AI real-time accurate weather forecast

Google's AI system accurately predicts rainfall 6 hours before the rainfall, with an accuracy of up to 1 km and a delay of just a few minutes.

Weather forecasting is hard work, but in recent years, experts say machine learning and artificial intelligence (AI) can help classify sunlight better. Recently, Google said it had new research to forecast the weather almost immediately.

Google real-time weather forecast system.(Photo: The Verge).

The research is at an early stage but the results are positive, but have not yet been integrated into any commercial system. Google scientists describe how they make accurate predictions of rainfall 6 hours before the rainfall, with an accuracy of up to 1km and a delay of just a few minutes. That's a huge leap forward compared to current techniques that take hours to make forecasts, and how to use data more complexly.

According to the team, rapid projections will be an essential tool to respond to current climate change, minimizing loss of life and property.

The biggest advantage of Google's approach over traditional forecasting techniques is speed. The scientists have compared the new method with two existing methods, optical flow prediction (OF), cloud movement and simulation predictions , creating physical simulations of the weather. They found that the old methods took a lot of time to process 100 terabytes of data from weather forecasting stations every day, and it took hours to process data on expensive supercomputers.

Fast weather forecasts will help limit loss of people and property.(Photo: AP).

Google's new weather forecast method will compare and produce results in minutes because it does not model complex weather systems, but predicts a specific data such as rainfall. Google's AI system uses data collected from 2017 to 2019 in the US provided by the National Oceanic and Atmospheric Administration (NOAA) to allow up to 10 days' forecasts.