Tamanho ideal de séries históricas de vazões: uma abordagem considerando parâmetros de médias e vazões mínimas de referências

Abstract

The present study seeks to contribute to the knowledge of the hydrological regime, which is very important for making decision about the available water resources management in a hydrographic basin. The use of historical series of hydrological records is one of the basic tools for such management, as it provides informations of water availability, checking trends over the years, in addition to being essential for the dimensioning of hydraulic works and for the management of water resources, among others. However, several regions do not have historical records with enough data to carry out these studies. Therefore, working with series with many years of daily data records is often not possible. Thus, with this study, we aim to define the ideal sample series size for historical series of minimum, maximum and average flows. To do so, 19 fluviometric stations from the Paraná Hydrographic Region (RHP) and another 35 from the São Francisco Hydrographic Region (RHSF) were used, totaling 54 stations, with the criterion of having at least 50 years of daily data record. Series of minimums, maximums and averages annual flows were generated for each of the stations. For the series of minimum, the non-parametric tests of randomness, stationarity, independence and homogeneity were applied in order to verify if the stations are able to be worked with probability distributions. Then, the Kolmogorov-Smirnov adherence test was applied in order to verify the best fitted distribution for each station. The series of minimum flows were analyzed from Q7, Q7,10, Q90% and mean. The maximum and average flow series were analyzed using Q90% and means. For all cases, the complete series (with all years available in the original historical series) and the partial series (randomly constituted with predefined sizes of 10; 20; 30; 40; 50; 60; 70; 80; and 90% of the complete series, in addition to the climatological normal) were analyzed. The results of the partial series were compared with the results of the complete series through two tests, namely: PBIAS; and Student's “t” test, in order to verify the statistical difference between them. For the parameters analyzed with the series of minimum flow rates, series with sizes from eight years on of data were statistically similar to the results obtained with the complete series. Regarding the maximum and average flow rates, the analysis of the average parameter of the partial series was statistically close to the complete series from seven years on of data. For the analysis of Q90%, however, there was great variation in the results, including for the larger series and for climatological normals. In this specific case, it was noticed that the longer the series used, the better.