ECCC to roll out AI‑assisted weather model this spring, aiming for earlier storm warnings

Environment and Climate Change Canada says it will begin rolling out an artificial intelligence–assisted weather model this spring that could deliver earlier warnings and more accurate forecasts for severe weather.

The department says the hybrid system combines artificial intelligence with its existing physics-based forecasting model to improve accuracy across short-, medium- and long-range forecasts, while giving forecasters more lead time to detect major weather systems.

According to the department, warning times for events such as winter storms, heat waves and atmospheric rivers could be extended by eight to more than 24 hours, and a six-day forecast could be as accurate as the current five-day outlook. Officials say the system is intended to improve confidence in predicting when weather events will begin and how storms will track, supporting earlier public alerts and emergency preparedness.

The model has been tested over the past year by running alongside the existing forecasting system and by using it to simulate past storms. Federal officials describe the approach as using artificial intelligence to identify patterns in large volumes of historical weather data, while traditional physics-based modelling continues to account for local conditions such as temperature, wind and precipitation.

An Environment and Climate Change Canada roadmap on artificial intelligence integration states that “physics-based models will remain at the core of our forecast systems” as the department gradually incorporates AI into operations.

Separate federal guidance on artificial intelligence notes that AI systems can produce inaccurate or misleading outputs and require human oversight, with meteorologists continuing to play a central role in interpreting forecasts and communicating risks to the public.

The department says it will continue to rely on meteorologists to assess model outputs and issue official forecasts and warnings.