High resolution direct numerical simulations are used to investigate the dynamics of turbulence in flows subject to strong stable stratification, which are common in natural settings. Results are presented for two categories of simulations, uniform and non-uniform density stratification. For all simulated flows, the density stratification was held constant in time, and there was no ambient shear. Flows with uniform density stratification are first analyzed to help provide clear insight to physical processes, followed by flows with non-uniform density stratification which better represent the stratification occurring in nature.

A research program was carried out to investigate turbulent mixing in stably stratified shear flows with the hope of gaining an improved understanding of stably stratified nocturnal boundary layers. The program was mainly laboratory experimental, supplemented by theoretical and numerical developments. The flow configuration consisted of a three-layer system, with upper turbulent layer driven over the lower stratified, quiescent, layer while an intermediate (inversion) layer sandwiched between these two layers. The studies included the nature of instabilities, intermittent generation of turbulence, sustenance and decay of turbulence under varying background conditions (essentially determined by the Richardson number) and ensuing turbulent mixing in the inversion layer. An unprecedented volume of laboratory data were gathered during the program, which enabled to delve into the mechanics and energetics of mixing in stable boundary layers. The laboratory results were compared with, and was used to gain insights on, field observations. Also, the parameterizations developed were compared with those currently used in numerical models. A meso-scale numerical model also was used to check the efficacy of some of the laboratory-based parameterizations.

Turbulence, mixing and the mutual interaction of turbulence and chemistry continue to remain perplexing and impregnable in the fron tiers of fluid mechanics. The past ten years have brought enormous advances in computers and computational techniques on the one hand and in measurements and data processing on the other. The impact of such capabilities has led to a revolution both in the understanding of the structure of turbulence as well as in the predictive methods for application in technology. The early ideas on turbulence being an array of complicated phenomena and having some form of reasonably strong coherent struc ture have become well substantiated in recent experimental work. We are still at the very beginning of understanding all of the aspects of such coherence and of the possibilities of incorporating such structure into the analytical models for even those cases where the thin shear layer approximation may be valid. Nevertheless a distinguished body of "eddy chasers" has come into existence. The structure of mixing layers which has been studied for some years in terms of correlations and spectral analysis is also getting better understood. Both probability concepts such as intermittency and conditional sampling as well as the concept of large scale structure and the associated strain seem to indicate possibilities of distinguishing and synthesizing 'engulfment' and molecular mixing.