Real-time flood forecasting and early warning are rapidly evolving through advances in Earth observation, sensor networks, data assimilation, artificial intelligence (AI), machine learning (ML), high-performance computing, and digital technologies. These developments offer new opportunities to improve forecast accuracy, lead time, spatial resolution, uncertainty estimation, and the translation of forecasts into actionable warnings.
This session invites contributions covering the full chain from real-time data to forecasting, impact assessment, warning, and early action. We particularly welcome studies combining process-based knowledge with emerging data-driven and hybrid approaches, as well as innovative operational applications.
Topics include, but are not limited to:
• Real-time data: ground observations, radar, satellite and remote sensing, IoT, crowdsourcing, data quality control, imputation, assimilation, and multi-source data fusion.
• Flood forecasting: process-based, conceptual, hybrid, AI/ML and physics-informed approaches; nowcasting; rapid forecasting; and forecasting in data-scarce regions.
• Uncertainty and reliability: ensemble and probabilistic forecasting, uncertainty quantification, explainable AI, model transferability, robustness, and extreme-event forecasting.
• Forecast-to-action: impact-based forecasting, inundation and damage prediction, early warning and anticipatory action, decision-support systems, and communication of forecast uncertainty.
• Digital innovations: digital twins, cloud and edge computing, high-performance computing, open-source platforms, and immersive visualisation.
• Operational and community applications: real-world forecasting systems, stakeholder engagement, citizen science, emergency response, and community-centred early warning.
• Emerging directions: foundation models, generative AI, multimodal data fusion, autonomous forecasting, and next-generation Earth-system and hydrological digital twins.