Google Develops AI Model That Outperforms Current Weather Forecasts

Researchers at Google DeepMind have developed a machine learning model capable of producing reliable probabilistic weather forecasts based on current and future climate.

The model, called 'GenCast', outperforms traditional medium-range weather forecasts and is also able to better predict extreme weather conditions, tropical cyclone paths and wind power generation.

The details of the model were released on Wednesday in an article published in the journal Nature, reported the Efe news agency.

Having accurate weather forecasts is essential for people, governments and organizations to make essential decisions in their daily lives, from carrying an umbrella to assessing wind power generation or planning for extreme weather conditions to avoid disasters.

Traditional weather forecasts rely on numerical weather prediction methods that estimate the current weather and map it to a prediction of the future weather over time (known as deterministic forecasts), but this generates numerous potential scenarios that are combined to produce a weather forecast.

Now, a team of scientists at Google has developed a machine learning-based weather forecasting method called GenCast that can generate a probabilistic forecast, which predicts the likelihood of future weather based on current and past climate states.

The authors trained GenCast using 40 years (1979 to 2018) of data analysis of the best estimates of weather events.

Thanks to this training, the model is able to generate 15-day global forecasts for more than 80 atmospheric and surface variables in eight minutes.

When compared to the European Centre for Medium-Range Weather Forecasts (ENS) forecast suite – currently the world’s best performing medium-range forecast – they found that GenCast outperformed ENS on 97.2% of the 1,320 targets used.

GenCast is also more effective at predicting extreme weather, tropical cyclone tracks and wind power generation.

The authors argue that GenCast can generate more efficient and effective weather forecasts to support effective planning.