Determinação de coeficientes de desagregação e relações IDF para o estado do Rio Grande do Sul por meio de séries de duração parcial

Abstract

Knowing the magnitude of heavy rainfalls is of extreme importance for the elaboration of several hydraulic projects, for the management of water resources and for the prevention of natural disasters in engineering. The most usual form of estimation of heavy rainfall is the use of an intense rainfall equation, which must be generated for sites provided with historical series obtained from pluviographs, but in many places these data are scarce, making it impossible to determine the relationships between intensity, duration and frequency of heavy rainfall. Due to this difficulty, some alternatives using pluviometer data that can be safely used in the hydraulic projects have been used, and the disaggregation rainfall method, through rainfall relationships of different durations, is the most used in Brazil, making possible the estimation of rainfall from the maximum daily precipitation per year through the use of disaggregation coefficients. The most used coefficients in Brazil are those proposed by CETESB (1986) obtained for a national average, and for this reason they may not be representing in the best way the intense rains in the state of Rio Grande do Sul. For this state, there is also a study by Beltrame et al. (1991), which presents disaggregation coefficients for several locations, however, it is an old work with rainfall data from more than 30 years ago. In this way, this study had as main objective to generate new coefficients of disaggregation of daily rain for the state of Rio Grande do Sul, through the establishment of an easy-to-apply methodological script for defining the partial series to be used and for selecting the probability density function that best suits partial series. For the application, the most recent rainfall data from 9 cities were considered by means of partial duration series, evaluating the density functions of probability of Gumbel, PoissonPareto and Poisson-Exponencial, in addition to 4 different rates of exceedance, guaranteeing greater precision in the estimates of intense rains and thus contributing to the reduction of uncertainties in hydraulic projects in the state. Through the Kolmogorov-Smirnov adherence test performed at 20% significance, it was found that the Poisson-Pareto theoretical function was the one that best fitted the empirical curves and that very high exceedance rates make it difficult to adjust the functions of probability. The results of the disaggregation coefficients showed that there are differences that reach 9.2% in the 10min/30min relation, when compared only with the averages of the other studies, reaching differences of more than 28% when comparing the results separately for each locality with the national average, confirming that local data should be preferred over average data. The relation for the precipitation of 24h/1 day in the state of Rio Grande do Sul varied from 1.01 to 1.18 with an average of 1.09, showing a difference of 4.6% from the national average presented by CETESB and widely used, which is 1.14. This study also obtained IDF equations for each analyzed location with determination coefficients above 0.988 and mean absolute percentage error of 3.22%.