Harnessing Ocean Data to Advance Weather Forecasting

Since 2022, Lockheed Martin has collaborated with NVIDIA to build a prototype of an AI-driven Earth and Space Observing Digital Twin. The software can process live streams of incoming weather data from satellites and other weather measurement tools, by applying AI and machine learning to analyze the data. The Digital Twin displays current global environmental conditions from satellite and ground-based observations and output from weather forecasting models.

With goal of attaining data for more accurate forecasting, the project over the past year has demonstrated one of NOAA’s critical data pipelines—sea surface temperatures—to highlight multi-sensor fusion from satellite and model data along with short-term temperature anomalies. Obtaining accurate sea surfaces temperature data helps provide a more complete picture of climatic conditions.

Satellites play an increasingly important role in weather forecasting. Both Lockheed Martin and Nvidia consider harnessing ocean surface temperature a key element to obtain a more accurate picture of climatic conditions that would aid predictive weather forecasting. The Earth and Space Observing Digital Twin aims to provide NOAA with an efficient and centralized approach to fuse and visualize data from various space and earth sensors. The demo showed NOAA and other government customers the potential of using AI to display high-resolution, accurate, and timely depictions of satellites and sensor data.

Related:Using the Digital Twin to Improve Weather Forecasting

More information can be found by viewing this video from Lockheed Martin.

The Earth and Space Observing Digital Twin project uses Lockheed Martin’s OpenRosetta3D software and the NVIDIA Omniverse development platform to build applications that aggregate data in real time. The Digital Twin uses these technologies to advance 4D visualizations and display this information in a digestible format. The OpenRosetta3D platform utilizes AI and machine learning (ML) to ingest, format, and fuse observations from multiple sources into a gridded data product and detect anomalies.

NVIDIA Omniverse Nucleus, the collaboration and database engine of its Omniverse world simulation platform, converts data into the Universal Scene Description framework, enabling data-sharing across multiple tools and between researchers. Agatha, a Lockheed Martin-developed visualization platform, will ingest this incoming data from Omniverse Nucleus and allow users to interact with it in an Earth-centric 3D environment.

Presently, NOAA receives terabytes of data about its five earth systems domains—the cryosphere, land, atmosphere, space weather and ocean—from numerous space and Earth-based sensor sources. The Earth Observations Digital Twin gives NOAA a high-resolution, accurate and timely depiction of global conditions, using current satellite and ground-based observations.

Related:Easing Battery Design Through Digital Twin and Simulation Tools

According to Lockheed Martin, its next plans are to incorporate additional data streams—such as space weather and sea ice concentrations—into the Earth and Space Observing Digital Twin next year.