This supporting text discusses the comparison of GRACE and GLDAS annual cycle data using degree correlations. The geopotential is described by spherical harmonic coefficients, and the degree correlation measures the similarity between two sets of coefficients. Comparisons between GRACE and GLDAS showed that they are well correlated at low degrees (large spatial scales) but become uncorrelated between degrees 30-40 (smaller spatial scales). The discrepancy between GRACE and GLDAS is due to GRACE errors increasing with degree and GLDAS not including certain gravity signals. Degree correlation alone does not validate GRACE results' accuracy at specific scales. The error in geoid height calculation depends on all spherical harmonic coefficients. However, practical error estimates are limited to a maximum degree and order. A Gaussian-weighted average can be used to calculate smoothed geoid height estimates, which is useful for visualizing gravity variability. The error propagation method can be used to calculate the global variance of geoid height errors for each weighting function. The error covariance matrix predicts error power and correlation between gravity coefficients. To illustrate possible geoid height errors, a realization of individual coefficient errors is needed. This involves factoring the covariance matrix into a triangular matrix and multiplying it by a vector of random numbers to generate coefficient errors. A geoid height error map can then be calculated. Figures S1, S2, and S3 illustrate the degree correlations between GRACE and GLDAS, the resolution of Gaussian weighting functions, and random realizations of geoid height errors, respectively. The error distribution varies with each realization, but the nature of the errors is similar across maps. Multiple realizations are generally useful for understanding error patterns.This supporting text discusses the comparison of GRACE and GLDAS annual cycle data using degree correlations. The geopotential is described by spherical harmonic coefficients, and the degree correlation measures the similarity between two sets of coefficients. Comparisons between GRACE and GLDAS showed that they are well correlated at low degrees (large spatial scales) but become uncorrelated between degrees 30-40 (smaller spatial scales). The discrepancy between GRACE and GLDAS is due to GRACE errors increasing with degree and GLDAS not including certain gravity signals. Degree correlation alone does not validate GRACE results' accuracy at specific scales. The error in geoid height calculation depends on all spherical harmonic coefficients. However, practical error estimates are limited to a maximum degree and order. A Gaussian-weighted average can be used to calculate smoothed geoid height estimates, which is useful for visualizing gravity variability. The error propagation method can be used to calculate the global variance of geoid height errors for each weighting function. The error covariance matrix predicts error power and correlation between gravity coefficients. To illustrate possible geoid height errors, a realization of individual coefficient errors is needed. This involves factoring the covariance matrix into a triangular matrix and multiplying it by a vector of random numbers to generate coefficient errors. A geoid height error map can then be calculated. Figures S1, S2, and S3 illustrate the degree correlations between GRACE and GLDAS, the resolution of Gaussian weighting functions, and random realizations of geoid height errors, respectively. The error distribution varies with each realization, but the nature of the errors is similar across maps. Multiple realizations are generally useful for understanding error patterns.