New simulations of semidiurnal thermospheric winds and perturbation temperatures calculated by the National Center for Atmospheric Research thermospheric general circulation model (TGCM) are presented for equinox conditions. The model now incorporates a self-consistent calculation of neutral composition and includes high-latitude energy and momentum sources associated with the aurora. The tidal waves propagating upward from the lower atmosphere are specified by wave like distributions of the geopotential at the model lower boundary; the boundary geopotential is related to the other model fields through Hough mode tidal functions. The specification of the waveamplitudes and phases is based on lower atmosphere model predictions and thus represents the coupling between the upper and lower atmosphere in a more physically meaningful manner than previous calculations. In these simulations, asymmetric semidiurnal modes are also included. The revised TGCM calculations at geographic longitude 75¿W are compared with observational data at low, middle, and high latitudes. In general, the agreement is reasonable, given the limited data available. The predicted solar cycle variation of the semidiurnal tides is consistent with previous works: the temperatures respond more strongly than the winds, and the solar cycle effects are most pronounced in the upper thermosphere.

The lower thermosphere is dominated by the upward propagating waves from the lower atmosphere, with little contribution from solar forcing. These upward propagating waves are damped at lower altitudes as solar forcing increases. The effects of coupled composition and auroral forcings on the semidiurnal tidal components are investigated. Both processes alter the simulated solar-driven tidal fields compared to calculations which neglected these processes: amplitudes may double or triple in restricted altitude and latitude regions, while the times of maxima may shift by a few hours. The TGCM simulations are compared with the Fourier-decomposed exospheric temperature measured during the September 1984 Equinox Transition Study period. The semidiurnal temperature phase in particular is found to be strongly dependent on the level of geomagnetic activity. Thus the simulation of individual day observations will require careful specification of the auroral forcings which are input into the theoretical models. Simultaneous data obtained at different latitudes are vital to test and confirm the model predictions. ¿1991 American Geophysical Union