Frequência de cheias no Brasil: análise local

Abstract

Floods are among the hydrological events that have great potential to cause environmental, economic and social damage in the world. Therefore, studies on the characteristics and frequency of these events are of paramount importance. Appropriate estimates of extreme hydrological events help mitigate the impacts of natural disasters and the knowledge of these events is a fundamental requirement for hydraulic structures projects. At-site flood frequency analysis is considered the most direct method for probabilistic modeling of these events and an important technique for estimating the magnitude and frequency associated with flood events. Few studies have focused on Probability Density Functions (PDFs) with more than three parameters due to the complexity of these functions, however, recent studies have demonstrated that multiparameter PDFs provide better results than those PDFs traditionally used in Brazil. It should be also noted that, even after an extensive literature review, it was not possible to find studies on this topic covering the entire Brazilian territory. Therefore, this study had as main objective to propose and evaluate a robust methodological framework for at-site flood frequency analysis that can be applied at the level of a continental country such as Brazil, allowing to indicate the most suitable PDFs to represent the maximum annual streamflow in Brazilian rivers. The methodology of this study consisted of the following steps: acquisition of streamflow data for the entire Brazilian territory and organization of the maximum annual streamflow series (MAS); screening of MAS series; fit of eleven PDFs (Normal, Gamma, Gumbel, GEV, GLO, GNO, GPA, LN3, PE3, Kappa and Wakeby) to the MAS series; and PDF’s goodness-of-fit and quantiles derived from the PDFs. As evidenced by this study, the more robust PDFs with a greater number of parameters outperformed those simpler PDFs (with 2 and 3 parameters). The 5-parameter Wakeby function provided the most accurate results, such that it was the most appropriated function for 345 MAS series in Brazil, followed by the Kappa function that had the best fit for 149 series. Therefore, it can be concluded that the Wakeby is the most suitable function for probabilistic modeling of maximum annual streamflow series in Brazil. The importance and scope of this study should be also highlighted, as it is the first to be carried out at this scale in the country. In this context, this study can serve as a basis for the probabilistic modeling of flood events throughout the country, thus supporting researchers and designers in their decision making.