Corrections for Accelerometer Noise for Data from the GRACE Follow-On Mission

A substantial reduction in the level of low-frequency noise is expected for the accelerometer data from the GRACE Follow-On satellites. However, a proceedure to at least partly correct for this noise and for calibration constant errors will still be needed. For timely analyses of perhaps monthly sets of data, the best available a priori information on variations in the Earth's mass distribution would be used in determining the corrections. But, if the hydrology, ice and snow data is not available promptly, it appears that the a priori uncertainties in the mass variations on land may be considerably higher than those in favorable low latitude regions of the oceans. In particular, nearly real time mass variation models such as the ECCO and OMCT ocean models appear to have quite high accuracy at latitudes between -30 and +30 deg for the Pacific Ocean and for parts of the Atlantic and Indian Oceans, provided that boundary current regions are not included. Thus, if only quite low frequency corrections are needed for the accelerometer data, it is possible about half of the time to use only the GRACE Follow-On data over such regions in solving for the accelerometer corrections. Such correction proceedures, which are called Ocean Calibration proceedures, need to be evaluated to determine if the higher accuracy data over the favorable areas makes up for the considerably smaller calibration area that would be used. For a single pair of satellites, temporal aliasing will still be the dominant error source, but reducing the accelerometer noise error contribution still could be useful. For a possible future Earth gravity variation program that might have two pairs of satellites, it would be useful to know how low the accelerometer noise or drag-free noise level would need to be in order to have such noise be only a quite minor contributor to the total mass variation uncertainties, with any of the possible instrumental noise correction proceedures. Iteration between the calibration proceedure and the mass variation solutions can help, but its effect is limited by the temporal aliasing. Weighting the data from some areas differently for the calibration proceedure but not for the main mass distribution solutions appears desirable if the instrumental noise error contributions don't vary substantially with location.