Improving Wind Power Forecasting with Meteomatics High-Resolution Models

Wind energy production is dependent on weather conditions. As renewable energy gains wider adoption, wind turbines are increasingly installed in complex terrain or in offshore locations. However, current numerical weather prediction models (NWP) often struggle to accurately forecast weather over complex terrain due to limitations in spatial resolution. Additionally, current NWP models do not account for the effects of turbines on wind flows, making it difficult to predict localized variations in wind speed and direction, both within and downstream from large-scale wind park. This leads to inaccurate predictions of power production, impacting the efficiency and reliability of renewable energy systems.

To address this challenge, Meteomatics has developed an operational high-resolution NWP model with a horizontal grid spacing of 1 km spatial resolution, a 72-hour forecast horizon, and an hourly refresh rate. In addition to standard data sources such as weather stations, radar, satellite data, and radiosondes, data from Meteomatics' network of Meteodrones are assimilated into this NWP model. Meteodrones are small, unmanned aircraft systems that collect vertical atmospheric profiles up to 20,000 feet above ground. Beside modifications to parameterization to better account for maritime PBL dynamics, the model parametrizes the influence of wind turbines on the flow for offshore wind parks, enabling accurate representation of wind wake effects. As a result, the model provides highly accurate and precise wind forecasts at turbine hub height, optimizing onshore and offshore wind power forecasts. The model currently covers the pan-European domain (EURO1k) and will be extended to North America in early 2025.

Data from Meteomatics' high-resolution model is ingested into a specialized wind power forecast system that combines various weather models, aiming to reduce the average error. The addition of the high-resolution NWP model reduced the normalized root mean square error (RMSE) by 8.1% for intraday and 8.5% for day-ahead wind power forecast. Moreover, a comparison of day-ahead forecasts with actual production data, combined with balancing energy costs, demonstrates improved earnings thanks to the addition of the high-resolution NWP model. This underscores the value of high-resolution NWP models and wind turbine parametrizations to enhance the cost efficiency of renewable energies and promote greater integration into the energy mix, thereby contributing to reduced CO2 emissions.