Hydrologic model calibration approaches for highly regulated river basin: A comprehensive assessment

Study region The Montreal River basin in Ontario, Canada, is snow-dominated and highly regulated. Study focus This study investigates the effectiveness of different calibration strategies for hydrological modeling in regulated watersheds. Using the semi-distributed RAVEN models, it explores single-site versus multi-site, single-variable versus multi-variable, and single-objective versus multi-objective calibration schemes. The calibration process is automated through the Dynamically Dimensioned Search (DDS) and Pareto Archive Dynamically Dimensioned Search (PADDS) algorithms. New hydrological insights for the region The results reveal that multi-site, multi-variable calibration significantly improves the simulation of inflow, outflow, and SWE, particularly in snow-dominated environments, underscoring the importance of accounting for multiple hydrological processes in the region. Including SWE data in the calibration process enhances the model's ability to capture the timing and magnitude of snowmelt, a critical factor in reservoir inflow. Multi-objective calibration further addresses parameter uncertainty, providing robust simulations for water management under varying conditions. The study highlights the challenges of using reservoir outflow as an operational control, which leads to poor water level predictions, reinforcing the need for more region-specific operational constraints. Naturalization of streamflow using the reconstitution method reveals higher peak flows during summer and lower flows during winter than regulated flows, with low-flow conditions particularly impacted by regulation. These findings provide critical hydrological insights for improving water resource management in snow-dominated, regulated basins, with broader applications to similar systems.