Abstract. The use of ground gravity data in wellsurveyed areas, continued upward to satellite altitude, is one of the most promising external absolute in-orbit calibration/ validation methods for GOCE satellite gravity gradient (SGG) observations. Based on a synthetic gravity test environment – providing in addition to statistical error information also absolute error estimates – several upward continuation methods, e.g. least squares collocation, equivalent source techniques using point masses or area density distributions defined on a spherical surface section, are described, assessed and compared. It turns out that all these strictly local approaches fail to work sufficiently accurate. Consequently, a combined adjustment strategy is proposed, supporting the high-quality gravity field information within the well-surveyed test area with a low-accuracy, but globally defined Earth model. Under quite realistic assumptions the upward continuation is performed with rms errors in the order of 1 mE. The most crucial limiting factor of this method is spectral leakage in the course of an adequate representation of the initial gravity information, which can be overcome by an enlargement of the parameter model in combination with a priori filtering of the initial gravity data.

Key words. Satellite gravity gradiometry – GOCE – calibration – least squares adjustment