times, turbulent Froude numbers in much of the upper-wate r column were less than one, indicating important stratification effects. T aken as a whole, the data show that stratification affects the turbulent ve locity variance q(2) most severely; that is, observed reductions in <(u'w') over bar> are largely associated with small values of q(2) rather than with a dramatic reduction in the efficiency with which turbulent motions produc e momentum fluxes. Finally, the dataset is compared to predictions made using the popular Mell or-Yamada level 2.5 closure. These comparisons show that the model tends to underestimate the turbulent kinetic energy in regions of strong stratifica tion where the turbulence is strongly inhomogeneous and to overestimate the turbulent kinetic energy in weakly stratified regions. The length scale do es not appear to compensate for these errors, and. as a result, similar err ors are seen in the eddy viscosity predictions. It is hypothesized that the underestimation of q(2) is due to an inaccurate parameterization of turbul ence self-transport from the near-bed region to the overlying stratificatio n.