Google recently shared new research according to which they can make “nearly instantaneous” weather forecasts possible, which would be an excellent tool in a world where climate change shapes the way things work.
In the paper, Google’s researchers confirmed in just minutes, they generated relevant rainfall predictions at a 1km resolution up to six hours ahead of time, which is more than the existing techniques can generate. Compared to the existing traditional forecasting techniques, optical flow (OF) predictions, and simulation forecasting, Google’s approach is more efficient. For example, simulations made by US federal agencies for weather forecasting occupy a lot of data and take hours to run.
According to Google software engineer Jason Hickey, “if it takes 6 hours to compute a forecast, that allows only 3-4 runs per day and resulting in forecasts based on 6+ hour old data, which limits our knowledge of what is happening right now.”
Researchers also believe that Google’s AI models provide “an essential tool needed for effective adaptation to climate change, particularly for extreme weather.”
The most beneficial thing about the use of machine learning in weather forecasts stays in short time, speedy predictions, while longer forecasts are more appropriate for powerful models.