next up previous
Next: 4. Summary Up: 3.2. Modelling the Previous: 3.2.3. Atmospheric density

3.2.4. Orbit reduction methods.

Despite the tremendous improvements in force modeling previously discussed, knowledge of the motion of a satellite is not absolute. Imperfect or incomplete non-conservative force models, rather than geopotential uncertainties, are now the dominant source of orbit positioning errors for complex Earth observing platforms. Even the most detailed and computationally intensive models for these surface forces will be limited by uncertainties in the time varying orientation, material properties and temperature of the satellite surfaces. However, improvements in modeling the full complement of forces can bring ephemeris knowledge close enough to truth to permit employment of empirical and/or stochastic accelerations adjustments. This approach accommodates the small remaining effects sensed by highly precise tracking data to heighten orbit accuracy.

Much of this mismodeling occurs at the once-per-orbital-revolution frequency. Accordingly, constant and one-cycle-per-revolution empirical accelerations are used to accommodate the effect [ Colombo, 1989]. This technique has been used successfully for T/P in both the SLR/DORIS [ Tapley et al., 1994a] and the GPS [ Bertiger et al., 1994] dynamic orbit solutions. Furthermore, in the GPS ``reduced dynamic'' approach, the converged dynamic orbit is used as an a priori ephemeris and constant, three-dimensional stochastic accelerations spaced at short intervals are estimated, virtually eliminating any residual dynamic force modeling error in the resulting orbits [ Wu et al., 1991; Yunck et al., 1990, 1994]. The extensive temporal and spatial tracking coverage provided by DORIS and GPS permit this routine recovery of extensive empirical/stochastic force model parameters. The use of these empirically determined effects, although effective, should not be considered as a substitute for correct modeling and can misposition the satellite during times of data anomalies. These approaches have allowed sub-4 cm radial accuracy to be achieved on T/P. Empirical parameterization in the dynamic approach is not a panacea for dynamic model deficiencies, however, since these parameters can absorb the signals arising from the geophysical processes one is trying to observe.



next up previous
Next: 4. Summary Up: 3.2. Modelling the Previous: 3.2.3. Atmospheric density



U.S. National Report to IUGG, 1991-1994
Rev. Geophys. Vol. 33 Suppl., © 1995 American Geophysical Union