Enhancing the simulation accuracy of long-term runoff models using digital filter methods

The objectives of this study propose a parameter estimation method that can consider both the total runoff and the runoff component by integrating the digital filter method and the long-term runoff models (SWAT, TANK), and evaluate the appropriateness of the applied methods. The study area is the Soyang River Dam basin, and parameter calibration and validation are performed by dividing it into a parameter estimation method considering the total runoff and a parameter estimation method considering the runoff component. In both methods, the fit between the observation and simulation runoff was excellent, and the model performance was found to be good with a coefficient of determination (R2) of 0.73~0.87, and NSE of 0.67~0.85. As a result of comparison with each method, it was confirmed that the simulation accuracy was improved when applying the method considering the runoff component in both the SWAT model and the TANK model. When comparing between the models, the SWAT model showed better statistics in both methods, but the effect of applying to the method was found to be insignificant. However, even though the TANK model did not specifically consider the physical characteristics of the methodology, the statistical value of NSE increased by 17% when integrating a method such as a digital filter. In other words, the applicability to the digital filter method was found to be better in the TANK model, and when a hydrograph separation method such as a digital filter is applied to a conceptual model such as this model, it is judged that more improved simulation results can be obtained than the physical model. Accordingly, it is judged that estimating the parameters by considering the runoff component will be more accurate than estimating the parameters by considering only the total runoff when simulating the hydrological model.