Assessment of modern lunar gravity field models through orbit analysis

Abstract

All present-day lunar gravity field models are — with the exception of a minor contribution from lunar laser ranging data to the second-degree harmonics — satellite-only models. Orbit analysis is therefore a fundamental tool for quality assessment, alongside arguments of compatibility imposed by selenophysics. A general problem in selenodesy and selenophysics is the lack of independent data to serve purposes of calibration and validation of basically any result. Model assessment techniques may therefore be quasi-circular and far from independent. Regarding the use of prior information from lunar physics, primarily the assumptions on coefficient amplitudes and the distribution of the selenopotential signal power over the degrees, the indicia of their correctness are for the larger part provided by the induced improvement in measurement fit of the satellite orbits. One prob-lem of satellite orbit and orbit error measures, on the other hand, is that they are “global” measures of the selenopotential quality, as opposed to a direct function of location on the lunar sphere, and that RMS-of-fit values over several-day satellite arcs actually contain very little spatial information on the intrinsic selenopotential model quality. Nevertheless, gravity field models capable of predicting reasonable mass anomaly structures and estimates of crustal thickness tend to be considered reliable. It should therefore be clear that it is only through continued iteration over the available data sets, along with improved modelling in several branches of selenoscience, that solid quality assessment and statistically significant quality improvement of lunar gravity field models may be achieved.

‘...the greatest usefulness of linear perturbation theories for orbital motion error studies is probably in understanding what information various satellite tracking data contribute to a particular gravity model solution. To this end, it can be seen that an accurate determination of the gravity coefficient error covariance is as important as the determination of the actual coefficient values.’

G. W. Rosborough, 1987