Time Scales in the Unstable Atmospheric Surface Layer

Abstract

Calculation of eddy covariances in the atmospheric surface layer (ASL) requires separating the instantaneous signal into mean and fluctuating components. Since the ASL is not statistically stationary, an inherent ambiguity exists in defining the mean quantities. The present study compares four methods of calculating physically relevant time scales in the unstable ASL that may be used to remove the unsteady mean components of instantaneous time signals, in order to yield local turbulent fluxes that appear to be statistically stationary. The four mean-removal time scales are: (t _c ) based on the location of the maximum in the ogive of the heat flux cospectra, ($\tilde t_{MR}$) the location of the zero crossing in the multiresolution decomposition of the heat flux, (t ^*) the ratio of the mixed-layer depth over the convective velocity, and ($\tilde t\,$) the convergence time of the vertical velocity and temperature variances. The four time scales are evaluated using high quality, three-dimensional sonic anemometry data acquired at the Surface Layer Turbulence and Environmental Science Test (SLTEST) facility located on the salt flats of Utah’s western desert. Results indicate that $t_c\approx t_{MR}$ and $t^*\approx \tilde t$ , with t _c achieving values about 2–3 times greater than t ^*. The sensitivity of the eddy covariances to the mean-removal time scale (given a fixed 4-h averaging period during midday) is also demonstrated.