Integration of river ice information in hydrodynamic models for enhanced and continuous streamflow forecast

River ice significantly influences streamflow in northern watersheds and leads to operational challenges for water resources managers due to the impact on water supply, especially during ice formation and breakup jams. This study focuses on the Upper Saint John River, in the state of Maine, US, an ice-prone segment of the river, specifically between Dickey and Fort Kent, which frequently experiences ice-induced flooding. The goal of this study is to enhance streamflow simulation in ice-affected regions through an integrated modeling approach that couples the NOAA National Water Model (NWM) and the Hydraulic Engineering Center's River Analysis System (HEC-RAS) to address complexities introduced by river ice dynamics. Streamflow estimates from the NWM Retrospective simulations were employed as boundary conditions to drive the HEC-RAS model. While previous investigations utilized the HEC-RAS 1-D steady model for river ice modeling, this study introduces composite roughness into a 2-D unsteady model and a novel framework to update roughness values automatically, leveraging remote sensing observations. Simulations were performed for 12 years, focusing on both open water and ice-affected conditions, guided by remote sensing and composite channel roughness updates according to the changes in ice conditions. The coupled NWM/HEC-RAS model demonstrated improved performance in terms of 12-year average KGE (0.50) compared to the standalone NWM model (0.35), which highlights the potential of the proposed framework to simulate streamflow under ice conditions. • Enhanced streamflow simulation in ice-affected regions using integrated modeling. • River ice significantly impacts streamflow and water resource management. • Novel framework couples NWM and HEC-RAS for better streamflow estimates. • Remote sensing aids in detecting and updating river ice conditions in the model. • Improved model performance with KGE of 0.50 compared to 0.35 for standalone NWM.