Processing Choices Affect Ocean Mass Estimates From GRACE

Abstract

Accurate estimates of ocean mass change are necessary to infer steric sea level change from sea level changes measured with satellite altimeters. Published studies using the Gravity Recovery and Climate Experiment (GRACE) satellite mission indicated a large range in trends (∼1–2 mm/year) with reported standard errors of 0.1–0.3 mm/year. Here we show that a large part of this discrepancy (up to 0.6 mm/year) can be explained by which model is used to account for the effect of glacial isostatic adjustment (GIA). The second largest contribution (0.3–0.4 mm/year) is related to the way how different studies have restored atmospheric and oceanic signals which have been removed during the GRACE gravity estimation process. Here two processing strategies, which previously resulted in differing ocean mass trends, are considered. The “direct” method uses the standard GRACE Stokes coefficients, while the “inverse” method applies a joint inversion of data from GRACE and altimetry. After accounting for differences in processing corrections, global mean ocean mass estimates from the direct, the mascon, and inverse approach agree with each other on global scales within less than 0.1 mm/year. Using the A et al. (2013; https://doi.org/10.1093/gji/ggs030) GIA model, we provide a reconciled monthly time series of global mean ocean mass, which suggests that ocean mass has increased by 1.43 mm/year over 2002.6–2014.5, with an amplified rate of 1.75 mm/year over 2002.6–2016.5 which covers almost the complete GRACE time span. However, we note that estimates as low as 1.05 mm/year cannot be ruled out when other published GIA corrections with lower mass‐equivalent signals over Antarctica are used.

Can We Resolve the Basin‐Scale Sea Level Trend Budget From GRACE Ocean Mass?

“…It is well understood that the choice of data center, processing choices, and GIA forward‐model can significantly affect the basin‐scale SSH trend, particularly in those areas where the GIA signal has a large amplitude (e.g., Marcos et al, ). It is also clear that the choice of GRACE processing produces significant variations in the trend, which corresponds with long‐wavelength or low‐degree spherical harmonic coefficient differences (similarly shown by Blazquez et al, , Uebbing et al, , and others). In our analysis, the subpolar and subtropical North Atlantic region means are most affected by the choice of GRACE product and GIA correction.…”

Improved Earth Oblateness Rate Reveals Increased Ice Sheet Losses and Mass‐Driven Sea Level Rise

“…To compute the mass component of GMSL, we use the CSR RL06 GRACE monthly gravity field product, apply a 300-km coastal buffer (Johnson & Chambers, 2013), and correct for the GIA model of Geruo et al (2013). Substituting the JPL RL06 monthly gravity field product has a negligible effect, and the impact of the selected ocean mask and GIA model is discussed by Uebbing et al (2019). When replacing the GRACE C 20 with the GSFC recommended and TN11 solutions, we obtain ocean mass trends and goodness of fit uncertainties of 2.31 ± 0.07 mm/year and 2.23 ± 0.07 mm/year, respectively.…”

“…Total GMSL variability has been continuously measured since 1992 with a series of sea surface altimetry satellites, and should equal the sum of the mass component measured by GRACE, and the steric component observed by a global network of profiling floats. Substituting the JPL RL06 monthly gravity field product has a negligible effect, and the impact of the selected ocean mask and GIA model is discussed by Uebbing et al (2019). For assessing the GMSL budget closure, we consider the recent international collaborative effort of the World Climate Research Programme (Group, WCRP Global Sea Level Budget, 2018), which presents a total GMSL trend of 3.70 ± 0.35 mm/year and full-depth steric trend of 1.31 ± 0.20 mm/year for their ensemble solutions from 2005-2015 (all uncertainties are 1-).…”

Improved Quantification of Global Mean Ocean Mass Change Using GRACE Satellite Gravimetry Measurements

“…We include two additional cases (8 and 9) to add GRACE GAD fields back to GSM (i.e., GSM + GAD, but with the GAD mean over the ocean removed), and compute GMOM change using ocean kernels (500 and 300 km) following the method of Uebbing et al (2019). We include two additional cases (8 and 9) to add GRACE GAD fields back to GSM (i.e., GSM + GAD, but with the GAD mean over the ocean removed), and compute GMOM change using ocean kernels (500 and 300 km) following the method of Uebbing et al (2019).…”

“…To better understand how various parameters affect estimates, we consider varying buffer zone size of the ocean mask, low-degree SH coefficients (in particular ΔJ 2 ), latitude range of GMOM integration (global or ±64.5°), and how an alternate treatment of spatial leakage (FM) would affect GRACE GMOM estimates. We include two additional cases (8 and 9) to add GRACE GAD fields back to GSM (i.e., GSM + GAD, but with the GAD mean over the ocean removed), and compute GMOM change using ocean kernels (500 and 300 km) following the method of Uebbing et al (2019). GAD fields are GRACE supplementary data products containing gravity change over the ocean due to nontidal atmospheric surface pressure and high frequency oceanic mass changes used in the GRACE dealiasing process (Bettadpur, 2018).…”