Skilful seasonal streamflow forecasting using a fully coupled global climate model

Abstract. The seasonal streamflow forecast (SSF) is a crucial decision-making, planning and management tool for disaster prevention, navigation, agriculture and hydropower generation. This study demonstrates for the first time the capacity of a fully coupled operational global forecast system to directly provide skilful seasonal streamflow predictions through a physically consistent and convenient single-step workflow for forecast production. We assess the skill of the SSF derived from the operational Météo France forecast system SYS8, based on the in-house fully coupled atmosphere-‐ocean–land general circulation model of the sixth generation, CNRM‐CM6‐1. An advanced river routing model interacts with the land and atmosphere via surface and/or sub-surface runoff, aquifer exchange, and open-water evaporation to predict river streamflow. The actual skill is evaluated against streamflow observations, with the ensemble streamflow prediction (ESP) approach being used as a benchmark. Results show that the online coupled forecast system is overall more skilful than ESP in predicting streamflow for the summer and winter seasons. This improvement is particularly notable with enhanced land water storage initial conditions, especially in summer and in large basins where the low-flow response is influenced by soil water storage. Predicting climate anomalies is crucial in winter forecasting, and results consistently suggest that the atmospheric forecast of the fully coupled CNRM‐CM6‐1 model contributes to better seasonal streamflow forecasts than the climatology-based ESP benchmark. This study showcases the capacity of an operational seasonal forecast system based on a general circulation model to deliver relevant streamflow predictions. Additionally, the positive response to enhanced initial hydrological conditions pinpoints the efforts still needed to further improve land initialization strategies, possibly through land data assimilation systems.