The upper atmosphere dynamics in the polar cap is mainly driven by ion-drag momentum sources imposed by the mapping of magnetosphere convection into the thermosphere/ionosphere and by Joule and auroral particle heating. Auroral particles also enhance conductivity particularly in the middle and lower ionosphere. Changes in the magnetospheric energy and momentum sources can significantly modify the wind circulation during geomagnetic storms. To observe these effects, a Michelson interferometer has been installed in Svalbard to measure winds in the thermosphere. Prior to 30 October 2003, cloud cover over Svalbard rendered the conditions unfavorable for optical observation. However, meteorological conditions improved after this date to enable the thermospheric response to the 28 October coronal mass ejection to be made. During quiet geomagnetic conditions measured wind velocities were in good agreement with those predicted by the Horizontal Wind Model (HWM). During disturbed geomagnetic conditions, HWM tended to underestimate the observed velocities. Comparison of the wind observations with a physical model tended to show reasonable agreement during both the strongly driven and recovery phase of the storm. Although the physical model did not always capture the timing of the rapid changes in the wind response in the early phase of the storm, the amplitudes of the fluctuations were in good agreement. After the initial phase the physical model agreed well with both the timing and amplitude of the meridional and zonal wind fluctuations. The meridional wind component was also derived from the EISCAT Svalbard Radar ion velocity and was found to be in close agreement with the optical winds observations.