Proposta de otimização de modelo de regionalização de curvas de permanência de vazões

Abstract

Due to lack of data flow in the State of Pará, the regionalization of streamflow duration curves, presents itself as an important technique, allowing the estimation of flow in sites with insufficient or no data. Hus, this paper aims to propose the optimization of a model of regionalization of streamflow retention curves for the Para region. The model was based on data at 43 gauged stations distributed in the state of Pará. The flow duration curves were calibrated using 5 regression models: power, exponential, logarithmic, quadratic and cubic. Regionalization model was established using the multiple regression technique. The spatial variation of each parameter was explained in terms of the drainage area, mean annual precipitation, length and slope of the main river. The model was validated using the Jack -knife procedure. The best fit of the cubic model was mathematically represented by the quadratic mean relative errors (ϵ%), coefficients of Nash-Sutcliffe (Nash) and the graphics settings the simulated and observed streamflow. The optimization model was achieved by insertion of synthetic stations and cluster stations by drainage area. The number of synthetic stations included in the model were evaluated by Root Mean Square Error (RMSE), the coefficient of Nash and the ϵ%. For groups I and II the optimal number of synthetic stations, who joined those already considered in the groups, it was 6 and 3, respectively. In group III only the clustering method in relation to drainage areas was sufficient for good performance of the regionalization model. The good performance of the calibrated, validated and optimized model demonstrated the potential of this in the estimation of retention curves of the rivers that cross the Para. The geographic coordinates of synthetic stations, which optimized the model can serve as a suggestion to the government where to install new stations. The number of new stations would be limited to the results of optimization, streamlining resources and building up the model developed to determine retention curves flow for the entire state of Para. In this case, the new installed stations, would also help in the future to improve the model performance.