Estimating GRACE aliasing errors

Abstract

Synthetic GRACE aliasing errors are examined for land, ocean and atmospheric signals. Simulated GRACE data are generated using least square fits of spherical harmonics to potential differences. Mass variations are produced by GLDAS (Global Land Data Assimilation Scheme), ECCO (Estimating the Circulation and Climate of Ocean), and the difference between NCEP and ECMWF (as an estimate of dealiasing error). Additional studies include single spherical harmonics and isolated (Gaussian) loads. Aliasing errors are measured by their degree-order spectrum, and arise from inadequate space-time sampling. Because atmospheric variations are rapid, they cause large aliasing errors at low degree, below 15. To better understand these, we examine aliasing due to sinusoidally oscillating single harmonics, and use Kaula's resonance solution to interpret the aliases. Single temporal frequency Gaussian mass load variations illustrate aliasing variations associated with geographical location.