Smooth Copula-based Generalized Extreme Value model and Spatial Interpolation for Sparse Extreme Rainfall in Central Eastern Canada

This paper proposes a smooth copula-based Generalized Extreme Value (GEV) model to map and predict extreme rainfall in central eastern Canada. Furthermore, we provide a comparison with different classical interpolation-based approaches. The considered data represents a station network particularly spatially sparse. Furthermore, one observes several missing values and non-concomitant record periods at different stations. We compare the classical GEV parameter interpolation approaches with our smooth GEV modeling approach, in which the parameters are modeled as smooth functions in space through the use of spatial covariates and by using copula-clustering techniques recently introduced in the literature.