The structure and evolution of a heated, sheared planetary boundary layer (PBL) similar to the one developed on July 11, 1987, during the First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment (FIFE) are investigated by a two- and a three-dimensional large eddy simulation (LES) model. The simulated results are utilized to compare with the single Doppler lidar data whose validity, however, has never been thoroughly verified due to insufficient independent observations. In contrast to the traditional research on a buoyancy-driven, windless convective boundary layer (CBL), the horizontal and vertical domains of this study are designed to be sufficiently large so that the principal physical processes involved in the development of this boundary layer can be successfully simulated. They include (1) large eddies generated by the surface sensible heat flux, (2) internal gravity waves excited in the stable layer due to the lifting mechanism exerted by the rising large eddies, (3) interaction between the waves and the background wind shear, and (4) wave breakdown caused by the convective instability when the flow speed exceeds the phase speed. The mean and statistical properties of this boundary layer turn out to be quite different from those of a CBL. Among them, the most unique feature is that in the stable layer the direction of the momentum flux (u'w'), which is transported by the gravity waves, is countergradient. The occurrence of this phenomenon, derived by lidar and confirmed by the simulation, is associated with the existence of a critical layer where the Doppler-shifted frequency vanishes. Under this critical layer the vertical profile of horizontal velocity variance (u'2) is found to exhibit a second maximum. The simulated results of potential temperature flux (w'&thgr;') as well as the potential temperature variance (&thgr;'2) also imply a second local absolute value maximum in the stable layer. The former is believed to be closely related to the formation of convective instability. Despite the model's ability in simulating the physical processes involved in the FIFE PBL, the large quantitative discrepancy discovered between the model-generated and the lidar-extracted statistical properties lead to a consistency check for the latter. It is shown that in the stable layer the magnitudes of the lidar-observed momentum flux (u'w') and the vertical velocity variance (w'2) appear to be too strong. Several possible error sources which can contaminate the lidar observations are discussed.¿ 1997 American Geophysical Union