Avaliação dos processos hidrológicos em uma bacia do semiárido com o modelo SWAT calibrado com evapotranspiração por satélite.

Abstract

The calibration of hydrological models is a lengthy process, and sometimes a challenge for a significant portion of the Brazilian watersheds, given the fragile spatial density and temporal discontinuity of hydrometeorological measurements, the uncertainties in the representation of physical characteristics, and human interventions such as, for example, dams and/or reservoirs. The objective of this study was to analyze the use of evapotranspiration estimated by remote sensing in the calibration of the distributed hydrological model, SWAT, for a watershed in the Brazilian semiarid region, in order to evaluate its hydrological processes. The model used was the Soil Water Assessment Tool (SWAT) that incorporates three methods, which were used in this study, to calculate evapotranspiration: Penman-Monteith, Priestley Taylor, and Hargreaves. The coefficients of the functions ME (Mean Error), RMSE (Root Mean-Square Error), KGE (Kling-Gupta efficiency), NSE (Nash-Sutcliffe Efficiency), the R² (determination coefficient) and the br2 coefficient were obtained. The results indicated that the Priestley-Taylor method was the method that presented the best performance in the ET simulation, while the Hargreaves method presented the worst. As for the simulated discharge, the model showed more favorable results in the simulation of minimum flows in relation to maximum flows. The Hargreaves method showed the greatest performance in the discharge simulation and the Priestley-Taylor showed the worst. After calibration, a study of the water balance components was carried out through the trend analysis estimated with the Theil-Sen method and it was found that the drought faced from 2012 to 2018 brought significant reductions to the water balance components in the basin of the river Piancó. Precipitation showed a reduction of 2.3%, evapotranspiration (1.4%) and flow showed a reduction of 4.7%. This study showed the potential of using hydrological models calibrated with data from remote sensing, and in particular, evapotranspiration, to estimate discharge in places with few in loco data. Furthermore, the evaluation of different methods for estimating potential evapotranspiration showed that the use of the average of the simulations (ensemble) can be more advantageous than the use of a single method.