First protein folding. Working with Met Office, weather services in the UK, DeepMind has developed an in-depth study tool called DGMR that can accurately predict rainfall potential in the next 90 minutes – one of the biggest challenges in climate forecasting.
Comparing rudely with available tools, the 12 experts considered DGMR’s predictions to be the best in a number of areas — including location, length, movement, and rain power — of 89% of the time. The result was published in Natural paper today.
DeepMind’s new tool no AlphaFold, which opened up a major problem in biology that scientists have struggled with them for years. Yet even a small change in the forecasts.
Predicting rainfall, especially heavy rainfall, is important for many industries, from external events to airplanes to emergency services. But doing well is hard work. Determining the amount of water in the atmosphere, and when and how it will fall, depends on a number of factors, such as temperature changes, cloud formation, and wind. All of these are complex enough in and of themselves, but they are even more complex when combined.
The best methods of prediction are the use of large-scale experiments on space science computers. This is good for predicting the future but it is not good enough to predict what will happen in the next hour or so, called nowcasting. In-depth study methods of the past have been developed, but this works best for one thing, such as location forecasting, and the destruction of something else, such as power forecasting.
THINK
“Rainfall changes are still a major problem for climate experts,” said Greg Carbin, chief of staff at the NOAA Weather Prediction Center in the US, who is not involved in the project.
The DeepMind team trained their AI on radar intelligence. Many countries produce frequent visuals throughout the day for radar measurements that follow the design and movement of clouds. In the UK, for example, a new reading is released every five minutes. Putting these displays together provides a stand-alone video that shows how the rain is going across the country, much like the shows you see on TV.