What is happening to the climate today is seriously different from the Earth’s average. We are seeing weather extremes – droughts of historic proportions, unprecedented heat waves, severe hurricanes, storms and catastrophic floods. Climate disasters are becoming a new reality.

The fight against climate change must begin today. But it is clear that we will not feel the results of our efforts for decades to come. It is very difficult to take action for a delayed result in the distant, distant future.

To understand the possible consequences and to develop better adaptation strategies, we need climate models that can predict the climate in different regions of the globe for decades.

Unlike weather forecasting, which primarily models atmospheric physics, climate models are multi-year simulations that model the physics, chemistry, and biology of the atmosphere, water, ice, land, and human activities.

Climate simulations today can be tuned to a resolution of 10 to 100 kilometers.

But to simulate changes in the global water cycle – the movement of water in the oceans, sea ice, land surface and groundwater through the atmosphere and clouds – a higher resolution is required. Changes in this system lead to increased storms and droughts.

Meter-scale resolution is needed to simulate clouds reflecting sunlight back into space. Scientists estimate that these resolutions would require millions and billions of times more computing power than what is currently available. It would take decades to achieve this with the normal dynamics of computational evolution (an acceleration of 10 times every five years).

Today, for the first time, we can say that we have the technology to model climate at ultra-high resolution to predict changes in extreme weather conditions in the region at the speed of light for decades to come.

We can achieve speedups of millions of times by combining three technologies: GPU-accelerated computing, deep learning and breakthroughs in neural networks with knowledge of the laws of physics, and AI-enabled supercomputers along with huge amounts of observed and synthetic data for training.

With ultra-high-resolution methods, we might get the jump of billions of times needed for ultra-high-resolution climate modeling. Countries, cities and towns will be able to get early warnings to adapt and make infrastructure more resilient. And with more accurate predictions, people and countries will finally be able to take action.

NVIDIA is willing to devote significant resources to this task. We will bring our computing capabilities and expertise to the global climate science community.

This week, NVIDIA announced plans to build the world’s most powerful artificial intelligence supercomputer for predicting climate change. The Earth-2 system, or E-2, will create a digital twin of the Earth in the Omniverse.

This system will be analogous to Cambridge-1, the world’s most powerful artificial intelligence supercomputer for medical research. We unveiled Cambridge-1 in the UK this summer, and it is already being used by a number of leading medical companies.

All the technology we’ve built over the years is needed to make Earth-2 a reality. I can’t think of a bigger or more important way to use them.

The post NVIDIA to build Earth-2 supercomputer to see our future appeared first on ZED.