The project is part of an initiative developed within the European Programme Destination Earth (DestinE), which in turn is included among the Pilot Services projects for DestinE Impact Sectors, specifically the Services supporting resilience and the mitigation of impacts related to extreme weather events.
Goals and expected results
This proposal focuses on the development and demonstration of SynCast, an advanced machine learning-based system designed to integrate multiple forecast model outputs into a single optimized product. The key objective is to improve the accuracy and reliability of forecasts for critical near-surface parameters. SynCast leverages the capabilities of Destination Earth’s Digital Twin infrastructure and high-performance computing (HPC) to provide real-time, high-resolution forecasts, addressing the growing need for accuracy in response to increasingly extreme weather events.
CIMA Research Foundation’s contribution
SynCast integrates forecast data from multiple models, including DestinE’s Extremes DT and Climate DT, along with operational models, e.g., ECMWF’s IFS and ICON. CIMA Research Foundation is responsible for collecting and aligning forecast data at different temporal and spatial resolutions to ensure compatibility with the SynCast architecture. To guarantee that the combined forecast for wind, temperature, and humidity is both accurate and reliable, CIMA also validates the results through:
- Cross-validation: comparison of historical forecast data with actual observations for wind, temperature, and humidity over a range of geographic locations and time periods
- Performance Metrics: measuring the accuracy and quality of the combined product through key metrics such as RMSE, bias reduction, and correlation with real-world observations
- Case studies: analysis of case studies in sectors such as renewable energy, agriculture, and disaster management
