2019
DOI: 10.1214/17-aos1652
|View full text |Cite
|
Sign up to set email alerts
|

Estimation and prediction using generalized Wendland covariance functions under fixed domain asymptotics

Abstract: We study estimation and prediction of Gaussian random fields with covariance models belonging to the generalized Wendland (GW) class, under fixed domain asymptotics. As for the Matérn case, this class allows for a continuous parameterization of smoothness of the underlying Gaussian random field, being additionally compactly supported. The paper is divided into three parts: first, we characterize the equivalence of two Gaussian measures with GW covariance function, and we provide sufficient conditions for the e… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

8
96
0

Year Published

2019
2019
2024
2024

Publication Types

Select...
8

Relationship

2
6

Authors

Journals

citations
Cited by 81 publications
(104 citation statements)
references
References 40 publications
8
96
0
Order By: Relevance
“…In each case, we estimated the parameters of the covariance function using ML and used them to compute the SCCC given in equation (6). Three set of parameters were considered, one set each for σ 1 = σ 2 = 1, ν = 4 and k = 1: The ML estimates of the parameters of the Wendland-Gneiting covariance function had low bias and standard errors, and agreed with previously published results (e.g., Bevilaqua et al, 2019). Using these estimates, we computed the SCCC in each case for 0 < ||h|| < 2.…”
Section: Monte Carlo Simulationssupporting
confidence: 70%
“…In each case, we estimated the parameters of the covariance function using ML and used them to compute the SCCC given in equation (6). Three set of parameters were considered, one set each for σ 1 = σ 2 = 1, ν = 4 and k = 1: The ML estimates of the parameters of the Wendland-Gneiting covariance function had low bias and standard errors, and agreed with previously published results (e.g., Bevilaqua et al, 2019). Using these estimates, we computed the SCCC in each case for 0 < ||h|| < 2.…”
Section: Monte Carlo Simulationssupporting
confidence: 70%
“…In this section, we let d = 1, 2 or 3 and extend the results for the Matérn covariance functions of Section 4 to the isotropic Wendland family of covariance functions on [0, 1] d (Gneiting, 2002;Bevilacqua et al, 2019). Here k θ = k θ,s,µ , with θ = (σ 2 , ρ), is given by…”
Section: Microergodic Parameter Estimation For the Isotropic Wendlandmentioning
confidence: 97%
“…The Wendland covariance function is given by k θ,s,µ (u, v) = k θ,s,µ (u − v). The parameter s drives the smoothness of the Wendland covariance function, similarly as for the Matérn covariance function (Bevilacqua et al, 2019). The parameters σ 2 > 0 and ρ > 0 are interpreted similarly as for the Matérn covariance functions and are to be estimated.…”
Section: Microergodic Parameter Estimation For the Isotropic Wendlandmentioning
confidence: 99%
“…Low-rank methods project the original process onto representative points or knots (see, e.g., Higdon, 2002;Banerjee, Gelfand, Finley, & Sang, 2008;Cressie & Johannesson, 2008;Stein, 2008); however, these approaches often perform poorly for prediction as they often over-smooth (see Stein, 2014). Alternatively, sparse methods induce zeros either in the covariance matrix using compactly supported covariance functions (see, e.g., Furrer, Genton, & Nychka, 2006;Kaufman, Schervish, & Nychka, 2008;Bevilacqua, Faouzi, Furrer, & Porcu, 2019) or in the precision matrix by assuming conditional independence (Vecchia, 1988;Stein et al, 2004). We ultimately favor approaches that assume conditional independence because predictive performance is generally better (Heaton et al, 2018) and the class of valid covariance models is more expansive.…”
Section: Nngp Modelmentioning
confidence: 99%