Detalhes bibliográficos
| Resumo: | The current research aims to characterize the multivariate spatial dependency by expressing the cross-covariance functions in terms of the direct covariances, under second-order stationarity. This approach reduces the dimensionality of the estimation problem and enables a separate specification of the different components, thus avoiding a joint estimation of all the covariance functions. We derive bounds for the approximations errors and establish sufficient conditions to ensure the positive-definiteness of the resulting multivariate covariance structure. Theoretical properties of the proposed approximations are studied within a multivariate stochastic process framework. Implications for spatial inference and kriging-based prediction are discussed. |
|---|
| Autores principais: | García-Soidán, Pilar |
|---|
| Outros Autores: | Menezes, Raquel; Tinoco, Daniel; Pilar, García-Soidán |
|---|
| Assunto: | covariance function covariate multivariate process positive-definiteness stationarity |
|---|
| Ano: | 2026 |
|---|
| País: | Portugal |
|---|
| Tipo de documento: | artigo |
|---|
| Tipo de acesso: | unknown |
|---|
| Instituição associada: | Instituto Nacional de Estatística |
|---|
| Idioma: | inglês |
|---|
| Origem: | REVSTAT-Statistical Journal |
|---|