Google DeepMind Releases New AI Weather Forecasting Model Capable of Predicting Global Weather in 1 Minute
-
Google DeepMind's newly launched global weather forecasting AI model, GraphCast, has drawn significant attention from the scientific community. This model can not only predict global weather for the next 10 days in just one minute but also accurately forecast extreme weather events. Trained on nearly 40 years of data from the European Centre for Medium-Range Weather Forecasts (ECMWF), GraphCast has demonstrated outstanding performance with an accuracy rate exceeding 90%.
Unlike traditional weather forecasting methods, GraphCast utilizes deep learning to predict weather conditions for the next 6 hours using only the past 6 hours and current weather data, with the capability to roll out predictions up to 10 days in advance. The model is built on a neural network architecture with a total of 36.7 million parameters. GraphCast's source code has been made publicly available, offering a powerful tool for scientists and forecasters.
<p class="article-content__img" style="margin-top: 0px; margin-bottom: 28px; padding: 0px; box-sizing: border-box; outline: 0px; border-width: 0px; border-style: solid; border-color: rgb(229, 231, 235); --tw-shadow: 0 0 #0000; --tw-ring-inset: var(--tw-empty, ); --tw-ring-offset-width: 0px; --tw-ring-offset-color: #fff; --tw-ring-color: rgba(41, 110, 228, 0.5); --tw-ring-offset-shadow: 0 0 #0000; --tw-ring-shadow: 0 0 #0000; line-height: 32px; text-align: center; color: rgb(59, 59, 59); word-break: break-word; font-family: 'PingFang SC', 'Microsoft YaHei', Helvetica, 'Hiragino Sans GB', 'WenQuanYi Micro Hei', sans-serif; letter-spacing: 0.5px; display: flex; -webkit-box-align: center; align-items: center; -webkit-box-pack: center; justify-content: center; flex-direction: column; text-wrap: wrap; background-color: rgb(255, 255, 255);"><img src="https://www.cy211.cn/uploads/allimg/20231115/1-231115153Kc50.png" title="image.png" alt="image.png" style="margin: 0px auto; padding: 0px; box-sizing: border-box; outline: 0px; border: 1px solid rgb(238, 238, 238); --tw-shadow: 0 0 #0000; --tw-ring-inset: var(--tw-empty, ); --tw-ring-offset-width: 0px; --tw-ring-offset-color: #fff; --tw-ring-color: rgba(41, 110, 228, 0.5); --tw-ring-offset-shadow: 0 0 #0000; --tw-ring-shadow: 0 0 #0000; max-width: 700px; background: url('../img/bglogo2.svg') center center no-repeat rgb(247, 248, 249); box-shadow: rgba(27, 95, 160, 0.1) 0px 1px 3px; display: inline-block;"/></p>
Paper address: https://www.science.org/doi/10.1126/science.adi2336
The research results indicate that GraphCast performs exceptionally well in the meteorologically important Z500 field, showing a 7%-14% improvement in skill scores. Compared to traditional models, GraphCast achieves 90.3% accuracy across 1380 test variables, demonstrating particular advantages in identifying extreme weather events. Early identification of severe weather events is crucial for mitigating the impact of storms and extreme weather on communities.
The model also excels in cyclone tracking, predicting hurricane landfall locations and times up to 9 days in advance, far exceeding traditional prediction models. GraphCast can also characterize atmospheric river features, helping to develop emergency plans alongside AI models for flood prediction. In the context of global warming, accurate prediction of extreme temperatures becomes increasingly important. GraphCast can identify when temperatures at any location will exceed historical record highs, providing strong support for heatwave forecasting.
The open-source nature of GraphCast means that scientists and forecasters globally can leverage this advanced weather forecasting tool, thereby advancing the field of weather prediction. Overall, GraphCast provides humanity with a more powerful tool to face natural disasters by delivering accurate and efficient early warnings in advance, promoting the progress of AI in weather forecasting. The launch of this model marks a significant step forward in weather forecasting, offering innovative solutions for future meteorological research and disaster prevention efforts.
