The problem of turbulence decay in the atmospheric boundary layer has not been extensively studied because of the observational and computational difficulties associated with the inherent non-stationarity of the process. Two previous studies [Nieuwstadt and Brost (1986) and Sorbjan (1997)], both based on large-eddy simulation, have proposed velocity and time scales for turbulence variables during the decay process associated with the afternoon transition from daytime convective conditions to nocturnal stable stratification. In this paper, we use turbulence measurements from a sonic anemometer, a minisodar, and a wind profiling radar to evaluate the behavior of these scales in a real atmospheric boundary layer. For example, Nieuwstadt and Brost found that the dissipation length scale in their simulations remained essentially constant during the decay process. Our data analysis indicates this is almost true in the real atmosphere, although there is a tendency for this length scale to decrease with time. The data we have used for this analysis were collected for three weeks during October 2000 in the central Salt Lake Valley as part of the first field campaign of the U.S. Department of Energy's Vertical Transport and Mixing (VTMX) program.

We also compare the previous scaling results and results from the observations with direct numerical simulations (DNSs) of turbulence driven by both idealized and observed time series of surface heat flux. The simulations allow us to calculate ensemble averages of turbulence variables and to determine the sensitivity of scales to the particular functional form of the surface heat flux.