An observational study of instability and turbulence in nighttime drainage winds

A detailed analysis of two nighttime drainage wind events that commenced on the evenings of 7 and 8 October, 1980 is presented. Data on wind velocity and temperature (10-s values), obtained from each of the eight instrumented levels of the Boulder Atmospheric Observatory, are used to construct 10-min means and root-mean-square values of all the variables. Additional information is provided by acoustic sounder data for 8 October.The analyses reveal that the passage of the drainage front occurs abruptly, between two 10-s observations, on both days. Relatively intense root-mean-square variability in the velocity and temperature fields accompanies the passage of the drainage front. In addition, the undercutting cold drainage air initiates significant variability in the vertical velocity field that extends above the 300 m level of the tower. The most significant variability in the other meteorological fields is primarily restricted to the lowest 150 m with the passage of the drainage front.A principal feature of the analysis is the delineation of Kelvin-Helmholtz instability and billow development, breakdown into turbulence and ultimate decay to a less turbulent state that occurs intermittently behind the drainage front. These features are interpreted in light of Thorpe's (1973a, b) experimental work on stability and turbulence in stratified shear flow and the predictions of linear instability theory. The interpretations are carried out by considering the distributions of the Richardson number, the peak shears of the mean flow and the vertical fluxes of horizontal momentum associated with the unstable growth of the disturbances. Additional comparisons are made between the turbulent structures in Thorpe's laboratory experiments and the turbulence exhibited in traces of the 10-s vertical velocity data measured at various levels both above and below the interface.The relevance of the present results to the design of future field programs, and to the observational data requirements that should be met to incorporate turbulent entrainment processes in models of pollutant dispersal is discussed.