The article discusses the development and application of the DEMETER (Development of a European Multimodel Ensemble System for Seasonal to Interannual Prediction) project, which aims to enhance seasonal climate prediction by combining multiple global coupled climate models. The project involves seven leading European models, each with its own ensemble of initial conditions to account for uncertainties in the atmosphere and oceans. The ensemble forecasts are used to produce hindcasts over a 40-year period, with a focus on skill in predicting El Niño, precipitation, and climate indices like the Pacific-North American pattern (PNA) and North Atlantic Oscillation (NAO). Key findings include: - The multimodel ensemble consistently outperforms single-model ensembles in terms of skill, particularly in predicting El Niño and precipitation. - The ensemble's probabilistic skill is higher than that of single-model ensembles, leading to greater economic value. - Downscaling techniques, including statistical and dynamical methods, are used to improve the spatial resolution of the model outputs for applications such as crop yield and malaria prediction. - Crop yield predictions using the DEMETER system show promising results, with the ensemble spread capturing interannual variations and providing reliable probability distributions. - A biological model driven by temperature and precipitation data successfully predicts malaria prevalence, demonstrating the potential for early warning systems in malaria-endemic regions. The article also discusses sensitivity experiments that assess the impact of different methods and model components on the ensemble forecasts, highlighting the benefits of a fully coupled system over an atmosphere-only model.The article discusses the development and application of the DEMETER (Development of a European Multimodel Ensemble System for Seasonal to Interannual Prediction) project, which aims to enhance seasonal climate prediction by combining multiple global coupled climate models. The project involves seven leading European models, each with its own ensemble of initial conditions to account for uncertainties in the atmosphere and oceans. The ensemble forecasts are used to produce hindcasts over a 40-year period, with a focus on skill in predicting El Niño, precipitation, and climate indices like the Pacific-North American pattern (PNA) and North Atlantic Oscillation (NAO). Key findings include: - The multimodel ensemble consistently outperforms single-model ensembles in terms of skill, particularly in predicting El Niño and precipitation. - The ensemble's probabilistic skill is higher than that of single-model ensembles, leading to greater economic value. - Downscaling techniques, including statistical and dynamical methods, are used to improve the spatial resolution of the model outputs for applications such as crop yield and malaria prediction. - Crop yield predictions using the DEMETER system show promising results, with the ensemble spread capturing interannual variations and providing reliable probability distributions. - A biological model driven by temperature and precipitation data successfully predicts malaria prevalence, demonstrating the potential for early warning systems in malaria-endemic regions. The article also discusses sensitivity experiments that assess the impact of different methods and model components on the ensemble forecasts, highlighting the benefits of a fully coupled system over an atmosphere-only model.