On the modeling of long wavelength systematic errors in surface gravimetric data

Abstract

A satellite-only model and a set of 1 degrees x1 degrees area-mean terrestrial gravity anomalies were used to estimate simultaneously geopotential coefficients to degree and order 70 and a set of spherical harmonic coefficients representing regional systematic errors in the gravity data. Several test solutions were developed whereby the weighting of the gravity data and the maximum degree and order of the systematic bias coefficient set were varied. The results were evaluated using both internal consistency statistics (e.g., a posteriori signal and error statistics, calibration factors) and comparisons with independent data (orbit fits, comparisons with GPS/leveling-derived geoid undulations). The global RMS error correlation between geopotential and bias coefficients was 6.7% for an unconstrained bias expansion to degree 20, and 7.8% for an expansion to degree 25, where an apriori constraint on the bias parameters was also employed. The bias expansion to degree 25 resulted in a slight degradation of the results from comparisons with GPS/leveling-derived geoid undulations or height anomalies. Optimization of this technique requires additional tests, especially with regard to terrestrial data weighting issues. Future satellite missions (e.g., CHAMP, GRACE and GOCE) will permit bias recovery at finer resolution, thereby improving the overall effectiveness of this technique.