2017/01/21 by Alfredo Alegría, Alegría, Alfredo, Emilio Porcu +5
Environmental Science · Mathematics · #FOS: Mathematics #Morphological variations and asymmetry #Remote Sensing in Agriculture #Soil Geostatistics and Mapping #Statistics Theory (math.ST)
paper · pdf · doi:10.48550/arxiv.1701.06010
openalex publication_date 2017/01/21 · openalex created_date 2022/10/02 · openalex updated_date 2026/07/28
The construction of valid and flexible cross-covariance functions is a\nfundamental task for modeling multivariate space-time data arising from\nclimatological and oceanographical phenomena. Indeed, a suitable specification\nof the covariance structure allows to capture both the space-time dependencies\nbetween the observations and the development of accurate predictions. For data\nobserved over large portions of planet Earth it is necessary to take into\naccount the curvature of the planet. Hence the need for random field models\ndefined over spheres across time. In particular, the associated covariance\nfunction should depend on the geodesic distance, which is the most natural\nmetric over the spherical surface. In this work, we propose a flexible\nparametric family of matrix-valued covariance functions, with both marginal and\ncross structure being of the Gneiting type. We additionally introduce a\ndifferent multivariate Gneiting model based on the adaptation of the latent\ndimension approach to the spherical context. Finally, we assess the performance\nof our models through the study of a bivariate space-time data set of surface\nair temperatures and precipitations.\n