Karhunen-Loeve Expansion for Massive Spatial Data

摘要

Massive spatial data are observed in environmental studies and present a computational challenge to statistical analysis because of the large covariance matrix involved. The KL expansion is a useful tool for analyzing massive spatial data because it can reduce the model complexity. This article develops and compares two algorithms to obtain the eigenvalues and eigenfunctions in the Karhunen-Loeve expansion for spatial processes. The results can be used to make spatial interpolations given massive spatial data.