A0802
Title: Tail estimation for the cross-spectral density of a bivariate stationary Gaussian random field
Authors: Joonho Shin - Department of Statistics, Seoul National University (Korea, South) [presenting]
Abstract: Multivariate stationary Gaussian random fields are widely used to fit multivariate spatial data. The one to one correspondence between (cross-)covariance functions and (cross-)spectral densities allows us to model (cross-)spectral densities instead of (cross-)covariance functions. We consider bivariate stationary Gaussian random field model. Under some assumptions on high-frequency behavior of (cross-)spectral densities, we introduce approach to estimate parameters that control tail behaviors by minimizing a modified version of multivariate Whittle likelihood objective function. We show consistency and asymptotic normality of the estimators with simulation results.
