Precise orbit determination for GRACE using undifferenced or doubly differenced GPS data

Abstract The two GRACE satellites provide the ideal platform to study the performance of different strategies for precise orbit determination using undifferenced or doubly differenced GPS data. We use pseudo-stochastic orbit modeling techniques in a batch least-squares environment for the two GRACE satellites to outline the mutual benefits of processing doubly differenced instead of undifferenced GPS data. We either process the space baseline only, the space-ground baselines only, or both types of baselines together, and show that the fixing of the GPS double difference carrier phase ambiguities has a significant impact on the space baseline, but also on the space-ground baselines. The validation of the relative orbit positions by inter-satellite K-band observations shows precisions of better than 1 mm in the case of fixed space baseline ambiguities, precisions of a few millimeter in the case of fixed space-ground baseline ambiguities, and precisions of about 1 cm in the case of float ambiguities. We discuss the differences between the various GRACE orbit solutions in order to formulate well suited orbit determination strategies tailored to the GRACE configuration. Satellite laser ranging observations indicate that accuracies between 2 cm and 2.5 cm are achieved.