The Method of Normalized Correlations: A Fast Parameter Estimation Method for Random Processes and Isotropic Random Fields That Focuses on Short-Range Dependence

摘要

We focus on the inference of covariance parameters from spatial or temporal data using the method of normalized correlations (MoNC). We compare the MoNC with maximum likelihood estimation and weighted least squares using simulated data, as well as gridded and scattered real data. For the synthetic experiments, none of the methods considered performed uniformly better than the others in terms of estimation accuracy and precision; however, for regularly spaced data, the MoNC was significantly faster and practically insensitive to sample size. For scattered data, the MoNC's speed was reduced but still competitive. Potential applications of the MoNC include those involving large data sets and fast preconditioning.