Observations of the cross-wind velocity variance in the stable boundary layer

Nine tower datasets over grassland, brush rangeland, snow covered plain, the ocean, three different pine forests, an aspen forest and an urban site, are used to document the scale-dependence of the cross-wind velocity variance in the stable boundary layer. The turbulence velocity variance scales with the surface momentum flux, as reported in previous studies. Such scaling removes the stability dependence of the variance at a given site, and also removes most of the differences between sites. The scaling is more effective with use of a record-dependent averaging time for defining the turbulent fluctuations. The variable averaging time is the timescale associated with the gap region in the heat flux multiresolution cospectra. On scales larger than turbulence and less than a few hours (mesoscale), variations in the cross-wind velocity variance at a given site are not related to local variables such as the friction velocity. Possible exceptions include suppression of turbulence and mesoscale motions in well-defined drainage flows and enhancement of turbulence and mesoscale motions in stronger winds downstream of a ridge. Larger mesoscale variance is associated with complex terrain and forested sites compared to the more homogeneous sites in flat terrain with short or no vegetation. These differences between sites are related to the absence of a gap region in the velocity spectra at the complex terrain and forested sites. The observed probability distribution functions of the total variance and the mesoscale variance are documented for different averaging times, stability classes and site characteristics.