A covariance analysis is presented for satellite tracking and gravity recovery with a differential Global Positioning System-based technique to be demonstrated on TOPEX in the early 1990s. The technique employs data from an ensemble of repeat ground tracks to recover a unique satellite epoch state for each track and a set of invariant positional parameters common to all tracks. The positional parameters represent the effect of mismodeled gravitational field on the satellite orbit. At an altitude of 1336 km, where gravity modeling is the dominant systematic error, averaging of random error over many arcs and adjustment of the gravity model reduce the final satellite position error. The positional parameters can then be used to produce a refined global gravity model. The analysis indicates that errors ranging from 5 to 8 cm in TOPEX altitude and 0.05 to 0.2 mGal for the gravity field can be achieved, depending on the number of repeat arcs used.
INTRODUCTIONNASA's Ocean Topography Experiment (TOPEX) will carry an orbiting radar altimeter to measure sea surface height, with the primary goal of charting global ocean circulation [Born et al., , 1985. TOPEX will fly in a 1336-km circular orbit inclined at 66.1 ø. The altimeter measurements, averaged over a several kilometer footprint, will be precise to about 2.5 cm. In order to make maximum use of the precise altimetry it is necessary to independently track the satellite geocentric altitude with comparably high precision. The baseline tracking system for TOPEX is a global network of laser-ranging systems which is expected to deliver about 13-cm altitude accuracy for TOPEX [Born et al., , 1985. In addition, an experimental tracking system employing the Global Positioning System (GPS) will be used in a special demonstration. The experimental system will make use of an on-board GPS receiver operating in concert with a network of six ground receivers distributed around the globe [Lichten et al., 1985]. A number of estimation techniques have been proposed to determine the TOPEX orbit using the GPS data [Yunck et al., 1990; Melbourne et al., 1986]. The most thoroughly studied techniques are variations on the traditional dynamic technique which requires the use of accurate force models. Systematic errors in the force models lead directly to systematic errors in the orbit solution. At the 1336-km TOPEX altitude, which is well above the regime of significant atmospheric drag, errors in the model for the Earth's gravity field are the major source of dynamic orbit error [Lichten et al., 1985; Wu et al., 1986].Because it provides continuous coverage with accurate measurements, the GPS system allows alternative tracking techniques which are less dependent on the nominal dynamic models. Two recently proposed alternatives to dynamic orbit estimation, the kinematic and reduced dynamic techniques Wu et al., 1991], make use of geometric information in GPS observations to greatly reduce the sensitivity to force model errors. A third ap-