Improved GRACE‐FO Gravity Field Solution by Combining Different Accelerometer Transplant Products

Abstract Gravity field solutions determined from the Gravity Recovery and Climate Experiment Follow‐on (GRACE‐FO) are affected by the unexpected performance degradation of one accelerometer onboard, which is circumvented by the accelerometer data transplant (ACT) technique. Three operational ACT products are presently available: JPL‐ACT and JPL‐ACH from NASA Jet Propulsion Laboratory (JPL) and TUG‐ACT from Graz University of Technology (TUG). Considering the heterogeneous qualities of individual ACT products, we propose to combine them to improve the gravity field models after a comprehensive assessment of individual products using a consistent Level‐1B data processing. The combination is carried out on the normal equation of gravity field solutions. Spectral, temporal, and spatial evaluations over the period October 2018 to December 2022 demonstrate that the combined solution significantly reduces the noise relative to solutions using individual ACT products, and the average noise reduction rates over JPL‐ACT range from 5% to 12% with varying post‐processing filters and C 20 and C 30 coefficient treatments, which are double those of the second‐best JPL‐ACH. Moreover, the C 20 estimate exhibits significant improvement in the combined solution with its deviation from the satellite laser ranging (SLR) solution showing a root mean square (RMS) value of 7.6 × 10 −11 , compared to similar RMS values of 15.9 × 10 −11 , 12.1 × 10 −11 , and 14.9 × 10 −11 for JPL‐ACT, JPL‐ACH, and TUG‐ACT, respectively. This improvement helps greatly reduce the amplitude of the 161‐day spurious signal over polar regions. For C 30 , the JPL‐ACT gives degraded estimates with an RMS of differences to SLR of 6.2 × 10 −11 , while the differences are greatly reduced to 2.6 × 10 −11 , 2.7 × 10 −11 and 3.6 × 10 −11 for the combined solution, JPL‐ACH and TUG‐ACT, respectively, approaching the formal error of SLR estimates.