Tests and comparisons of gravity models

Optical observations of the GEOS satellites were used to obtain orbital solutions with different sets of geopotential coefficients. The solutions were compared before and after modification to high order terms (necessary because of resonance) and then analyzed by comparing subsequent observations with predicted trajectories. The most important source of error in orbit determination and prediction for the GEOS satellites is the effect of resonance found in most published sets of geopotential coefficients. Modifications to the sets yield greatly improved orbits in most cases.The sets of coefficients analyzed are APL 3.5, NWL5E-6, Köhnlein (1967), Rapp (1967), Kaula (1967), Smithsonian Astrophysical Observatory (SAO)M-1 (1966), SAO AGU (1969), SAO COSPAR (1969) and SAO 1969 Standard Earth. The SAO 1969 models generally give better orbital fits and prediction results than the other models above. However these models can be improved by corrections to resonant coefficients.The results of these comparisons suggest that with the best optical tracking systems and gravity models, satellite position error due to gravity model uncertainty can reach 50–100 m during a heavily observed 5–6 day orbital arc. If resonant coefficients are estimated, the uncertainty is reduced considerably.