Adv. Space Res. Vol. 11, No.6, Pp. (6)189— (6)192, 1991 0273—1177/91 $0.00 + .50 Printed in Great Britain. All rights reserved. Copyright © 1991 COSPAR ACCURACY ASSESSMENT OF GPS SATELLITE ORBITS B. E. Schutz, B. D. Tapley, P. A. M. Abusali and C. S. Ho Center for Space Research, The University of Texas at Austin, Austin, TX 78712, USA. ABSTRACT Global Positioning System (GPS) orbit determination using data collected by the civilian tracking network CIGNET have been used ina variety of experiments, along with data collected from special campaigns to augment the global tracking data. These experiments have been used to investigate the fidelity of the satellite force and kinematic models with the goal of improving the GPS ephemeris accuracy. Software has been developed for the preprocessing of GPS pseudo-range and phase measurements, including procedures for cycle slip correction. A multi-satellite orbit determination program, MSODP, allows for satellite-dependent force models, including high- low satellite constellations. Assessments of orbit accuracies derived from measurements collected in campaigns with nearly global networks have been performed with a variety of techniques, including determination of baselines known from SLR or VLBI, the computation of long, multi-week arcs and the comparison of predicted ephemerides with determined ephemerides. These results have provided indications of inadequacies in the force models, especially with the nongravitational forces, particularly those aspects associated with eclipsing. INTRODUCTION The importance of accurate GPS orbits has been widely acknowledged by an abundance of papers /1, 2/. The required orbit accuracy is dependent on the particular application and is further complicated by the fact that orbit accuracy does not always map into comparable accuracy in terms of geodetic parameters, such as interstation baseline vectors. While the orbit accuracy can be described in terms of, for example, the positional accuracy in an adopted reference frame, the more commonly used measure of accuracy is the ability of an orbit, or ephemeris, to support geodetic parameter estimation. The latter application describes the accuracy in terms of the baseline length, such as 1 part in 108 or 1 cm on a 1000 km baseline. The positional accuracy of an orbit is dependent on a variety of factors, including the quality of the force and kinematic models as well as the precision and geographical distribution of the measurements used to determine the orbit. The quality of the models interacts in a complex way with the measurement quality and distribution over the interval of time selected for the orbit determination. This paper examines the GPS orbit accuracy using several evaluation procedures. The data used for the study is drawn from the CASA Uno experiment /3/. Furthermore, this paper provides additional detail to the results published in /4/. DATA The CASA Uno experiment was conducted from January 18 through February 5, 1988, using a tracking network based on the CIGNET, but augmented with several additional sites in the Pacific. The complete network of stations is given in Table 1, which consisted of 11-4100 receivers. Most of the sites were located near VLBI or SLR sites and the position vectors of the GPS receiver reference point with respect to the VLBIJSLR reference points has been documented (e.g., /4/). For all of the studies described in this paper, ionospherically corrected measurements were used. In Table 1, a few sites could not be satisfactorily specified a priori with respect to SLR and VLBI reference points as noted with the asterisk. These sites were treated as unknown and estimated during the solution process. In addition, a few sites were selected to be estimated for the purpose of aiding in orbit accuracy assessment even though their coordinates are accurately known with respect to the SLRJVLBI systems.