DERIVATION AND OBSERVABILITY OF UPPER ATMOSPHERIC DENSITY VARIATIONS UTILIZING PRECISION ORBIT EPHEMERIDES

Several models of atmospheric density exist in today's world, yet most possess significant errors when compared to data determined from actual satellite measurements. This research utilizes precision orbit ephemerides (POE) in an optimal orbit determination scheme to generate corrections to existing density models to better characterize observations of satellites in low earth orbit (LEO). These corrections are compared against accelerometer derived densities that are available for a few select satellites, notably, the CHAMP and GRACE satellites. These corrections are analyzed by determining the cross correlation coefficients and root-mean-squared values of the estimated corrected densities as compared to the accelerometer derived densities for these satellites. The POE derived densities showed marked improvement using these methods of comparison over the existing empirical density models for all examined time periods and solar and geomagnetic activity levels. The cross correlation values for the POE derived densities also consistently out-performed the High Accuracy Satellite Drag Model (HASDM). This research examines the ability of POE derived densities to characterize short term variations in atmospheric density that occur on short time scales. The specific phenomena examined were travelling atmospheric disturbances (TAD) and geomagnetic cusps, which had temporal spans of less than half the period of the satellite's orbit, more specifically spans of between four and ten minutes, and less than three minutes respectively. Density variations of shorter duration are more difficult to observe even in accelerometer data due to diurnal variations that arise from cyclical increases due to the satellite passing from the darkened side of the earth to the lit side. This research also examines the effects of a vertically propagating atmospheric densities by looking at periods of time during which both the GRACE and CHAMP satellites have coplanar orbits, during which perturbations can be examined for their capability to extend vertically through the atmosphere, as well as their observability in POE derived densities. Additionally, this research extends the application of optimal orbit determination techniques to an additional satellite, the TerraSAR-X, which lacks an accelerometer. For LEO, one of the greatest uncertainties in orbit determination is drag, which is largely influenced by atmospheric density. There are many factors which affect the variability of atmospheric densities, and some of these factors are well modeled, such as atmospheric heating and to some degree, the solar and geomagnetic activity levels, though some variations are not modeled at all. The orbit determination scheme parameters found to perform best for most cases were a baseline model of one of the three Jacchia based baseline models, a density correlation half-life of 18 or 180 minutes, and a ballistic coefficient correlation half life of 1.8 minutes. All three Jacchia based models performed very similarly, with the CIRA-1972 model edging out the other two overall. The density correlation half-life's optimal value was usually 180 minutes, though for specific levels of geomagnetic activity, a half-life of 18 minutes was preferable. During the coplanar periods for both the GRACE and CHAMP satellites, both satellites showed minor density increases that occur on the unlit side of the earth near the equator. These increases were mostly unseen in the precision orbit ephemeris (POE) derived densities, though the POE derived densities did show a slight response to these perturbations. The secondary density increases were seen in both GRACE and CHAMP accelerometer data, and likely existed both above and below the orbits of these two satellites. The TerraSAR-X densities found for the time period examined in this study using POE data showed deviations from empirical density models of up to 10% for peak atmospheric density values. The CHAMP and GRACE POE derived densities showed a greater relative deviation from the empirical density models during peak density periods, and the deviations for the CHAMP and GRACE satellites' empirically predicted densities much better approximated the density values found using the accelerometers aboard both satellites. As the TerraSAR-X satellite lacks its own accelerometer, the POE derived densities are assumed to be a more accurate representation of the atmospheric densities