Relação intensidade-duração-frequência de chuvas extremas na região Nordeste do Brasil

Abstract

The lack of data regarding rainfall-runoff in Brazil, of which territorial extension hinders conventional hydrological measuring programs, led to the development and of mathematic models that use intensity-duration-frequency (IDF) relations. The IDF relations allow the determination of project rainfall in location, which dispose of no flow data, causing the dimensioning of hydrologic and agricultural constructions to be reliable. To define the IDF relations, data obtained from pluviographs, capable of measuring rain intensity, duration and height are used. However, such equipment are scarce in Brazil. On the other hand, national pluviometric network, which provides only rain height, is ample and well distributed, leading to the development of methodologies that allow the estimation of the IDF relation by means of rainfall disaggregation techniques. In this context, we aimed at generating IDF relations for pluviometric stations located in the northeastern region of Brazil, by means of a 24 hours rainfall disaggregation technique, developed by CETESB, and, by adjusting the IDF relations, space and interpolate its parameters by use of Kriging. For such, we chose stations that presented historical series with a minimum of 15 years of daily observations, totalizing 2,042 stations with pluviometric records extracted from the Hidroweb, which were disaggregated, generating durations of 5, 10, 15, 20, 25, 30, 60, 360, 480, 600, 720 and 1440 minutes. The Gumbel probability model was employed in the frequency distribution analysis and in the estimation of intense rainfall for the return periods of 5, 10, 25, 50 and 100 years. For verifying the adherence of the frequencies estimated by the Gumbel model to the observed frequencies, using the Chi-square test at a level of 5% of significance. Subsequently, the IDF relations were adjusted and the performance of the IDF models was evaluated by the coefficient of determination (r 2 ). For all employed stations, we verified an optimum adherence of the theoretical frequencies obtained by the Gumbel model to the observed frequencies using the Chi-square test and the performance coefficient, which presented values above 0.85. The adjustments of the IDF relations for all evaluated stations presented values for r 2 close to 1.0, allowing a perfect adjustment. For analyzing the error generated by cross validation, we verified that the semivariogram model that presented the best results was the Gaussian model, allowing for thematic maps of the IDF relation parameters using kriging, thus the obtaining of these relations for locations lacking monitoring.