Dynamic and reduced-dynamic precise orbit determination of satellites in low earth orbits

The precise positioning of satellites in Low Earth Orbits (LEO) has become a key technology for advanced space missions. Dedicated satellite missions, such as CHAMP, GRACE and GOCE, that aim to map the Earths gravity field and its variation over time with unprecedented accuracy, initiated the demand for highly precise orbit solutions of LEO satellites. Furthermore, a wide range of additional science opportunities opens up with the capability to generate accurate LEO orbits. For all considered satellite missions, the primary measurement system for navigation is a spaceborne GPS receiver. The goal of this thesis is to establish and implement methods for Precise Orbit Determination (POD) of LEO satellites using GPS. Striving for highest precision using yet efficient orbit generation strategies, the attained orbit solutions are aimed to be competitive with the most advanced solutions of other institutions. Dynamic and reduced-dynamic orbit models provide the basic concepts of this work. These orbit models are subsequently adjusted to the highly accurate GPS measurements. The GPS measurements are introduced at the zero difference level in the ionosphere free linear combination. Appropriate procedures for GPS data screening and editing are established to detect erroneous data and to employ measurements of good quality only. For the dynamic orbit model a sophisticated force model, especially designed for LEO satellites, has been developed. In order to overcome the limitations that are induced by the deficiencies of the purely dynamical model, two different types of empirical parameters are introduced into the force model. These reduced-dynamic orbit models allow for the generation of much longer orbital arcs while preserving the spacecraft dynamics to the most possible extent. The two methods for reduced-dynamic orbit modeling are instantaneous velocity changes (pulses) or piecewise constant accelerations. For both techniques highly efficient modeling algorithms are presented. The additional parameters are estimated within the adjustment process. In order to regulate their impact on the dynamic solution, the empirical parameters are assigned statistical a priori information. This allows for a perfect synergy of the advanced knowledge of spacecraft dynamics and the high accuracy of the GPS measurements. The developed routines have been implemented in a computer program and tested for CHAMP and GRACE. The applied strategies proved to be highly efficient, robust and flexible. The attained orbit solutions are validated against solutions of other well-established POD methods. It is shown, that dynamic POD delivers accuracies at the 1 dm level for orbital lengths of up to 90 minutes. Furthermore, employing an adequate number of pseudo-stochastic parameters with optimal weighting, reduced-dynamic orbit determination is a powerful strategy for LEO POD. Orbital arcs over a whole day can be generated with an accuracy of up to 4.5 cm RMS. (author)