J. L. Romero Béjar, J. M. Angulo Ibáñez
Risk assessment of real phenomena modeled by spatial or spatiotemporal random fields is an area of increasing research interest. Evaluation of different classes of excursion probabilities, in relation to structural properties of excursion sets, is one of the most significant topics in this context. There are key references providing error bound approximations for excursion probabilities based on the Euler-Poincaré characteristic of excursion sets. Gaussianity and/or stationarity assumptions, jointly with suitable regularity conditions, are considered. Spatial deformation and local smoothing by means of kernel–based blurring transformations in terms of convolution operators provide significant classes of non-stationary random fields. In this work, extensions of classical asymptotic error bound approximation results are provided in such more general scenarios.
Work supported by MINECO/FEDER grant MTM2015-70840-P.
Palabras clave / Keywords: blurring, deformation, Euler-Poincaré characteristic, non-stationary random fields, threshold exceedances
Programado
Sesión V08 Estadística Espacial y Espacio-Temporal II
1 de junio de 2018 17:20
Sala 1
