Abstract A model of ocean tides is generally used to remove the dynamic tidal signal in satellite gravimetric observations, such as GRACE(‐FO). Model imperfections can cause stripes and long period aliasing errors in monthly gravity solutions. Recent developments in ocean tide modeling have substantially improved accuracy at high latitudes and in shallow water regions. In this study, we compared the 8 major tidal constituents from three recent ocean tide models, namely EOT20, FES22 and GOT5.6, as well as GOT4.8, using 13‐year long GRACE ranging observations. The improvements of the three recent models are significant, showing more than 2 postfit variance reduction. Among them, GOT5.6 generally performs the best in terms of postfit variance reduction (11% vs. GOT4.8) and residual ocean tide fitting, particularly near West Antarctica. The residual ocean mass RMS, computed from monthly gravity solutions using GOT5.6, is on average 1 mm (GOT4.8), 0.3 mm (EOT20) and 0.1 mm (FES22) smaller. These results make it our preferred model for the upcoming GRACE(‐FO) Release 07 (RL07) re‐processing. Regarding implementation, utilizing the minor tides in GOT5.6 and long period tides in FES22 as well as considering geographically varying seawater density can further enhance model performance (denoted GOT5.6p3), which on average reduces the residual ocean mass RMS by an additional 7%. Notably, using lateral varying seawater density largely reduces the anomalous M 2 residuals in the North Atlantic Ocean. We anticipate that GOT5.6p3 will substantially reduce stripes and aliasing errors in forthcoming RL07 monthly solutions. The improvements will also benefit GOCE reprocessing and GRACE(‐FO) sub‐monthly solutions.
Zhang et al. (Fri,) studied this question.