Mon. Nov 21st, 2022

The US National Oceanic and Atmospheric Administration selected Lockheed Martin and NVIDIA to build a prototype system to accelerate the delivery of Earth environmental monitoring results and related imaging.

Using artificial intelligence techniques, such a system can reduce the time required to create complex weather visualizations by an order of magnitude.

A first-of-its-kind U.S. federal agency project, the Global Earth Observation Digital Twin, or EODT, will provide a prototype for visualizing terabytes of geophysical data from land, ocean, cryosphere, atmosphere, and space.

By combining data from a wide variety of sensor sources, the system will be able to provide information that is not only relevant, but also trusted by decision makers, says Lockheed Martin Space Senior Scientist Lynn Montgomery.

“We provide a one-stop shop for researchers and next-generation systems for not only current but also recent past environmental data,” Montgomery said. “Our collaboration with NVIDIA will provide NOAA with a timely global visualization of their massive datasets.”

Building on NVIDIA Omniverse

Based on NVIDIA Omniverse, the system can serve as a clearing house for scientists and researchers from a wide range of government agencies, which can be expanded over time to support a wide range of applications.

Support for the EODT pilot project is one of several NVIDIA initiatives to develop tools and technologies for large-scale, even planetary simulations.

Last November, NVIDIA announced the creation of a supercomputer called Earth-2, designed to predict climate change by creating a digital twin of the planet.

Last year, NVIDIA and Lockheed Martin announced they were partnering with the USDA Forest Service and the Colorado Division of Fire Prevention and Control to use artificial intelligence and digital dual modeling to better understand wildfires and prevent their spread.

And in March, NVIDIA announced an accelerated digital twin platform for scientific computing, consisting of the NVIDIA Modulus AI infrastructure for developing ML neural network models and the NVIDIA Omniverse 3D virtual world modeling platform.

The EODT project builds on these initiatives by relying on the NVIDIA Omniverse Nucleus to enable various applications to quickly import and export custom renderable assets to and from a central data store.

“This is a blueprint for a complex system using Omniverse where we will combine sensor data, architectural data, and AI-derived data, combined with a variety of visualization capabilities deployed in the cloud and across multiple workstations,” said Peter Messmer, director of HPC. Technology for Developers at NVIDIA. “This is a fantastic opportunity to show all these components in a real-world example.”

Rushing Effort

Efforts will move quickly, with a demonstration of the system’s ability to visualize sea surface temperature data scheduled for September next year. The system will take advantage of GPU compute instances from Amazon Web Services and on-premise NVIDIA DGX and OVX servers.

The fast and flexible system will be the prototype for visualizing geophysical variables from NOAA satellites and ground data sources from a wide range of sources.

These include temperature and humidity profiles, sea surface temperatures, sea ice concentrations, and solar wind data, among other sources.

This data will be collected by Lockheed Martin’s OpenRosetta3D software, which is widely used for complex large-scale image analysis, workflow and sensor fusion by government agencies such as NASA and the private sector.

NVIDIA will support the development of one-way connectors for importing “snapshots” of processed geospatial datasets from Lockheed’s OpenRosetta3D technology into NVIDIA Omniverse Nucleus as UDS input.

USD is an extensible open source ecosystem for describing, creating, modeling and collaborating in 3D worlds, originally invented by Pixar Animation Studios.

Montgomery explained that the Nucleus Omniverse will be vital to provide data quickly, in part because of Nucleus’ ability to only transfer what has changed in the dataset.

Nucleus, in turn, will feed these US dollar datasets to the Unity-based Lockheed Agatha 3D viewer, allowing users to quickly view data from multiple sensors on an interactive 3D earth and space platform.

The result is a system that will help researchers at NOAA, and eventually elsewhere, make faster decisions based on the latest data available.