Comparison of Satellite-Only Gravity Field Models Constructed with All and Parts of the GOCE Gravity Gradient Dataset

ABSTRACT

The impact of GOCE Satellite Gravity Gradiometer data on gravity field models was tested. All models were constructed with the same Laser Geodynamics Satellite (LAGEOS) and Gravity Recovery and Climate Experiment (GRACE) data, which were combined with one or two of the diagonal gravity gradient components for the entire GOCE mission (November 2009 to October 2013). The Stokes coefficients were estimated by solving large normal equation (NE) systems (i.e., the direct numerical approach). The models were evaluated through comparisons with the European Space Agency's (ESA) gravity field model DIR-R5, by GPS/Leveling, GOCE orbit determination, and geostrophic current evaluations. Among the single gradient models, only the model constructed with the vertical ZZ gradients gave good results that were in agreement with the formal errors. The model based only on XX gradients is the least accurate. The orbit results for all models are very close and confirm this finding. All models constructed with two diagonal gradient components are more accurate than the ZZ-only model due to doubling the amount of data and having two complementary observation directions. This translates also to a slower increase of model errors with spatial resolution. The different evaluation methods cannot unambiguously identify the most accurate two-component model. They do not always agree, emphasizing the importance of evaluating models using many different methods. The XZ gravity gradient gives a small positive contribution to model accuracy.

KEYWORDS:

• GOCE
• gravity gradients
• gravity model accuracy
• geostrophic currents

Funding

This work was partly financed through ESA contract no. 18308/04/NL/MM. CNES TOSCA supported this study.