Applying the SAC-SMA hydrologic model to a tropical watershed in Brazil

Abstract. Brazil has an extensive runoff measurement network, but they have issues related to the extension of historical series, heterogeneous distribution, systematic and gross errors in the records, as well as to lack of sporadic measurements. A hydrologic conceptual rainfall-runoff model, duly calibrated and validated, may be used in filling runoff failures, in the analysis of the observed levels consistency and even in the extension of historical series, since they require rainfall as input data, since it typically has longer series. The aim of the current study is to evaluate the performance of the Sacramento Soil Moisture Accounting model (SAC-SMA) in the estimation of the daily runoff of a Brazilian river basin, so that it can be incorporated as a water resources management tool. The study was conducted in three runoff-monitoring sections of Piracicaba River basin, which has an area of 5,304.0 km² and is located in Minas Gerais State-Brazil. The rainfall and evapotranspiration reference data required as input data in the SAC-SMA model were obtained by Thiessen method and by Hargreaves-Samani method, respectively. The model calibration was performed through the Shuffled Complex Evolution algorithm using the Nash-Sutcliffe efficiency index (E) as objective function. The following statistical measurements were used to check the Sacramento model performance: mean absolute error; root mean square error; bias; paired t-test at 5% significance; Willmott's concordance index; and Nash-Sutcliffe efficiency index. By analyzing these measurements and the hydrographs, it was possible to see concordance between the runoff values estimated by the Sacramento model and those observed in the measurement sections. However, the model showed greater difficulty to simulate peak values. The Nash-Sutcliffe efficiency index values obtained in the three sections were 0.87, 0.78 and 0.71, which confirms that the model is “fit and proper" to estimate the daily runoffs. The difficulty to determine the representative rainfall of the hydrographic basin is among the possible causes for low accuracy in estimating peak runoffs, due to the basin‘s high spatial and temporal variability. The SAC-SMA model is suitable to estimate the daily flow of water courses in Piracicaba River basin and it has potential to be used in studies related to hydrological simulation and to the management of water resources in tropical basins.