Hydrology Evaluation of the Soil and Water Assessment Tool Considering Measurement Uncertainty for a Small Watershed in Southern Brazil

Abstract. Areas under tobacco crop cultivation have impacted the water balance and have become sources of environmental contamination in Southern Brazil. Correct determination of the area’s hydrology is essential since it is the driving force of sediment and nutrient loading dynamics. The Soil and Water Assessment Tool (SWAT), version 2009, was used to evaluate hydrological processes for the Arroio Lino watershed, located in the Brazilian state of Rio Grande do Sul. The measured discharge at the watershed outlet was used to evaluate model streamflow sensitivity to selected parameters, and for calibration and validation from 2001 to 2005. A Latin Hypercube (LH) and One-factor-At-a-Time (OAT) sensitivity analysis was performed. Model calibration was performed with the Shuffled Complex Evolution Algorithm-Uncertainty Analysis (SCE-UA) combined with manual adjustment on model simulation results. Time series plots and standard statistical measures were used to verify model predictions. The statistical indicators were modified to reflect measurement uncertainty. The most sensitive parameters were curve number (CN2), soil evaporation compensation factor (ESCO), and baseflow alpha factor (ALPHA_BF). The predicted monthly streamflow matched the observed values with a Nash-Sutcliffe coefficient of 0.87 and 0.76 during calibration and validation, respectively. Daily indicators of goodness of fit were lower but improved when uncertainty of measured data was taken into account. The results suggest that the SWAT model is a promising tool to evaluate hydrology in Brazilian watersheds. However, especially in those watersheds with limited data availability, the evaluation of model goodness-of-fit needs to consider the uncertainty related to data measurements.