A1598
Title: Statistical analysis of extreme geostatistical data: Challenges and advances
Authors: Clara Cordeiro - FCiencias.ID, Associacao para a Investigacao e Desenvolvimento de Ciencias (Portugal) (Portugal)
Manuela Neves - FCiencias.ID, Associacao para a Investigacao e Desenvolvimento de Ciencias (Portugal)
Dora Prata Gomes - NOVA.ID.FCT FCT-UNL (Portugal) [presenting]
Abstract: The quantification and characterization of extreme meteorological events are crucial due to their significant impacts on human life, agricultural productivity, and economic stability. Extreme value theory (EVT) is a theory for modeling and quantifying events that have a very low probability of occurrence but can result in significant impacts. Given that a single extreme weather event can influence multiple locations, the assumption of temporal independence is often unrealistic. While stationary and weakly dependent series share the same limiting distribution as independent ones, their parameters are affected by dependence. Therefore, when analyzing data from multiple locations, it is essential to account for spatial dependence appropriately. Traditional geostatistics, which predominantly relies on Gaussian distribution, is inadequate for modeling tail behavior. The application of max-stable processes is explored, a natural extension of multivariate extremes in a spatial context. An exploratory analysis of annual maximum monthly precipitation data recorded from 1941 to 2023 across 19 locations in northern Portugal is conducted. The results include reviewing dependence measures, estimating parameters of interest, and simulating rainfall prediction maps. The R software will be explored, and packages and functions related to this topic will be used.
