NOAA dives into AI
As Earth science researchers get greater access to data from satellites, unmanned systems and commercial sources, they are increasingly relying on artificial intelligence to help them better interpret that data to understand ocean resources.
The National Oceanic and Atmospheric Administration’s AI strategy is designed to reduce the cost of data processing and deliver higher quality and more timely services to advance the agency’s mission.
“AI capabilities are already demonstrating significant improvements in performance and skill at vastly reduced costs and compute time,” the report said, and NOAA wants to accelerate the technology’s adoption.
The strategy has five goals that target improving NOAA’s computational and analytical capacities, enhancing AI research, transitioning technology to the marketplace, boosting AI know-how of its workforce and creating partnerships across the agency, the federal government and with external researchers and commercial communities.
To build the kind of organization that can advance AI, NOAA will explore establishing an AI center to coordinate research among partners and a portal to host shared applications. It will also embrace cloud and big data technologies and prioritize AI-based approaches when drafting its budget.
To keep AI front and center in its mission, NOAA plans to prioritize and track AI research in its funding opportunities, grants and requests for proposals, and also establish an annual AI competition. It plans building test beds to develop best practices and training data to improve algorithms along with evaluating performance of its AI models.
As NOAA works to move its AI research to other government, academic and commercial partners, it will develop technical guidelines covering best practices and standards for AI training data, training practices as well as guidelines to evaluate AI models to ensure their integrity and reliability.
To increase AI proficiency in the workforce, the agency envisions more on-scene training, assigning staff to offices where cross-pollination of AI expertise would raise the overall proficiency of the workforce. It will also support education, training courses and internships and improve recruitment, retention and the hiring pool for AI proficient NOAA workforce.
The science agency currently uses AI for a number of applications, including assessing marine mammal and fish populations, exploring the deep sea with robotics, improving quality control of satellite observations and weather models and using machine learning for interpretation of earth observations.
“NOAA already has robust experience with these applications across a range of mission areas and is delivering successful operational systems that are beginning to pay dividends, whether in helping to improve weather forecasts or fish stock assessments,” said Tim Gallaudet, the Commerce Department's assistant secretary for oceans and atmosphere and deputy NOAA administrator.
NOAA also released its strategies for unmanned systems and “omics,” the technologies for analyzing biomaterials such as DNA, RNA or proteins. A draft version of a fourth strategy, NOAA’s Cloud Strategy, is being finalized for a later release.
Connect with the GCN staff on Twitter @GCNtech.