| dc.description.abstract | The estimation of unit hydrographs (UH) for flood flow studies is a current practice in engineering hydrology. Conceptual UH models, such as the instantaneous unit hydrograph Nash (IUH-Nash), as well as the conventional Least-Squares UH are based on historical rainfall-runoff data and are usually applied in gauged basins. For ungauged or poorly gauged basins, the concept of geomorphologic instantaneous unit hydrograph (GIUH) can be used to relate runoff with the catchment geomorphological characteristics. For the latter, digital terrain models are used, as associated with automatic extraction techniques of geomorphological features for obtaining the required parameters. The method presented in this MSc dissertation consists basically of two phases. The first relates a comparison of the results of applying these three models to twenty basins with drainage area between 500 and 5,000 km2, as defined by existing streamflow gauging stations in the upper basin of the São Francisco River, located in the Brazilian state of Minas Gerais. For that purpose, 287 pairs of rainfall-runoff events have been collected and examined, from which their respective direct runoff hydrographs and effective rainfall hietographs were derived, thus resulting in the median HU of each of the basins. In the sequence, representative events have been selected in order to compare the flood volumes associated with the estimated and observed hydrographs, in order to evaluate the models applicability and robustness for the Brazilian watersheds. In the second phase there has been proposed a method for regionalizing the HU parameters, with the purpose of transferring them to ungauged basins. The proposed methodology was verified by the jack-knife cross validation technique in order to assess the reliability of the obtained regional models. The results showed a satisfactory efficiency in the application of these models for estimating flood hydrographs as compared to the observed ones. The proposed regional models can be considered robust, as it was possible to define homogeneous regions through regional equations using multiple linear regression, which link the catchment geomorphological characteristics to the model parameters. | |
