Title: Censored local likelihood inference for modeling non-stationarity in spatial extremes
Authors: Daniela Castro-Camilo - King Abdullah University of Science and Technology (Saudi Arabia) [presenting]
Raphael Huser - King Abdullah University of Science and Technology (Saudi Arabia)
Abstract: In order to model complex dependence structures in spatial extremes, we propose an approach based on factor copula models. The latter, which can be seen as Gaussian location mixture processes, assume the presence of a common factor affecting the joint dependence of all measurements. When the common factor is exponentially distributed, the resulting copula is asymptotically equivalent to the Husler-Reiss copula; therefore, the so-called exponential factor model is suitable to capture tail dependence. Under the assumption of local stationarity, the exponential factor model is used to model non-stationary extreme measurements over high thresholds. Inference is performed using a censored local likelihood. Performance is assessed using simulation experiments, and illustrated using a daily rainfall dataset.