On accuracy of current geopotential models estimated through comparison of quasi-geoid models and GPS/levelling data

Gravity-dedicated satellite missions CHAMP and GRACE provide accurate data that can be inverted into geopotential coefficients forming global geopotential models (GGM). Through an estimation process, mean square errors of these coefficients can be elvaluated. In most cases, even the entire variance-covariance matrices are computed. However, these errors are only formaI and do not represent weIl the actual accuracy of estimated coefficients, thus also of various gravity field parameters synthesized from these coefficients. For this reason validation procedures for the new GGMs are being sought since the classical validation methocts reached their limits. This article discusses the validation procedure of the GGM through its comparison with idependent data estimated at selected GPS/leveIling stations over the area of Central Europe. Due to a different spectral content of height anomalies synthesized from the GGM and of those derived from combination of ellipsoidal and normal heights at selected points of the European Vertical Reference Network, the GGM-based low-frequency height anomaly is completed for a missing higth-frequency component based on high resolution and high accuracy ground gravity and elevation data. The methodology is also applied on a set of GPS/leveIling stations in the Czech Republic. In accordance with previous validation tests of GGMs based on data of crossover altimetry, obtained results indicate that the current GGMs estimated from single/dual-sateIlite data seem to have significant problems namely with low-order and degree coefficients.