Calibração e validação de um modelo hidrológico para a bacia hidrográfica do rio Ivaí.

Abstract

Numerical models are widely used to study important issues related to the hydrological cycle, especially those related to changes in land use and occupation and to climate change. Hydrological models in particular have proven to be useful tools for policy-making by managers and decision makers. Thus, the proper representation of the physical processes by a model is extremely important for the quality of the response-variables of interest in a given basin. In this sense, this work calibrated and validated the SWAT (Soil Water Assessment Tool) hydrological model for the Ivai River Basin (IRB). For the calibration and validation, five fluviometric monitoring stations were used, distributed along the central course of the river. Based on the indicators proposed in the literature, for the performance evaluation of hydrological models, the results obtained for the calibration indicate that the SWAT model adequately simulated monthly flows. Based on the Nash-Sutcliffe efficiency index (NS), four of the fluviometric stations considered were rated Very Good (NS between 0.86 and 0.89) and only one with Good index (0.70). The applicability of the model to the IRB was confirmed through the simulated responses during the validation step, based on a period different from that used in the calibration. In this case, the five stations were evaluated as Very Good (NS between 0.81 and 0.86). The calibrated model was also used in the evaluation of flow responses under different arrangements of rainfall stations. In this case, we considered the calibrated and validated simulation with the largest number of stations as the control scenario. This procedure allowed to investigate the impact of the density of stations and their spatial distributions in the IRB and to evaluate the minimum density of stations to support hydrological simulations with satisfactory quality. Considering a relative mean square error (RRMSE) of 0.05 as acceptable in the literature, the results showed that the model is able to satisfactorily represent the IRB flow with minimum arrangements between 11 and 20 rainfall stations (one rainfall gauge controls 923.11 km² and 744.44 km², respectively). The difference in the minimum number of stations is associated to the great spatial variability of the physical and climatic characteristics of the IRB, which entails different values of RRMSE due to the spatial arrangement of the stations. Although improvements are still possible, the quality of the simulated flows shows that the hydrological model using SWAT is perfectly suited for immediate use by managers and policy makers for the water resources area within the IRB.