In 2013, the World Weather Research Programme (WWRP) and the World Climate Research Programme (WCRP) initiated the Sub-seasonal to Seasonal Prediction (S2S) Project.
A significant milestone of S2S project has been the establishment of a database of near-real-time forecasts (2 days to three weeks behind real time) and reforecasts from 11 operational centres across the world. This database has supported a range of research on understanding the sources of predictability on S2S time scales and in the quantification of the skill of operational sub-seasonal to seasonal prediction models.
Additionally, sixteen real-time pilot projects (RTPs)3 were given access to the S2S database to explore the potential for applications of sub-seasonal to seasonal forecasts in decision-making. These RTPs highlighted the challenges of developing new services with users unfamiliar with sub-seasonal forecasting including the challenges of co-production. Through webinars and training courses the S2S Project built capacity in both research on prediction and the use of S2S forecasts.
Whilst the S2S Project significantly advanced understanding of the sources of S2S predictability and the skill of prediction systems, advances in the uptake of S2S forecasts to support decision-making were limited.
S2S forecast skill varies by region, season, weather regime, forecast variable, and event type. SAGE addresses the need to better understand and quantify these variations so that forecast information can be communicated with appropriate levels of confidence and uncertainty.
Realizing the value of S2S forecasts requires methods for translating probabilistic information into decision-relevant guidance. SAGE focuses on how forecast information can be incorporated into operational decisions, including when forecasts are skilful, when they are uncertain, and when they should not be used.
As Artificial Intelligence and Machine Learning become more widely used in forecast production and scientific analysis, SAGE considers how data-driven forecasts can be evaluated on S2S timescales and how AI/ML techniques may support calibration, post-processing, and communication of forecast information.
Agricultural decisions are sensitive to weather and climate variability across planting, crop management, irrigation, pest and disease risk, and harvest planning. SAGE supports co-production with users to explore how probabilistic S2S information can be tailored for agricultural decision-making.
Energy systems are increasingly affected by weather and climate variability as power systems decarbonize and rely more heavily on weather-dependent generation. SAGE examines how S2S information can support decisions related to demand, renewable generation, system resilience, and planning.
For disaster risk reduction, S2S forecasts offer an opportunity to improve preparedness for high-impact weather and climate events before shorter-range warnings are available. SAGE explores how forecast products can provide decision-relevant information for anticipatory action and early warning.
Health risks such as heat-related mortality and infectious disease transmission can be influenced by weather and climate conditions. SAGE considers how S2S forecasts can be integrated with health impact models and early warning systems to support preparedness and response.