U. Meyer

AIUB-GRACE02S: Status of GRACE Gravity Field Recovery Using the Celestial Mechanics Approach

The gravity field model AIUB-GRACE02S is the second release of a model generated with the Celestial Mechanics Approach using GRACE data. Inter-satellite K-band range-rate measurements and GPS-derived kinematic positions serve as observations to solve for the Earth’s static gravity field in a generalized orbit determination problem. Apart from the normalized spherical harmonic coefficients up to degree 150, arc-specific parameters like initial conditions and pseudo-stochastic parameters are solved for in a rigorous least-squares adjustment based on both types of observations. The quality of AIUB-GRACE02S has significantly improved with respect to the earlier release 01 due to a refined orbit parametrization and the implementation of all relevant background models. AIUB-GRACE02S is based on 2 years of data and was derived in one iteration step from EGM96, which served as a priori gravity field model. Comparisons with levelling data and models from other groups are used to assess the suitability of the Celestial Mechanics Approach for GRACE gravity field determination.

GOCE: assessment of GPS-only gravity field determination

The GOCE satellite was orbiting the Earth in a Sun-synchronous orbit at a very low altitude for more than 4xa0years. This low orbit and the availability of high-quality data make it worthwhile to assess the contribution of GOCE GPS data to the recovery of both the static and time-variable gravity fields. We use the kinematic positions of the official GOCE precise science orbit (PSO) product to perform gravity field determination using the Celestial Mechanics Approach. The generated gravity field solutions reveal severe systematic errors centered along the geomagnetic equator. Their size is significantly coupled with the ionospheric density and thus generally increasing over the mission period. The systematic errors may be traced back to the kinematic positions of the PSO product and eventually to the ionosphere-free GPS carrier phase observations used for orbit determination. As they cannot be explained by the current higher order ionospheric correction model recommended by the IERS Conventions 2010, an empirical approach is presented by discarding GPS data affected by large ionospheric changes. Such a measure yields a strong reduction of the systematic errors along the geomagnetic equator in the gravity field recovery, and only marginally reduces the set of useable kinematic positions by at maximum 6xa0% for severe ionosphere conditions. Eventually it is shown that GOCE gravity field solutions based on kinematic positions have a limited sensitivity to the largest annual signal related to land hydrology.

The Effect of Pseudo-Stochastic Orbit Parameters on GRACE Monthly Gravity Fields: Insights from Lumped Coefficients

The official GFZ RL05 monthly GRACE gravity models were processed in a two-step approach. In the first step the orbits were determined. In the second step corrections to the gravity field parameters were estimated, while the orbits were kept fixed. This led to a significant de-noising of the resulting monthly models, but accidentally also to a regularization, i.e., the estimated gravity field coefficients were biased towards the a priori model. We compare the GFZ RL05 models to a revised version RL05a that was determined in a common estimation of orbit and force model parameters. A large number of gravity field coefficients is significantly affected. We relate this effect to the one-hourly stochastic accelerations estimated for orbit determination, and to ignoring the correlations.The official GFZ RL05 monthly GRACE gravity models were processed in a two-step approach. In the first step the orbits were determined. In the second step corrections to the gravity field parameters were estimated, while the orbits were kept fixed. This led to a significant de-noising of the resulting monthly models, but accidentally also to a regularization, i.e., the estimated gravity field coefficients were biased towards the a priori model. We compare the GFZ RL05 models to a revised version RL05a that was determined in a common estimation of orbit and force model parameters. A large number of gravity field coefficients is significantly affected. We relate this effect to the one-hourly stochastic accelerations estimated for orbit determination, and to ignoring the correlations. In the main part of this paper we study the interaction between pseudo-stochastic orbit parameters and gravity field coefficients. To explain this interaction we make use of a time-wise approach to gravity field determination. We apply the linear perturbation theory developed by Hill for circular orbits to compute lumped coefficients of the inter-satellite range-rate observations. We illustrate that the pseudo-stochastic orbit parameters act as a high-pass filter on the lumped coefficients spectra of the range-rates. Because the lumped coefficients are related to the spherical harmonics coefficients via a summation over all degrees, the whole range of gravity field coefficients is affected. This result is of relevance for all approaches to gravity field estimation from orbit observations, where dynamic orbits are introduced a priori and the arc-specific parameters are kept fixed.