Regionalização de chuva intensa para o estado de Minas Gerais

Abstract

The study of intense rains is important to know the spatial and temporal characteristics, besides the essential standard in the steps of designing large works and drainage projects. The greatest problems in hydrology are the availability of historical series of long pluviograms or absence of local information. The alternatives to dribbling are those that are regionalized to rain. In view of the above, the present work was developed in order to regionalize the intensities of the maximum time limit for the state of Minas Gerais, under the homogeneous forms with the indices of "flood-index" to fill the gaps in the local information. For this purpose, 494 rainfall stations associated with the network of the National Water Agency were used. The Gumbel distribution was observed to the precipitations observed, whose parameters were adjusted by the Moment (MOM), Maximum Likelihood (MVS) and Moments-L (MML) methods. The adhesion tests used to verify the fit of the distribution to the historical series were Chi-Square, Anderson-Darling and Filliben. The MVS and MML methods were the ones that resulted in better adjustments. It was verified by r² that the adjustment of the IDF relations was shown to be quite satisfactory, since the values were close to 1.0. In a second stage, the state of Minas Gerais was separated into hydrologically homogeneous regions to obtain their respective "Index-Flood" according to the times of return and duration in order to contribute to the estimation of the project rains. The separation of the homogeneous regions was done by the cluster analysis with k-medium groupings, generating a total of three groups in the state. For each group the regional IDF relationships were generated, which related to the values of precipitation intensity for various duration times resulted in the "Index-Flood". Subsequently, the regional equations of dependent variable, associated to the "Index-Flood" and the local precipitation value, were obtained to obtain the local design rain used for the hydraulic design. The adjustments of these equations were evaluated by the coefficient of determination, and for all equations the r² values were close to 1.0, indicating a good fit. With this it is possible to conclude that the use of the local IDF and Index Flood ratios is satisfactory and contributes to the estimation of the project rains to locations without rainfall stations.