Probabilistic modelling of extreme rainfall events in the Rio Grande do Sul State/Modelagem probabilistica de eventos de precipitacao extrema no estado do Rio Grande do Sul

Flood management in Brazilian watersheds must be discussed and prioritized, however, the current scenario indicates that there are gaps in hydrological information with respect to its spatial and temporal variability. The probabilistic modelling of extreme rainfall events, having as goal to extrapolate values for a given frequency and duration, can be used as an excellent tool for analysis and decision-making. The main objective of this study was to analyse the adjustment of different probabilistic models for series of annual maximum daily rainfall in Rio Grande do Sul. Series of 342 rain gauges were adjusted to 2-parameter Log-Normal, 3-parameter Log-Normal and Gumbel probability distributions and goodness-of-fit tests were based on the Kolmogorov-Smirnov and Chi-Square procedures. It was found that all the distributions presented adequate results, however, 3- parameter Log-Normal distribution had the best performance in accordance with the Chi-Square test. The parameters of probability distribution presented variability over the state and a pronounced relationship with the location of rain gauges. This suggests that regionalization of high-intensity rainfall can be employed in Rio Grande do Sul as an excellent management tool. Key words: high-intensity rainfall probability distribution hydrological regionalization A gestao de cheias em bacias hidrograficas brasileiras deve ser discutida e priorizada, porem o cenario atual indica lacunas quanto as informacoes hidrologicas com variabilidade espacial e temporal condizentes. A modelagem probabilistica de eventos extremos de precipitacao, buscando a extrapolacao para uma frequencia e duracao, pode servir como excelente ferramenta de analise e tomada de decisoes. O objetivo principal deste trabalho foi analisar o ajuste de diferentes modelos probabilisticos a series de precipitacao maxima diaria anual no Rio Grande do Sul. Series pluviometricas de 342 estacoes foram ajustadas as distribuicoes Log-Normal a 2 e 3 parametros e Gumbel e a adequacao foi realizada pelos testes de Kolmogorov-Smirnov e Qui-Quadrado. Todas as distribuicoes de probabilidade consideradas foram adequadas; entretanto, a distribuicao Log-Normal a 3 parametros apresentou os melhores ajustes segundo os resultados do teste Qui-Quadrado. Os parametros das distribuicoes de probabilidades apresentaram variabilidade ao longo do estado e forte relacao com a localizacao, sugerindo que a regionalizacao de chuvas intensas pode ser empregada no Rio Gra