he Global Positioning System (GPS) satellite orbits were estimated using the National Aeronautics and Space Administration's GEODYNII software, a batch least-squares estimator. Three deterministic models were tested: the solar radiation pressure scale factor and y-bias acceleration; the solar radiation pressure scale factor and radial, GPS X and GPS Y axes accelerations; and Colombo's resonant harmonic model which supplements the radial, GPS X and GPS Y axes accelerations with sine and cosine terms with frequencies equal to the nominal GPS orbital frequency. For one-day GPS orbits all models had an RMS of fit to the Final International GPS Geodynamics Service (IGS) orbit of 26 cm.;A square root information filter and smoother (SRIF/SRIS) algorithm was implemented as a sequential processor using the reference orbit, residuals, measurement partial derivatives and variational partial derivatives from GEODYNII, Bierman's one-at-a-time propagation algorithm was shown to neglect the upper triangular elements of R;For the tropospheric refraction zenith delay, a first-order Gauss-Markov model with continuous process noise standard deviation (;The solar radiation pressure scale factor and y-bias acceleration, and the solar radiation pressure scale factor and radial, GPS X and GPS Y axes accelerations were also modeled as stochastic processes. The first-order Gauss-Markov model for the acceleration terms with... |