The role of accelerometer data calibration within the ITSG-Grace2016 release: impact on C20 coefficients

We present a two-step approach used for accelerometer data calibration within the ITSG-Grace2016 release and analyze its impact on the recovered gravity field solutions, especially on the C20 coefficients.
Within the ITSG-Grace2016 release, the accelerometer biases are estimated daily using uniform cubic basis splines (UCBS), the scale factors are also estimated daily using a fully-populated scale factor matrix. Therefore, not only the sale factors in along-track, cross-track and radial direction are estimated, but also the non-orthogonality of the accelerometer axes and the misalignment between the Accelerometer Frame (AF) and the Science Reference Frame (SRF) are taken into account. This approach aims at improving the gravity field recovery, hence it does not guarantee a physically correct model. The setup of the calibration parameters is likely to also absorb mismodeled or non-accelerometer induced spurious signals that otherwise map into the gravity field coefficients.
For the ITSG-Grace2016 release, the improved calibration parametrization not only contributed to a noise reduction, but also significantly improved the estimates of the C20 coefficients. We show that the offset between SLR and GRACE derived C20 time series can be reduced remarkably by the use of a fully-populated scale factor matrix, demonstrating the merit of this new approach. Based on our results, we suggest the presence of a
clear temperature-dependent behavior (biases and scale factors) and the presence of off-diagonal elements within the accelerometer scale factor matrix.