Search

The traditional approach to characterize extreme rainfall events is based on using past observed data to fit a probability distribution, which is used to associate a certain probability of occurrence to a rainfall magnitude, under a stationary assumption. Observed series, however, may exhibit trends and abrupt changes, resulting from the natural hydro climatic variability and the climate change context, which are not usually considered in the stationary modelling procedure and which may affect the characterization of extreme events. In this paper, it was evaluated the presence of these temporal changes in rainfall series and their impact to estimate design quantiles, using rainfall data of 317 stations, located in São Paulo State, Brazil. The results showed that extreme rainfall events are changing in the state. Low-return period events (up to 10 years) are intensifying. Moderate- and high-return period events, on the other hand, did not present a clear spatial pattern. Trends and abrupt changes may significantly influence the estimation of design quantiles. The presence of the former in rainfall series exhibited more influence on low-return period events, showing, however, no clear relationship with high-return period design quantiles. The presence of the latter seems to influence the characterization process, affecting estimated quantiles of all evaluated return periods. Such findings highlight the need to consider the presence of trends and abrupt changes and their influence on design quantiles for a more reliable characterization. [ABSTRACT FROM AUTHOR]

The popular approach to select a suitable distribution to characterize extreme rainfall events relies on the assessment of its descriptive performance. This study examines an alternative approach to this task that evaluates, in addition to the descriptive performance of the models, their performance in estimating out-of-sample events (predictive performance). With a numerical experiment and a study case in São Paulo state, Brazil, we evaluated the adequacy of seven probability distributions widely used in hydrological analysis to characterize extreme events in the region and compared the selection process of both popular and altenative frameworks. The results indicate that (1) the popular approach is not capable of selecting distributions with good predictive performance and (2) combining different predictive and descriptive tests can improve the reliability of extreme event prediction. The proposed framework allowed the assessment of model suitability from a regional perspective, identifying the Generalized Extreme Value (GEV) distribution as the most adequate to characterize extreme rainfall events in the region. [ABSTRACT FROM AUTHOR]