Enhancing SWAT model predictivity using multi-objective calibration: effects of integrating remotely sensed evapotranspiration and leaf area index

Hydrological process in a catchment consists of multiple internal processes with complex interactions. The conventional practice of using a single variable to calibrate a hydrological model can lead to equifinality, which can result in substantial uncertainty in capturing the internal processes. Since multi-objective calibration process can be a way to solve this problem, a novel stepwise series of calibration experiments are employed by integrating (i) spatiotemporal remotely sensed actual evapotranspiration (AET) data in multi objective calibration process, (ii) Leaf area index (LAI) data, and (iii) AET and LAI data together in multi-objective calibration with Soil and Water Assessment Tool (SWAT). This calibration process aims for reducing the equifinality by improving the connected hydrologic internal processes rather than improvement in the assessment metrics at the catchment outlet. The results suggest that the integration of remotely sensed AET and LAI data in multi-objective calibration tend to increase model accuracy and reduce the prediction uncertainty, including significant improvement in simulated evapotranspiration and streamflow. In comparison with the traditional SWAT model, which is based on user input and semiempirical equations to simulate hydrologic processes, AET and LAI integration in multi-objective calibration notably captured the improved evapotranspiration and actual vegetation dynamics estimations, which resulted improvement in streamflow calibration. The findings showed the necessity of integrating remotely sensed AET and LAI data to improve the calibration and reduce the equifinality of hydrologic model simulations. The presented methodology can be replicated through other hydrological models and remote sensing data sources in diverse hydrological systems in the world.