An alternative explanation for the systematic height variation of normalized vertical velocity variance in the near-neutral surface layer

A systematic variation of the normalized vertical velocity, (s_w)/u_* with height above the ground was observed in the near neutral atmospheric surface layer by Högström (1990). He found that the data collapsed remarkably well when the height was normalized with a length scale proportional to the friction velocity u_*. He interpreted this result as an effect of boundary-layer-scale, 'inactive' turbulence, implying that the scale length should be equivalent to the height of the neutral boundary layer. Analysis of data from several field experiments since have given results in remarkable agreement with the data in Högström (1990). It appears, however, that the interpretation of the mechanism responsible for the variation with height of (s_w)/u_* is likely to be different from that originally proposed by Högström. Thus, there are other possible length scales than the height of the entire near-neutral boundary layer which are also proportional to u_*.

In a recent paper, Hunt and Morrison (2000) argue that the very high Reynolds number neutral boundary layer is to a large extent controlled by 'top-down' processes as opposed to what is known to be the case for moderately high Reynolds number boundary layers. This means that high velocity downdrafts or 'sweeps' play a significant role for the turbulence dynamics of the atmospheric surface layer. As the sweeps impinge on the surface, the vertical velocity component becomes blocked by the surface, and the corresponding vertical velocity variance caused by the sweeps becomes: (s_wo)²(z/L₀)^2/3. Here (s_wo)² is the vertical velocity variance in the layer just above the surface layer and L₀ the corresponding length scale. But the local shear in the surface layer also produces vertical velocity fluctuations, which means that the expression for the total normalized velocity variance becomes: (s_w)²=u_*² + (s_wo)²(z/L₀)^2/3.

A previous experimental study by Högström and Bergström (1996) showed that during near-neutral conditions the mean duration of sweeps in the surface layer varies systematically with height but is independent of wind speed or, equivalently, friction velocity. But from this constant time scale Dt it is possible to construct a length scale, which is Au_*Dt, where A=constant. It is argued that L₀=Au_*Dt. It is shown that the experimental data support this model very well, with A and Dt taken from the paper by Högström and Bergström (1996).