This session focuses on recent advances in probabilistic modelling and ensemble forecasting of streamflow and other environmental variables using machine learning (ML) and deep learning (DL) approaches. Contributions are invited on uncertainty quantification, probabilistic prediction, ensemble generation and post-processing, the use of meteorological and climate data for both model training and real-time forecasting, and decision-support frameworks applicable in contexts of limited information and uncertainty. Topics include hybrid process-based and ML/DL models, explainable AI, foundation models, ensemble learning, mixed-model approaches, and operational forecasting systems. Applications may address streamflow, floods, droughts, water quality, sediment transport, groundwater, ecohydrological variables, and other climate-sensitive environmental processes. Emphasis is placed on reducing, explaining, and communicating predictive uncertainty, and on the use of ensemble and probabilistic models in risk-informed operational decision-making. Participations that clarify the relationship between ensemble spread and predictive uncertainty, or that benchmark probabilistic methods against each other, are particularly welcome.