Utilization of simulated GRACE inter-satellite range-accelerations to estimate Earth's gravity field

Smith, Matthew Scott
Journal Title
Journal ISSN
Volume Title

The Gravity Recovery and Climate Experiment (GRACE) provides high-precision K-band Ranging (KBR) data which has been instrumental in improving our understanding of the monthly mass redistribution within the Earth system, and consequently its static and time-varying gravity fields. In practice, estimation of the Earth's gravity field with data from GRACE-like missions is typically done via the range-rate pseudo-observations. This approach is widely used and produces high-quality solutions, however there does exist a well-known North-South striping error in the resulting gravity field. It is thought there may be a potential benefit from utilizing instead the range-acceleration pseudo-observations, which should be sensitive to more spatially-localized mass variations in the signal, thereby reducing the N-S errors in the gravity field and facilitating more precise estimation to higher degrees. Most solutions obtained from range-accelerations to date have been unusable at worst and lesser in quality at best when compared to range-rate derived gravity field solutions. Current understanding is that this is due to the time-differentiation of the KBR signal required to obtain the range-acceleration measurements. The differentiation process acts as a high-pass filter, degrading the signal-to-noise ratio (SNR) at high frequencies, and thus the quality of the solution. The purpose of this work, which explores variational methods solely, is to discover what conditions, if any, make it possible to generate feasible solutions via range-accelerations, and to compare them to one obtained via range-rate. A 180x180 range-rate based gravity field solution produced from simulated August 2008 data was used as a baseline for these comparisons. It is demonstrated that adjusting the parameters of the currently-used filter for obtaining the range-accelerations provides some improvement in the resulting solutions. Conversely, attempts with an alternative approach to filtering the range measurements yielded no benefit over the current method, and only served to degrade the solutions further. However, through an application of filtering the range residuals instead, this research suggests that the culprit is not solely the noise induced by differentiation, but the inclusion of other noisy measurements necessary for the computation of the range-acceleration measurement equation. Through this new method, it is shown that not only are range-accelerations viable for estimating the gravity field, but they can produce solutions more accurate at higher degrees than their range-rate counterparts. While these results are encouraging for processing the range-accelerations, the same technique can be applied to range-rate based solutions, which produces similar improvements and again establishes that quantity as the most suitable for estimating the gravity field, for now.