Nonparametric estimation of isotropic covariance function

A nonparametric model using a sequence of Bernstein polynomials is constructed to approximate arbitrary isotropic covariance functions valid in R-alpha and related approximation properties are investigated using the popular L-alpha norm and L-2 norms. A computationally efficient sieve maximum likelihood (sML) estimation is then developed to nonparametrically estimate the unknown isotropic covariance function valid in R-alpha. Consistency of the proposed sieve ML estimator is established under increasing domain regime. The proposed methodology is compared numerically with couple of existing nonparametric as well as with commonly used parametric methods. Numerical results based on simulated data show that our approach outperforms the parametric methods in reducing bias due to model misspecification and also the nonparametric methods in terms of having significantly lower values of expected L-alpha and L-2 norms. Application to precipitation data is illustrated to showcase a real case study. Additional technical details and numerical illustrations are also made available.

Automated summary

This paper proposes a nonparametric model using Bernstein polynomials to approximate arbitrary isotropic covariance functions. The popular L-alpha and L-2 norms are used to investigate the approximation properties. A computationally efficient sieve maximum likelihood (sML) estimation method is developed to estimate the unknown isotropic covariance function. Numerical results show that the proposed approach outperforms both parametric and nonparametric methods in terms of reducing bias and having lower norms.