María del Carmen Bueso Sánchez , José Miguel Angulo Ibáñez , Francisco J. Alonso, María Dolores Ruiz Medina
Risk assessment in environmental applications involving spatiotemporal pro- cesses requires, in most cases, a dynamic evaluation and detection of the even- tually critical behavior in relation to the spatial variability. In this paper, the problem of designing adaptive sampling strategies for prediction in a spatiotem- poral state-space model framework is considered. Speci cally, at each time, the region of interest for the unobservable process is re-de ned based on the updated historical observations, which provide information on relative local uncertainty.
A dynamic optimal selection of spatial sampling con gurations is then proposed based on entropy criteria. Performance is illustrated with an empirical study.
© 2008-2024 Fundación Dialnet · Todos los derechos reservados