Answers are grounded in this paper and cite their sources.
Analysis reveals asymptotic bias in covariance estimators and inconsistency in cross-variograms for axially symmetric Gaussian processes.
Axially symmetric processes, those stationary in longitude but nonstationary across latitude, provide a flexible and physically meaningful class of models for global environmental data. Despite their wide use, the asymptotic properties of classical method-of-moments (MOM) estimators for these processes remain largely unexamined. In this work, we investigate MOM estimators of covariances and cross-variograms for axially symmetric Gaussian processes observed on regular latitude-longitude grids. First, we show that MOM covariance estimators are asymptotically biased. We then examine MOM estimators of cross-variograms, and prove that they are unbiased. However, using the block circulant structure of the covariance matrix and its Fourier diagonalization, we establish MOM cross-variogram estimators are not consistent. These findings illustrate intrinsic limitations of MOM-type estimators on compact manifolds and emphasize that Euclidean intuition does not carry over to spherical settings.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: