Metodologias na derivação das relações de intensidade-duração-frequência de chuvas intensas

Abstract

Intense rainfall events are extreme occurrences characterized by high amounts of water precipitated in a short period of time, resulting in floods and major disasters. Therefore, studying intense rainfall is important for planning preventive and mitigative measures against the impacts such events can cause, as well as aiding in the execution of safer and more cost effective infrastructure works and hydraulic projects. In this context, Intensity-Duration Frequency (IDF) relationships are employed to mathematically describe the relationship between rainfall intensity and duration for a given return period. While various methodologies exist in the literature for deriving IDF relationships, there is a shortage of comparative studies among such approaches. In Brazil, studies related to obtaining IDF relationships are still limited, with predominant usage of the rainfall disaggregation method. Hence, the objective of this study was to apply and compare the methodologies of standard procedure (PP), single step procedure (PEU), robust procedure (PR), simple generalized procedure (PGS), intensity transformation (TI), and third-order polynomial normal transformation (TNPT) in deriving IDF relationships using rainfall data from a pluviographic station located in the city of Porto Alegre, Rio Grande do Sul. The data are available in the Brazilian Pluviometric Atlas of the Mineral Resources Research Company and cover a 40-year period. This research considered 12 durations and 13 return periods, and the tested methodologies were quantitatively evaluated through error estimation, including the calculation of root mean square error (RMSE), mean absolute error (MAE), and mean error (ME). The main results indicate that all tested methodologies generated IDF curves that fit better for short durations than for medium to long durations; the number of steps each methodology employs to derive IDF relationships, in addition to the type of probability distribution and parameter estimation method used in the analysis, impact the results; the sample size of the dataset influences the quality of the estimates obtained. Among the tested methodologies, the single-step procedure demonstrated the best performance in terms of root mean squared error (RMSE).