Observed Inconsistencies between Satellite-Only and Surface Gravity-Only Geopotential Models

Abstract

The long wavelength inconsistencies observed between satellite-only and terrestrial-only gravitational solutions were re-examined, in view of the recent release by NIMA of an updated 1° mean gravity anomaly file (which was used in the development of the EGM96 geopotential model). The differences between the satellite-only model EGM96S and corresponding solutions obtained from 1°×1° terrestrial mean Ag were examined both spectrally and geographically. Up to N _max = 20, the global RMS geoid undulation difference (δN) between these models was ±3.7 m, for the NIMA 1° Ag data. This is an improvement over the ±4.4 m, obtained when the older OSU 1° Ag data were used. In some geographic regions however, the NIMA Δg data produce larger δN values with EGM96S, than the corresponding OSU anomalies. When the marine Δg data were replaced by altimetric values, the RMS δN dropped to ±1.6 m, indicating that more than 50% of the observed differences at long wavelengths is due to the poor quality of the available marine Ag. The spectrum of δN (EGM96S minus a terrestrial-only solution) exceeds by more than an order of magnitude the undulation spectra predicted by some postulated models of vertical datum inconsistencies. The spectrum of undulation effects implied by the approximation H* ≈ H, is quite similar to that predicted by one of the vertical datum inconsistency models postulated by Laskowski [1983].