The chapter discusses the analysis of mass variability in the Earth system using GRACE measurements. The authors compare the degree correlations between GRACE and GLDAS (Global Land Data Assimilation System) for the annual geoid variations. They use spherical harmonic coefficients to describe the geopotential and measure the power as a function of spatial scale through degree variance. The degree correlation between GRACE and GLDAS is found to be relatively high at the lowest degrees (longest spatial scales) but becomes uncorrelated between degrees 30-40 (spatial features of 500-650 km). The discrepancy between GRACE and GLDAS is attributed to GRACE errors growing with increasing degree, the incomplete representation of the GLDAS model, and unmodeled or incorrectly modeled atmospheric and ocean mass variations. The chapter also explores the error versus smoothing radius, explaining how the error in geoid height calculation depends on the spherical harmonic coefficients and how Gaussian weighted averages can be used to smooth these errors. Additionally, it discusses the generation of geoid height errors through the factorization of the error covariance matrix and the use of random realizations to illustrate the possible errors in geoid height maps. Supporting figures and references are provided to illustrate these concepts.The chapter discusses the analysis of mass variability in the Earth system using GRACE measurements. The authors compare the degree correlations between GRACE and GLDAS (Global Land Data Assimilation System) for the annual geoid variations. They use spherical harmonic coefficients to describe the geopotential and measure the power as a function of spatial scale through degree variance. The degree correlation between GRACE and GLDAS is found to be relatively high at the lowest degrees (longest spatial scales) but becomes uncorrelated between degrees 30-40 (spatial features of 500-650 km). The discrepancy between GRACE and GLDAS is attributed to GRACE errors growing with increasing degree, the incomplete representation of the GLDAS model, and unmodeled or incorrectly modeled atmospheric and ocean mass variations. The chapter also explores the error versus smoothing radius, explaining how the error in geoid height calculation depends on the spherical harmonic coefficients and how Gaussian weighted averages can be used to smooth these errors. Additionally, it discusses the generation of geoid height errors through the factorization of the error covariance matrix and the use of random realizations to illustrate the possible errors in geoid height maps. Supporting figures and references are provided to illustrate these concepts.