ECMWF’s AIFS-ENS Model Now Available via Meteomatics API and MetX

Meteomatics customers can now access ECMWF’s AI-based ensemble weather model, AIFS-ENS, directly through the Weather API and MetX with 90 m downscaling. Designed for medium-range forecasting (up to 15 days), the model supports scenario planning and risk mitigation, particularly valuable in sectors like energy trading.

Why AIFS-ENS?

According to ECMWF, “the new ensemble model outperforms state-of-the-art physics-based models for many measures, including air temperature, with gains of up to 20%.” Similar to ECMWF’s numerical ensemble forecast, AIFS-ENS relies on physics-based data assimilation to initialize its forecasts. However, once the initial conditions are set, AIFS-ENS can produce 51 ensemble members more than ten times faster, while reducing energy consumption by approximately 1,000 times.

Ensemble forecasts provide a probabilistic view of future weather conditions, enabling better assessment of uncertainty, particularly valuable for medium-range forecasting (3 to 15 days). In energy trading, this approach helps reduce risk by quantifying potential outcomes rather than relying on single-scenario forecasts. Industry experts are showing strong interest in AIFS-ENS, recognizing its potential to assess the likelihood of extreme events and evaluate scenario spreads, ultimately supporting more informed and resilient decision-making.

Added Value from Meteomatics: Enhancing Resolution from 20 km to 90 m

ECMWF’s native AIFS-ENS resolution is 0.25° (~20 km), but we apply our proprietary downscaling post-processing technology to deliver forecasts at an ultra-fine 90-meter resolution.

This means:

• Sharper forecasts for specific sites or assets

• More reliable input for simulations, analytics, and decision-making

• Better alignment with real-world terrain

AIFS-ENS Is Not Suited for Short-Term Forecasting

While AIFS-ENS offers advantages for medium-range probabilistic forecasting, it is not designed for short-term or high-resolution local forecasts (e.g., 0–48 hours) since it is limited by its temporal resolution of 6 hours. For users needing accurate, high-resolution weather data on short time horizons, we recommend switching to our EURO1k and US1k models. These models are physics-based, run at a native spatial resolution of 1 km, native temporal resolution of 15 minutes and are updated hourly, making them well-suited for forecasting local weather events such as thunderstorms, frontal passages or fog.

In summary:

• Use AIFS-ENS: for medium-range probabilistic planning, scenario analysis, and risk evaluation.

• Use EURO1k / US1k: for short-term operational decision-making where local accuracy and rapid update cycles are essential.

If you’re unsure which model best fits your use case, your Customer Success Manager can provide guidance based on your industry and application needs.

The AIFS-ENS model is now fully integrated into both the API and MetX interfaces.

• If your account includes model selection access, AIFS-ENS is already available. Simply select it from the model selector dropdown in MetX or specify it in your API call

• If your account only includes access to the model mix, you’ll need to upgrade your plan to access AIFS-ENS.

Reach out to your Customer Success Manager for more details or to activate access.