Recent basic research on turbulent shear flow has illustrated the fundamental effect of a non-uniform shear layer in producing a partial barrier to penetration of low-amplitude turbulent fluctuations from above. In Hunt and Morrison (2000) it is demonstrated that the length scale over which eddies de-correlate by a factor of 1/3, Lx=u0/dU/dz, where u0 is the rms turbulent velocity above the shear zone. For a neutral surface layer, the integral scale of the normal component of the vertical velocity Lz can be expressed as the a combination of this 'shear sheltering' or 'shear suppression' scale Lx and a length scale relevant for the blocking effect of the surface, Lb ~ z, where z is the height above the ground. Near the surface, both the effects of blocking and shear sheltering reduce the integral scale of w. Therefore to estimate the effect that makes the integral scale smallest, the harmonic mean is the appropriate interpolation, namely:
Lz-1=1/z + dU/dz/sw. This expression is valid for the neutral surface layer. But if there is stable stratification, a third term must be added to this expression: N/sw, where N=(g/T0dQ/dz) 1/2 is the Brunt-Vaisala frequency. In cases with stable stratification and a wind maximum situated close to the surface (< 100 m), the last two terms dominate and we get:
(LUT)-1=dU/dz//sw + N/sw (Smedman et al., 1997).
Measurements taken at a marine site in the Baltic Sea, the Östergarnsholm station, during stable conditions with frequent low level wind maxima are used to test the validity of the above predictions. It is found that a more or less pronounced spectral minimum ('spectral gap') is found in most of the measurements from 9 m above the water surface at frequencies around 10-2 Hz. In some cases the minimum is replaced by an inflexion. The ratio of the spectral energy at the maximum and at the minimum (or, as the case may be, inflexion) was found to be a strong function of the wind gradient dU/dz, indicating increasingly strong suppression of turbulent energy around n=10-2 Hz with increasing magnitude of the wind gradient. Thus eddies in the boundary layer above the measuring height, 9 m, could not make it into the surface layer. Also it was found that the drag coefficient, defined as the ratio of the measured stress at 9 m and the wind speed squared, was a strong function of the length scale LUT, the drag coefficient having a very small value for small LUT and increasing rapidly with increasing LUT.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner