The Intriguing Structure of Stripes in GRACE Geopotential Models

Abstract

Geopotential models derived from Gravity Recovery and Climate Experiment (GRACE) mission measurements are significantly obscured by the presence of a systematic artifact, known as longitudinal stripes. Based on our previous work (Peidou and Pagiatakis, 2020) we provide an in-depth analysis of the latitudinal sampling characteristics of GRACE and we reveal the intriguing sampling mechanism that creates sub-Nyquist artifacts (stripes). Because the sub-Nyquist artifacts are poorly understood, we provide a simple simulation example to elucidate the mechanism of the sub-Nyquist artifact generation. Subsequently, we randomly select June 2009 daily GPS precise science orbits for GRACE-A to produce ground tracks to sample the low frequency disturbing potential (geoid) along the parallel of ϕ=10° N. The sampled geoid is then deinterlaced in space to produce a monthly data sequence whose detailed analysis shows that the sub-Nyquist artifacts (stripes) are produced from a critical sampling rate of the low degree gravitational field that is related to the ratio m/n of two mutually prime integers, where m is the number of days it takes to have a nearly repeat orbit and n is the number of complete orbits in one day. We perform extensive analyses of GRACE Level-2 data over a period of eight years to show the variability in the orbital characteristics that are directly linked to the orbit resonances (via integers m and n). It turns out that during short repeat cycle resonances the stripes are amplified. Finally, to minimize the presence of stripes in Level-2 data products, it is recommended that orbits of future missions should be designed to avoid the critical m/n ratios while appropriately monitoring and adjusting them during the mission. For completed missions, or missions that are already active, force modelling the latitudinal low frequency disturbing potential may be a viable and most preferred approach to filtering.