Using GRACE to Improve Altimetry's Ocean De-Aliasing Model

We demonstrate that daily Gravity Recovery And Climate Experiment (GRACE) data can be used to improve the non-tidal Dynamic Atmospheric Correction (DAC) used to de-alias Jason altimetry and reduce high-frequency errors. To accomplish this, we compute empirical orthogonal function (EOF) maps based on the ocean-model part of the DAC, then fit the GRACE and Jason data to those maps. The resulting DAC and GRACE EOF time-series are compared to the altimetry EOF time-series, to determine which best represents the state of the ocean. We find that, for the first six modes in the North Pacific, 19 modes in the Southern Ocean, and three modes in the combined Indian/Atlantic/Equatorial-Pacific region, GRACE is able to explain more of the 10-20-day altimetry signal than the original de-aliasing model can. We recombine these first GRACE modes with the remaining ocean-only DAC data to produce an improved 10-20-day model during times when GRACE data exists. We smoothly blend that series with the original model during times when there is no GRACE data, also restoring the faster frequencies from the original. The result is a blended model which explains 10%-30% more of altimetry's 10-20-day variance in the Southern Ocean than the original DAC model did, 5%-10% more in the Atlantic and Indian Oceans, and is functionally equivalent in the Equatorial Pacific. The removal of the additional sub-Nyquist-frequency signals will result in better Jason altimetry data with fewer aliased signals incorrectly propagated to longer periods.