When the upper atmosphere is eclipsed by the sun, most of the radiation causing ionisation and heating is blocked by the moon. As a result, the atmosphere in the shadow region cools and as it does so, it contracts downwards. This contraction draws in air from around the eclipsed region and as a result, the horizontal winds are expected to converge on this region.
Recent modelling work carried out by Ingo Mueller-Wodarg at the Atmospheric Physics Laboratory at UCL suggests that these temperature and wind changes will be measurable.
The following images link to animations of the zonal and meridional winds respectively.
Note that the southward and eastward winds converge as the eclipse starts at around 10.30. This is because the cooler air under the eclipse shadow contracts, drawing air in from around it.
These results were obtained by comparing a model run for August 11 1999 with no eclipse with another model run which included a simulation of the eclipse. The plots above are the difference between these two model runs. From these it is possible to predict what will be seen by an instrument at a given location. For example, winds at a 52 degrees latitude, zero longitude and an altitude of 240 km.
An unexpected prediction is that a significant disturbance in the wind field then propagates southward, reaching the south pole by 14.00. Although these differences in the wind field are very small by this point, a measurable effect can be seen in the southward wind as far as the equator.