Análise estatística de extremos de precipitação via Distribuição Generalizada de Pareto (GPD). Estudo de casos: Natal, Fortaleza e Teresina

Abstract

Global climate change, evidenced by rising temperatures, irregular rainfall patterns and more frequent extreme weather events, is causing significant impacts on people's lives and on various economic sectors. Occurrences of extreme precipitation have the potential to drastically change the typical characteristics of a given region, resulting in serious challenges for the population. The extreme value theory approach focuses explicitly on extreme outcomes or values, offering a range of natural models to deal with these occurrences. Extreme value theory focuses on specific types of probability distributions such as the GPD (Generalized Pareto Distribution). This paper presents an analysis of climatic extremes of precipitation using the GPD, with daily data from Natal, Fortaleza and Teresina. The POT (Peaks Over Threshold) method adjusts the GPD to the values that exceed a certain threshold. It then calculated the return levels for 2, 5, 10, 15, 20, 30, 50, 70 and 100 years. The study reveals that Fortaleza has the highest number of extreme events and the highest accumulations of rain per day, closely followed by Natal, which is justified by the proximity of both to the coast and the influence of meteorological systems such as the ITCZ. In contrast, Teresina, which is more continental, has lower rainfall rates and a lower frequency of extreme events. What predominates in the expected levels over different return periods, highlighting the impact of time on the return and the uncertainty associated with the estimates. Understanding return levels helps to prepare for extreme precipitation events, build appropriate infrastructure and mitigate flood risks.