An Investigation Of Vertical Mixing Processes In Shelf Seas

Vertical mixing in shelf seas plays a crucial role in determining the dispersion of material, thetransfer of momentum and the variation of ocean circulation. An abundance of thermohalinestructures and dynamical processes in shelf seas provides a good laboratory for scientific studies onvertical mixing. In this dissertation, we focus on vertical mixing processes over a wide range oftime and spacial scales in different layers of water column. Field observations in Bohai Sea, YellowSea and the New England Shelf are analyzed. A one-dimensional model is also used to interpret theobservational data and to illustrate the dynamics.The microstructure data in strong stratified and weak dissipated Yellow Sea are re-processed.Optimal fits of the measured shear spectra to the empirical universal spectra are provided to avoidthe difficulty when microstructure profilers resolve the low wavenumbers contributing to smallvalues of the turbulent kinetic energy (TKE) dissipation rates (). This method is proved to be validto reduce the estimation errors of, especially for measurements made in strong stratified waters.The turbulent mixing in the bottom boundary layer (BBL) is mainly produced by shearinstability. Strong shear in the BBL leads to gradient Richardson number (Ri) less than0.25,corresponding to high values of. Observations in Bohai Bay and the New England Shelf show theinteractions of the BBL mixing with strong and weak stratification, respectively. The conclusionscan be summarized as follows.(1) Under weak stratification, strong shear and mixing can extendupward into water column, but strong stratification confines the strong mixing in a near bed layerthrough inhibiting the occurrence of shear instability.(2) In weak stratified water, depth-dependentphase delays in shear and velocity were observed; however, the phase delay of shear is larger thanthat of velocity. At the height (max) which the maximum flow magnitude was observed, shearshould be approximately equal to zero, hence phase delay of shear can be represented by thevariation ofmax. With increasing height from bed, the decreasing Reynold stress causes the phasedelay of velocity and decreasing amplitude of flow magnitude, which work together to determinethe variations ofmaxand shear phase delay. Weak stratification reduces the magnitude of the Reynold stress away from the bottom, hence increases the phase and amplitude differences ofvelocity between the upper and lower layers, which results in increasing depth-dependent shearphase delay.(3) The combination of near bed strong stratification and shelf slope can significantlyinfluence the BBL mixing. Differential transports of water masses with different salinity along theslope result in unstable stratification and convective mixing during onshore flow, and stablestratification and decreasing during offshore flow. The commonly employed local balance ofturbulent production and dissipation do not hold for our observations of stratified BBL flow. Twopossible explanations for elevated levels of are horizontal TKE advection and negative buoyancyflux resulted from convection mixing or local re-stratification.Turbulent mixing in the midcolumn is also discussed through analyzing the data obtained onthe New England Shelf. has no Ri dependence, but scaled positively with squared buoyancyfrequency (N~2) and squared shear (S~2). This is in contrast to many statistically based turbulencemodels that predict to increase with diminishing stability. The midcolumn can be explained bywave-wave interactions. is equated to the rate of energy transfer from large to small scale motionsand N~2represents the energy density of the internal waves. The spectral peaks of baroclinic velocityand shear occurred near inertial frequency, suggesting that shear was originated from near inertialmotions. The MG model is used to reproduce the observed. The model results show the correctfunctional dependence of on N~2and S~2, but there are two problems. First, significant variation ofmodel parameter0makes the MG model of limited applicability. Values of N~2and S~2in six datasetson the New England Shelf differ significantly, but their averages of midcolumn are similar. Thispossibly suggests that the effects of stratification are twofold; it provides the energy for internalwaves and also inhibits overturning of internal waves. Similar ranges of midcolumn suggest that0can be scaled negatively with increasing N~2and S~2. Second, values obtained from the MG modelsometimes deviate from observed values. Three methods are suggested to improve the scalingaccuracy:1) the MG model should be applied only to water column which are dominated by shearstability;2) the MG model should be used for water masses which have similar buoyancy structure;3) the MG model can be modified through adding an exponential parameter over combined terms of N~2and S~2.Simultaneous field measurements of the diffusive boundary layer (DBL) and the BBL undertwo distinctly different coastal ocean conditions are made. Of the two experiment sites, one is in anintertidal mudflat in Huichang Bay; one is near6m isobath of Bohai Bay. Observational data showthat the thickness of DBL (DBL) is less than1mm and the oxygen concentrations linearly decreasein the DBL until reaching the sediment-water interface (SWI). The diffusive flux across the SWI ismainly related toDBL. Variations ofDBLand diffusive flux are strongly influenced by those of flowspeed and in the BBL. Increasing BBL dynamics correspond to decreasingDBLand increasingdiffusive flux. According to dimensional analysis,DBL=aDU-1+b (where a and b are parameters, Dis molecular coefficient) is first proposed to scale observedDBL. This estimation has a highcorrelation with observedDBL; however, the parameters significantly vary at both two sites. Laterthree estimates of the Batchelor length obtained from measured, friction velocity frommeasurement (u*) and inferred from velocity measurement (u#) are proposed to scaleDBL. Due topossible larger uncertainties in the measurements of and u*, the estimation of the Batchelor lengthfrom u#has a higher correlation with observedDBLat both two sites. Compared with the formerestimation, the Batchelor length from u#considers more quantities, including the bottom roughnessand the kinematic viscosity. The Batchelor length from u#is a simple and universal scaling ofDBL.In conclusion, the vertical mixing in the BBL and the midcolumn are controlled by turbulence.Turbulence in the BBL is mainly produced by shear instability, and turbulence is the stratifiedmidcolumn can be explained by wave-wave interactions. The vertical mixing in the DBL iscontrolled by molecular diffusion, which influenced by the BBL dynamics.DBLcan be scaled bythe Batchelor length obtained from velocity.