Non-dimensional wind and temperature profiles in the atmospheric surface layer: A re-evaluation

Abstract

Previous results of non-dimensional wind and temperature profiles as functions of ζ( = z/L) show systematic deviations between different experiments. These discrepancies are generally believed not to reflect real differences but rather instrumental shortcomings. In particular, it is clear that flow distortion has not been adequately treated in most previous experiments. In the present paper, results are presented from a surface-layer field experiment where great care was taken to remove any effects from this kind of error and also to minimize other measuring errors. Data from about 90 30-min runs with turbulence measurements at three levels (3, 6, and 14 m) and simultaneous profile data have been analysed to yield information on flux-gradient relationships for wind and temperature.

The flux measurements themselves show that the fluxes of momentum and sensible heat are constant within ± 7% on average for the entire 14 m layer in daytime conditions and when the stratification is slightly stable. For more stable conditions, the flux starts to decrease systematically somewhere in the layer 6 to 14 m. From a large body of data for near-neutral conditions (¦ζ¦ ≦ 0.1), values are derived for von Kármán's constant: 0.40 ± 0.01 and for Φ _h at neutrally, 0.95 ± 0.04. The range of uncertainty indicated here is meant to include statistical uncertainty as well as the effect of possible systematic errors.

Data for Φ _m and Φ _h for an extended stability range (1 > ζ > − 3) are presented. Several formulas for Φ _m and Φ _h appearing in the literature have been used in a comparative study. But first all the formulas have been modified in accordance with the following assumptions: κ = 0.40 and (Φ _h )_{ζ = 0} = 0.95; deviations from this result in the various studies are due to incomplete correction for flow distortion. After new corrections are introduced, the various formulas were compared with the present measurements and with each other. It is found that after this modification, the most generally used formulas for Φ _m and Φ _h for unstable conditions, i.e., those of Businger et al. (1971) and Dyer (1974) agree with each other to within ± 10% and with the present data. For stable conditions, the various formulas still disagree to some extent. The conclusion in relation to the present data is not as clear as for the unstable runs, because of increased scatter. It is, however, found that the modified curve of Businger et al. (1971) for Φ _h fits the data well, whereas for Φ _m , Dyer's (1974) curve appears to give slightly better agreement.