Abstract. Year-round river discharge estimation and forecasting is a critical component of sustainable water resource management. In cold-climate regions such as Canada, this task is complicated by dynamic river-ice conditions which alter channel hydraulics and render open-water rating curves invalid. Some methods such as backwater-adjusted rating curves and ice thickness-based rating curves have been developed. However, these methods are site specific and subjective to human judgement. It is therefore an active field of research and development. The current study develops and assesses the performance of a river ice forecasting system for the Chaudière River in Quebec based on coupled hydrologic-hydraulic modelling approach within the Delft Flood Early Warning System (Delft-FEWS) platform. The current configuration of the system integrates (i) meteorological products such as the Regional Ensemble Prediction System (REPS); (ii) a hydrological module implemented through the HydrOlOgical Prediction LAboratory (HOOPLA), a multi-model based hydrological modelling framework; and (iii) hydraulic module implemented through a 1D steady and unsteady HEC-RAS river ice models. The system produces ensemble forecasts for discharge and water level and provides flexibility to modify various dynamic parameters such as discharge timeseries, and ice properties. Performance of the coupled modelling approach was assessed against “Perfect Forecast” for selected winter events between 2020 and 2023 using the root mean square error (RMSE) and percent bias (PBias) metrics. The hydrologic module of the system consistently underestimated the under-ice discharges. This can be explained by the inherent uncertainty in the under-ice discharge estimates used as observations as well as uncertainty in the model states. The hydraulic module had an RMSE between 0.1–0.5 m. The higher error in RMSE can be attributed to the uncertainty in the ice thickness estimates at one of the stations. The PBias analysis of the hydraulic module also confirmed that the forecasted discharges were under-estimated by the hydrologic module.
Usman et al. (Tue,) studied this question.