Enhancing physically based and distributed hydrological model calibration through internal state variable constraints

Abstract. Accurately representing hydrological processes remains a major challenge in hydrological modeling. Recent studies have demonstrated the benefits of multi-variable calibration, which integrates additional hydrological variables such as evapotranspiration and soil moisture alongside streamflow to improve model realism. However, groundwater recharge as a calibration variable remains relatively underexplored. This study evaluates how incorporating groundwater recharge into the calibration of the Water Balance Simulation Model (WaSiM) affects hydrological variables representation. Three configurations were tested: Baseline (BL) with streamflow-only calibration, Physical Groundwater Model (GW) with physically-based groundwater flow, and Physical Groundwater with Recharge Calibration (GW-RC), which further constrains groundwater recharge during calibration. The models were calibrated and applied to 34 catchments in Southern Québec. Their performance was evaluated using the Kling-Gupta Efficiency (KGE) for streamflow and spatial estimates of groundwater recharge derived from a previous research project conducted in the same region. Results indicate that while calibrating on streamflow alone produces high KGE values (median KGE = 0.83 for GW and 0.82 for BL), but it comes at the cost of misrepresenting subsurface hydrological processes. Adding groundwater recharge constraints (GW-RC) reduce streamflow performance, with a median KGE of 0.77 for GW-RC, but improves hydrological variable representation, especially in seasonal runoff patterns, where it better captures the balance between surface runoff and interflow during snowmelt. Additionally, GW-RC showed the smallest differences with the groundwater recharge estimates. These findings illustrate the consequence of equifinality in streamflow-based calibration, where multiple parameter sets can yield similar streamflow outputs while misrepresenting internal hydrological processes. Incorporating groundwater recharge constraints improves the representation of internal hydrological processes while maintaining strong streamflow simulation performance, which could ultimately enhance reliability of climate change adaptation and water resource management strategies.