About the S2S API

Climavision's Horizon AI Subseasonal to Seasonal (S2S) API leverages cutting-edge machine learning models and vast observational datasets to deliver highly accurate and granular climate forecasts. Our advanced technology integrates millions of observations, including ERA5 data, with proprietary climate simulations, feeding into our robust Climavision Data Lake. Through the use of advanced interpretable neural networks and extensive back-testing since 1940, our system generates forecasts with 500 ensemble members, ensuring reliability and precision.

The API includes over 50,000 energy assets across six continents and provides a probability density function for wind up to 100 meters above ground level (AGL). It features downscaled hourly temporal resolution, making it highly detailed and practical for various applications. Our rigorous validation process, available within the API going back to 1993, includes CRPSS, ACC, and MAE metrics broken down by month, further guaranteeing the reliability of our forecasts.

The point-based downscaling technology ensures that forecasts are not only accurate on a global scale but also finely tuned to local conditions, making them ideal for power generation forecasts and other energy-related applications. The global grid, gridded at 25km, provides comprehensive coverage and can be further refined to meet specific user needs. This innovative approach supports diverse needs, from high-impact event prediction to long-term energy asset management.

Our S2S API offers a global domain with forecasts available in hourly increments, ensuring timely and actionable insights. The forecasts are updated daily to provide the most current and accurate data. Outputs are probabilistic, allowing for a nuanced understanding of potential outcomes. The API provides gridded outputs at a 25km resolution, with point optimization available for specific locations. Furthermore, our validation stats, including CRPSS, MAE and ACC back to 1993, are accessible within the API, showcasing the long-term reliability and accuracy of our forecasts.

Each variable in our forecasts, such as power solar, power wind, solar irradiance, precipitation, wind speed and direction, and temperature, has specific time horizons based on scientific principles. These principles take into account the inherent predictability of each variable and the atmospheric dynamics that influence them.