Spatial interpolation using mixture distributions: A Best Linear Unbiased Predictor
Interpolation spatiale de distributions de mélanges : Un meilleur prédicteur linéaire non biaisé
Abstract
This paper deals with three related problems in a geostatistical context. First, some data are available for given areas of the space rather than for some point locations which creates problems of multiscale areal data. Second, some uncertainties rely both on the input locations and on measured quantities at these locations, involving uncertainty propagation problems. Third, multidimensional outputs can be observed, with sometimes missing data.
These three problems are addressed simultaneously here by considering mixtures of multivariate random fields and by adapting standard Kriging methodology to this context. While the usual Gaussian setting is lost, we show that conditional mean, variance and covariance can be derived from this specific setting. Case studies are presented both with simulated data and real data. In particular, we discuss the question of information loss in learning buildings energy efficiency.
| Origin | Files produced by the author(s) |
|---|
