Precise satellite orbit determination with particular application to ERS-1

The motivation behind this study is twofold. First to assess the accuracy of ERS-1 long arc ephemerides using state of the art models. Second, to develop improved methods for determining precise ERS-1 orbits using either short or long arc techniques. The SATAN programs, for the computation of satellite orbits using laser data were used. Several facilities were added to the original programs: the processing of PRARE range and altimeter data, and a number of algorithms that allow more flexible solutions by adjusting a number of additional parameters. The first part of this study, before the launch of ERS-1, was done with SEAS AT data. The accuracy of SEASAT orbits computed with PRARE simulated data has been determined. The effect of temporal distribution of tracking data along the arc and the extent to which altimetry can replace range data have been investigated. The second part starts with the computation of ERS-1 long arc solutions using laser data. Some aspects of modelling the two main forces affecting ERS-l's orbit are investigated. With regard to the gravitational forces, the adjustment of a set of geopotential coefficients has been considered. With respect to atmospheric drag, extensive research has been carried out on determining the influence on orbit accuracy of the measurements of solar fluxes (P10.7 indices) and geomagnetic activity (Kp indices) used by the atmospheric model in the computation of atmospheric density at satellite height. Two new short arc methods have been developed: the Constrained and the Bayesian method. Both methods are dynamic and consist of solving for the 6 osculating elements. Using different techniques, both methods overcome the problem of normal matrix ill- conditioning by constraining the solution. The accuracy and applicability of these methods are discussed and compared with the traditional non-dynamic TAR method.