Estudo de previsão de cheias em tempo real com modelos empíricos e semi-conceitual

Abstract

Mathematical models are important parts in flood-forecasting systems. Usually empirical and semi-conceptual models are used independently. In order to minimize the errors induced by the independent use of the models, this dissertation suggests the simultaneous use of the two kinds of models. Three models were tested: a semi-conceptual rainfall run-off model, and ARIMAX empirical model and a conjugation: a rainfall run-off model for streamflow forecasting and and ARIMAX for error forecasting of the semi-conceptual model. The models were tested with date from Ibirama basin, in the Itajaí Valley, in the State of Santa Catarina. The results showed that the effectiveness of the models depend on the forecasting horizon. For the very short run, the ARIMAX model can be used. For the long run, forecastings like the estimation of the peak time and its value, the semi-conceptual model was superior. For short run, the conjugated model seens to be the best.