Measured E region neutral winds from the Atmospheric Response in Aurora (ARIA 1) rocket campaign are compared with winds predicted by a high-resolution nonhydrostatic dynamical thermosphere model. The ARIA 1 rockets were launched into the postmidnight diffuse aurora during the recovery phase of a substorm. Simulations have shown that electrodynamical coupling between the auroral ionosphere and the thermosphere was expected to be strong during active diffuse auroral conditions (Walterscheid and Lyons, 1989). This is the first time that simulations using the time history of detailed specifications of the magnitude and latitudinal variation of the auroral forcing based on measurements have been compared to simultaneous wind measurements. Model inputs included electron densities derived from ground-based airglow measurements, precipitating electron fluxes measured by the rocket, electron densities measured on the rocket, electric fields derived from magnetometer and satellite ion drift measurements, and large-scale background winds from a thermospheric general circulation model.

Our model predicted a strong jet of eastward winds at E region heights. A comparison between model predicted and observed winds showed modest agreement. Above 135 km the model predicted zonal winds with the correct sense, the correct profile shape, and the correct altitude of the peak wind. However, it overpredicted the magnitude of the eastward winds by more than a factor or 2. For the meridional winds the model predicted the general sense of the winds but was unable to predict the structure or strength of the winds seen in the observations. Uncertainties in the magnitude and latitudinal structure of the electric field and in the magnitude of the background winds are the most likely sources of error contributing to the differences between model and observed winds. Between 110 and 135 km the agreement between the model and observations was poor because of a large unmodeled jetlike feature in the observed winds (140 m s-1). Agreement between the present simulation and the earlier simulations of Walterscheid and Lyons (1989) is favorable, although the winds in the present simulation are considerably weaker for particle precipitation of similar characteristic energy and flux. The reasons for the difference were the smaller latitudinal extent of the model diffuse aurora and the weaker electric fields in our simulation.

We have shown that the enhanced electron densities and electric fields associated with the postmidnight diffuse aurora provide the potential for a rapid acceleration of the zonal winds as shown by Walterscheid and Lyons (1989). However, the modeled response to the large-scale electric field is too great. This suggests that the assimilated mapping of ionospheric electrodynamics (AMIE) electric field is also too large. The actual electric field is most likely reduced locally in regions of enhanced ionization and conductivity within the diffuse aurora. In addition, we have shown that the ''exotic'' jetlike wind feature between 110 and 135 km is not aurorally forced. However, it may be the result of an enhancement of the Hall drag relative to the Coriolis force that modifies the geostrophic balance with the large-scale pressure gradient. ¿ American Geophysical Union 1995