Nvidia on Monday unveiled a platform with a digital twin of Earth for climate research. The purpose of Earth-2 is to deliver the most realistic simulation of weather conditions possible. With this, countries can avoid and minimize damage from extreme weather events, such as hurricanes and strong storms — which are increasingly recurrent today.
The Weather Company and Taiwan’s meteorological agency will be the first companies to use Earth-2’s API. Following the focus of the market and Nvidia, the digital twin of our planet uses the DGX Cloud AI platform to train artificial intelligence. Earth-2 uses Nvidia’s CorrDiff generative AI model, which promises to generate images with resolutions 12.5 times higher than current competitors.
As Jensen Huang pointed out in the Earth-2 announcement, the technology aims to improve users’ preparedness for extreme weather conditions. “Climate disasters are now normal: historic droughts, catastrophic hurricanes and generational floods appear in the news with alarming frequency,” Huang said.
In the announcement, Nvidia recalls an article in Nature that argues that these extreme events generated an economic loss of $140 billion per year in the economy. Of this total, 63% are human lives (the study uses a specific methodology to make this calculation).
According to Nvidia, the combination of Earth-2 with services from weather companies could reduce the time it takes to update forecasts and deliver dangerous weather warnings to seconds. Today’s technology produces these results in minutes or hours.
The example presented by Nvidia uses the scenario of Taiwan, in which typhoons can generate flooding and landslides.
