High-resolution gravity field modeling with full variance–covariance matrices

Abstract.

 High-resolution gravity field models are currently computed by a combined least-squares adjustment with the full variance–covariance matrix for the lower degrees and a simplified approach for the higher degrees. Simplification for the high degrees means that numerical quadrature is applied or that the structure of the covariance matrix is reduced to block diagonals. Both methods have been used for several years to compute high-resolution gravity field models. With recent improvements in algorithms and with the use of parallel computers, the degree and order for full variance–covariance matrices could be increased to 180. Several test solutions combining full and block-diagonal normals were computed and compared to independent data sets. These comparisons were made in order to quantify the impact of reducing the normal structures on the gravity field solution. Results showed that, particularly in the spectral domain (spherical harmonic coefficients), the impact cannot be neglected, while in the space domain (e.g. geoid height grid) the impact is almost negligible.