The Implications of the Use of ET Remote Sensing Data for Calibrating Hydrological Models: A Comparison of Single and Multi-criteria Calibration in SWAT + 

The availability of remotely sensed data has opened new possibilities for research and development of hydrological models. Satellite-derived evapotranspiration (ET) data are useful for improving calibration in ungauged catchments and for performing multi-criteria calibrations in gauged catchments. In this study, SWAT + has been used to analyze different calibration strategies in the watershed of Lake Erken Sweden. Two sensitivity analyses were carried out to detect flow-sensitive and ET-sensitive parameters. A calibration has been performed with observed discharge data (Q scenario), with ET data obtained from the Amsterdam Global Terrestrial Evaporation Model (ET scenario), and a multi-criteria calibration using both data (Q-ET scenario). The calibrations were performed using the SWAT Toolbox software. Results show that all three simulations are statistically acceptable for flow and ET (NSE = 0.773 for Q and NSE = 0.765 for ET in the Q scenario; NSE = 0.712 for Q and NSE = 0.779 for ET in the ET scenario; NSE = 0.764 for Q and NSE = 0.771 for ET in the Q-ET scenario). In the single calibrations, we observed an improved result for the calibrated variable but a reduced fit for the other. The combined calibration approach, however, produces intermediate values for both streamflow and ET, demonstrating its utility. Notably, calibrating with ET alone exhibits some limitations in simulating lateral flow and soil water retention but shows promise as a robust calibration method in ungauged basins. Our study suggests that a mixed calibration, prioritizing streamflow while using ET data to further refine relevant parameters, offers a realistic and effective approach to improve calibration of hydrologic models.