A new breakthrough in artificial intelligence by Google Deepmind could potentially revolutionize weather forecasts, according to the company. The GraphCast system, built by Deepmind, has the capability to produce 10-day weather forecasts in just one minute, with an unparalleled level of accuracy. This not only allows for more efficient forecasting, but also enables early identification of extreme weather events such as cyclones and floods. By providing advanced warnings, this technology has the potential to save lives and help people better prepare for hazardous conditions.
Current weather forecasts typically rely on Numerical Weather Prediction (NWP), a system that combines physics equations with computer algorithms to run on supercomputers. This approach requires significant computing resources and expertise from weather forecasters. In contrast, the GraphCast system utilizes deep learning techniques that learn from historical weather data to model how the Earth’s weather changes over time. While the creation of the model was intensive, involving training on decades of weather data, it now significantly reduces the resources required for predicting the weather. The time needed for a 10-day forecast has been reduced from hours on hundreds of machines in a supercomputer to just one minute on a single machine.
In tests, the GraphCast system outperformed the traditional NWP system in 90% of cases. It was even able to accurately predict the path of Hurricane Lee nine days before it arrived, compared to the traditional forecasts which provided a six-day lead time. This system’s ability to identify extreme weather events without specifically being trained on them is a remarkable advancement. For instance, it can predict areas with temperatures exceeding historical records, allowing people to anticipate and prepare for heat waves.
Deepmind also highlights the importance of this technology in the context of the climate crisis. The ability to predict extreme temperatures can be particularly valuable in efforts to mitigate the consequences of global warming. The company plans to open source the GraphCast system, enabling others to benefit from its capabilities. This move may foster further research and tools to address the various environmental challenges we face.
“GraphCast’s prediction of extreme temperatures could be particularly useful given the climate crisis. The system can predict areas where the heat will arrive above the historical top temperatures, allowing people to anticipate heat waves and prepare for them,” notes Deepmind.
Google Deepmind’s AI breakthrough presents a significant leap forward in weather forecasting. By combining deep learning techniques and historical weather data, this system has demonstrated remarkable accuracy and efficiency. With the potential to save lives and aid in climate change mitigation efforts, this technology offers unprecedented capabilities to enhance our understanding and preparation for weather events.
Use the share button below if you liked it.