The Sensitivity of Upper Ocean Structure to Time Varying Wind Direction

Observations and models show that es lasting one-half inertial period. Thus, the sudden chanMes in the magnitude of the wind stress intent is to isolate one effect rather than to which occur within a time interval of one-half insimulate a particular data set which would be difertial period are most effective in increasing ficult to interpret because of competing processsurface current speeds and mixing the upper layers es. of the ocean. The purpose of the present study is to quantify the effects of concurrent time depenModel dent wind direction. The Mellor-Yamada level 2 1/2 turbulence closure model is used. A series of The Mellor and Yamada (1982) level 2 1/2 turbumodel runs was executed in order to determine the lence closure model was chosen for the present relative sensitivity of mixed layer depth and sea study. This model and its simpler counterpart, surface temperature to wind speed as compared with level 2, have been quite successful in simulating the rate of change of wind direction. The results observations of upper ocean structure (e.g., indicate that the accuracy and time resolution of Mellor and Durbin, 1975). The model includes wind direction should be given special consideraequations for mean zonal and meridional momentum, tion in the design and interpretation of field exheat conservation, turbulent kinetic energy, turperiments which will be used for testing prognosbulent lenMth scale, and an equation of state. tic mixed layer models. Model assumptions include: hydrostatic balance, Boussinesq approximation, low Rossby number, bounIntroduction dary layer approximation, and no vertical advection. The expressions for turbulent fluxes may be Observations and numerical experiments of the written in a classical K-theory form. The vertiL•pper ocean show that sudden changes in the magnical eddy diffusivity of momentum, KM, for example, tude of the wind stress (within one-half inertial may be written as a product of a turbulent velocperiod) correlate with increased surface current ity, a turbulent length scale, and a stability speeds and rapid deepening of the mixed layer function which approaches zero as the flux (e.g., Pollard, 1970; and Mellor and Durbin, Richardson umber (the ratio of negative buoyancy 1975). Klein (1980) found that wind stress pulses production to shear production) exceeds a value of applied at the inertial period result in enhanced approximately 0.2. The details of the model have mixed layer depths because of a resonance mechabeen described extensively elsewhere (e.g., nism. However, Klein's simulations were restrictBlumberg and Mellor, 1981). The model runs utilize ed to uni-directional wind stress conditions. The 101 vertical grid points for the upper 60 m of the onset of major wind events commonly is characterwater column and the time step is 1 h. The duraized by changing wind direction (e.g., Hertzman et tion of each run is 48 h. The model is driven by al., 1974). While several simulations of the mixa wind stress obtained from the quadratic drag ed layer have been quite successful, the relative law. Its magnitude is given by i•portance of wind stress magnitude compared with 2 the rate of change of wind stress direction has Zo=0 a C10U10 (1) not been examined in detail to our knowledge. This question is relevant to the experimental dewhere 0 a is the density of air, C10 is the drag sign (e.g., required accuracy and time resolution coefficient, and U10 is the mean wind speed at 10 of wind stress direction) of upper ocean experim above the ocean surface. The value of C10 is ments and the interpretation of resulting data. set equal to 1.4 X 10-3 for all runs (see Paulson For example, satellite-borne scatterometers and Simpson, 1981, and Dickey and Simpson, 1982, (sidelookin• radar with large incidence angles) for further discussion of C10 variability). The have been used to determine vector wind stress surface heat flux is set equal to zero for all fields over the world's oceans (e.g., Huang, 1979•; runs because only high wind speed cases are conO'Brien, 1981; and Satellite Surface Stress (S) sidered and attention is focused on the short term Working Group, 1982). These data are used as inresponse. The effects of solar heating under conput for models of upper ocean dynamics and ocean ditions of high wind speed (U10>20 ms -1) are gencirculation. The tremendous potential of this erally negligible (Simpson and Dickey, 1981a,b). technique is compelling. The focus of this study Clearly, surface heat fluxes could not be excluded is to quantify the effects of variability in wind if the long term relaxation were of interest. The direction which accompanies mpulsive wind stresssame initial linear stratification (Brunt-V•is•l• period equals 15.3 min) used by Simpson and Copyright 1983 by the American Geophysical Union. Dickey (1981a) and Dickey and Simpson (1982) is employed and the mean currents are initially zero. Paper number 2L1910. For simplicity, geostrophic velocities are ne0094-8276/83/002L-191053.00 glected and uniform salinity is assumed. The lat-