Accelerated computing, AI, and virtualization technologies are getting closer to the point of enabling accurate, high-resolution climate modeling and forecasts at an extremely detailed level that will drive more precise forecasts and more informed decisions about climate change policy, according to Nvidia founder and CEO Jensen Huang, who delivered a keynote speech at the this week’s Berlin Summit for the international Earth Virtualization Engines (EVE) initiative.
Since announcing a vision to create the Earth-2 digital twin back in 2021, Huang said Nvidia has been making progress on helping the EVE community achieve their goals in three areas: simulation, emulation, and visualization. The first has to do with achieving high enough resolution for accurate climate simulation at a scale down to under “a couple of square kilometers,” Huang said. The second is the ability to emulate the physics of climate systems to drive accurate weather modeling and predictions at that same scale, and the third is creating better ways for climate scientists and, especially, policymakers around the world to visualize the effects of climate change on the planet over time.
In a post keynote Q&A, Huang elaborated: “There are three technologies that Nvidia specializes in that we’re building in service of EVE and in climate science,” he said, according to a transcript of the Q&A. “One is accelerated computing. It's the same foundational technology that is used for artificial intelligence today. We're going to adapt that technology for climate simulation. Second is new artificial intelligence [generative AI] that learns the laws of physics, not just the laws of language… there's all kinds of artificial intelligence capability for understanding language, like ChatGPT, self-driving cars… In this particular case, we would like the AI to learn climate. And the third is a digital twin technology that can be planetary scale to be able to ingest a massive amount of data and make it possible at cloud level to allow people to interact with it. The three technologies that I mentioned represent basically 100% of Nvidia’s R&D.”
Regarding the progress being made on emulation and visualization, Huang said the Nvidia Modulus open-source framework for working with physics-based machine learning models has been matched with FourCastNet, a global, data-driven weather forecasting model, to use raw data to emulate weather patterns. During the keynote, Huang demonstrated how FourCastNet was able to accurately predict the path of Hurricane Harvey, which caused massive damage in Texas in 2017, by modeling the Coriolis force, the effect of the Earth’s rotation, on the storm. Huang also showed how these technologies could be used to visualize patterns not only on a worldwide scale and down to a much smaller city neighborhood scale, as he demonstrated a high-resolution interactive visualization of global-scale climate data in the cloud, zooming in from a view of the globe to a detailed view of Berlin.
Being able to see these patterns will help more people “internalize” the effects of climate change that may seem more abstract and esoteric coming out of the mouth of a climate researcher. Ultimately, that will help humanity prepare for what is coming, not with the notion that it still can reverse climate change, but that it must now find ways of adapting to climate change.
“We have to acknowledge that the world needs to start planning for adaptation, that extreme events are happening more and more frequently,” Huang said during the Q&A. “All of our might and mitigation is necessary and we need to accelerate that. But even with all of that, we have to deal with adaptation. And so EVE is a wonderful system that… will to do some of these things.”