Analysis of precipitation data in Rio de Janeiro city using Extreme Value Theory

Every year the city of Rio de Janeiro is hit by heavy rains that cause several disasters. To be able to reduce potential damages that may arise due to extreme rainfall, it is important to make accurate estimates of such events. In this paper, we present preliminary work on using Extreme Value Theory (EVT) to model rainfall data in Rio de Janeiro municipality. We use twenty-five years of historical data coming from rain gauges spread throughout the city. We investigate the behavior of both approaches in EVT to identify extreme values, namely, Block Maxima and Peeks-over-Threshold. After determining the best sampling distributions given the available data, we present an analysis of their goodness of fit and of their corresponding return period plots.