Análise do desempenho do modelo SWMM5 acoplado ao calibrador Pest na Bacia do Arroio Cancela/RS

Abstract

This dissertation presents the results of the qualitative and quantitative modeling of the Arroio Cancela urban basin, having an area of 4.35 km², with the use of the Storm Water Management Model (SWMM5). The generation and routing of the runoff, base flows and the processes of accumulation and washoff of the total suspended sediments (TSS) and organic matter represented by biochemical oxygen demand (BOD5) were modeled. The package PEST (Parameter Estimator) was used in the calibrations of the most sensitive parameters of the SWMM5. Eight months of monitored rainfall and runoff data containing 34 rainfall events were calibrated. Calibration of the qualitative processes used 10 events containing monitored data with concentrations of TSS and BOD5. The validation of the average calibrated parameters was carried out over a period of three months, having 16 rainfall events, 4 of which containing monitored data of TSS and BOD5. The results indicate that the SWMM5 is more sensitive to parameters related with the impermeable areas of the basin. The parameters of the permeable areas were more sensitive in events of greater magnitude. The use of PEST proved to be valuable in optimizing the model, considering the speed at which the algorithm converges to a satisfactory solution. The runoff calibration of the events reached very good Nash-Sutcliffe efficiencies (ENS) (average of 0.92). For the continuous simulations, calculated ENS reached a value of 0.72. The average errors in the flow volume for both cases were less than 14%. The calibration of TSS reached an average ENS value of 0.56 and, for BOD5, an average ENS of -0.75, with a high dispersion of the washoff parameters for both pollutants. The runoff validation for events produced an average ENS equal to 0.47, calculated median equal to 0.87 and hydrograms with good reproduction of the shape of the observed data. The validation of the continuous series presented ENS equal to 0.74 and an underestimation of the flow volume equal to 7.7%. The validation of the quality processes resulted very poor ENS indexes, with deficiently representation of the variation of TSS and BOD5 concentrations. The results indicate that the use of the SWMM5 model coupled with the PEST calibrator can produce good results in the prediction of runoff events and continuous flow series. However, the representation of the qualitative processes require better initial parameter estimations for buildup and washoff, in addition to improvements in the calculation algorithm of buildup and washoff of pollutants.