Modeling Vector Random Fields on a Sphere: Application on Geomagnetic Field (306547)*Amy Kim, University of California, Davis
Thomas Lee, University of California, Davis
Debashis Paul, University of California, Davis
Keywords: multivariate analysis, spatial temporal; geomagnetic model, vector random field
Multivariate processes on a sphere have been investigated with the dual goal of prediction and modeling fluctuations. The random variations in the process are typically modeled through specifying the cross-covariance functions. We propose a framework for modeling vector random fields on a sphere through vector spherical harmonics. This approach enables incorporation of natural physical constraints on the process, and has the added advantage of providing a unified framework for modeling both mean and residual fields. We illustrate the proposed method with simulations and perform modeling geomagnetic field of Orsted Satellite data.