The Impact of New Accelerometer Transplant Data (ACH) on GRACE Follow‐On Along‐Orbit Inter‐Satellite Laser Ranging Observations and Monthly Time‐Variable Gravity and Mascon Solutions

Abstract

GRACE‐D accelerometer data show significant bias jumps since one month after the launch of the GRACE Follow‐On (GRACE‐FO) satellites in May 2018, making them inapplicable for correcting GRACE‐FO products for non‐gravitational accelerations. The GRACE‐FO Science Data System (SDS) compensated this issue by transplanting the GRACE‐C accelerometer data toward that of GRACE‐D. Recently, GRACE‐FO SDS implemented an updated transplant method, used in the latest release of GRACE‐FO data. Here, we examine the impact of updated accelerometer transplant data (ACH) on GRACE‐FO measurements at all levels: (a) Level‐1B inter‐satellite laser ranging residuals measured along satellite orbit, (b) Level‐2 time‐variable gravity solutions from all SDS centers (JPL, CSR, and GFZ), and (c) Level‐3 mascon solutions. We show that inter‐satellite laser ranging residuals are modified at low frequencies below 1 mHz, affecting the along‐orbit analysis of large‐scale time‐variable gravity signals. When mapped into monthly Level‐2 spherical harmonic coefficients of geopotential, the low‐frequency change in inter‐satellite ranging residuals leads to substantial improvement of coefficients associated with resonant orders (i.e., 15, 30, 45, etc.) and C30. We also present an improved SLR‐derived C30 which significantly improves the agreement with updated GRACE‐FO C30 at seasonal and interannual timescales. Moreover, we demonstrate the noise reduction in mass change estimates from new GRACE‐FO Level‐2 data over oceans, Greenland, and Antarctica for all SDS solutions. GRACE‐FO mascon solutions show a moderate change in the updated release. Our comprehensive analyses demonstrate high‐quality estimates of non‐gravitational accelerations by the updated transplant method, resulting in more accurate GRACE‐FO time‐variable gravity and mass change observations.

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

“…In general, improvements of JPL‐ACH over its predecessor JPL‐ACT are apparent, consistent with the results of Ghobadi‐Far et al. (2023), while the combined solution has the lowest noise level for most of the months. When C 20 and C 30 coefficients are excluded, as shown in Figure 4b, OOR become slightly smaller and closer among different solutions.…”

“…From a visual inspection, the largest noise reduction is achieved when β ′ is close to zero, as pointed out by Ghobadi‐Far et al. (2023). Furthermore, we present the noise reduction rates of each month with respect to the corresponding β ′ angles in Figure 6.…”

“…(2021) and Ghobadi‐Far et al. (2023). Among the four, JPL‐ACH and COMB come closest to the SLR with RMS values less than 3 × 10 −11 in Table 3, which is at the same level as the C 30 formal error of the SLR.…”

“…It clearly demonstrates that the largest along‐track and radial acceleration differences occur near zero β ′. During this period, satellites undergo significant temperature fluctuations, resulting in substantial variations in non‐gravitational forces, particularly SRP, and therefore challenging the accelerometer data transplant (Behzadpour et al., 2021; Cheng & Ries, 2017; Ghobadi‐Far et al., 2023; Harvey et al., 2022; Klinger & Mayer‐Gürr, 2016). This largely explains why the most noticeable improvements or degradations in gravity field solutions using different ACT products are observed with zero β ′.…”

“…Ghobadi‐Far et al. (2023) found that JPL‐ACH leads to significantly improved C 30 estimates and consistently reduces solution noise compared to its predecessor, JPL‐ACT, through systematic comparisons between the SDS RL06 and RL06.1 solutions. However, independent assessments of all three operational ACT products are still lacking in the current literature, which will require consistent L1B data processing to give fair and meaningful results.…”