# Estimating weights for the use of time‐dependent gravity recovery and climate experiment data in constraining ocean models

Type: Journal Article

Venue: Journal of Geophysical Research

Citation:

Quinn, K. J. and R. M. Ponte (2008), Estimating weights for the use of time‐dependent gravity recovery and climate experiment data in constraining ocean models, J. Geophys. Res., 113, C12013, doi:10.1029/2008JC004903.

Using Gravity Recovery And Climate Experiment (GRACE) data to constrain ocean general circulation models requires quantitative knowledge of the errors in GRACE‐derived estimates of ocean bottom pressure (pb) change, which for our purposes include not only instrument noise but also variability not represented in the models (e.g., post‐glacial rebound and self‐gravitation effects). We attempt a spatial mapping of these errors by comparing several GRACE data products to pb simulations from an ocean model. Uncertainties in the global ocean mean $\overline{p_{b}}$, partly related to the net freshwater flux into the ocean, and in the regional pb anomalies about that mean are considered separately. The resultant regional error estimates (∼1–3 cm), when zonally averaged, are comparable to the calibrated errors provided by the GRACE processing centers, except for enhanced errors near some continental regions with high seasonal hydrology signals or large mass trends. Errors in the GRACE‐derived $\overline{p_{b}}$ values estimated from model‐data differences (∼0.2 cm) are also comparable with those from the calibrated errors. For both pb and $\overline{p_{b}}$ estimates, accounting for the effects of geocenter noise is important. Replacing the C20 harmonic term in the GRACE data with estimates derived from satellite laser ranging results in significantly lower errors in the Southern Ocean. We also find lower errors at high latitudes when the variability of the atmospheric pressure over the land is removed from the data. Given the estimated errors and model‐data comparisons, GRACE data should be useful for constraining estimates of $\overline{p_{b}}$, particularly at interannual periods, but less so when considering regional pb variability.