Thursday August 27, 2020 0 comments
FORT COLLINS -- Colorado State University has teamed up with other leading organizations to help run a new $20 million National Science Foundation-funded institute focused on creating trustworthy artificial intelligence to study climate, weather, and coastal oceanography.
"In recent years AI has been proven to be a potential game changer for many applications in the environmental sciences," said Imme Ebert-Uphoff, a research professor in the Department of Electrical and Computer Engineering and the Cooperative Institute for Research in the Atmosphere, who will lead CSU's part of the institute, with assistance from Associate Professor Elizabeth Barnes of the Department of Atmospheric Science, Professor Chuck Anderson of the Department of Computer Science, and other CSU researchers and staff.
Although AI is ubiquitous in daily life, from smart cars to smartphones, the use of AI in scientific research and risk communications is relatively new, and not generally well-understood by some scientific researchers who see the method as a ‘black-box’ solution that potentially obscures the underlying science.
Ensuring that new applications and methods of artificial intelligence are accessible to the public and well-understood by researchers will be one of the key goals of the new institute.
"Now is the time to focus on how to use AI responsibly in these areas, namely to build on our current work using explainable AI for weather and climate to develop methods of trustworthy AI that are dependable and whose reasoning can be explained to the stakeholders," Ebert-Uphoff said.
The new NSF Institute for Research on Trustworthy AI in Weather, Climate, and Coastal Oceanography is a first-of-its-kind competitive program for the foundation and is being led by the University of Oklahoma.
Colorado State University is joining as a partner in the institute, which comprises a five-year, $20 million program, and is one of five such institutes funded by the NSF nationwide.
Over the next five years, the new AI techniques developed by the institute will provide benefits and advances in forecasting severe weather from days to weeks in advance, coastal changes, and enhanced community resilience in the face of natural disasters.
CSU has developed world-leading expertise in developing AI tools for forecasting, including applications that use satellite remote sensing to improve storm tracking and intensity prediction, as well as AI techniques to understand and visualize climate change.
‘The field already has a lot of tools for forecasting weather across time scales," Barnes said.
"Our work shows that AI methods are yet another great tool for the toolbox, both for forecasting, but also scientific understanding," said Barnes.
With experience in artificial intelligence and machine learning going back to the 1980s, CSU’s expertise has a long history to leverage as well.
Speaking to his own long research experience, Anderson noted: "Our research in AI over 30 years has taught me that the key to collaboration is understanding how to translate AI methods into the language of the discipline.
“In the current collaboration of the CSU team this common ground has and will continue to lead to custom AI methods that reveal novel answers to key questions in environmental science."
Joining with partners at the University at Albany, the University of Washington, North Carolina State University, Texas A&M Corpus Christi, Del Mar College, the National Center for Atmospheric Research, Google, the IBM Weather company, NVIDIA Corporation, Disaster Tech, and NOAA, CSU will assist the University of Oklahoma through providing expertise in development of artificial intelligence algorithms for use in the atmospheric and environmental sciences.
In particular, CSU will lead two "use-case" studies, seeking to better understand tropical storms and prediction of severe weather events at short- and long-term time scales.
The institute will also work to better understand artificial intelligence and how it can be relied upon as a trustworthy tool for environmental science – and to better translate the scientific results.
To do this, it will leverage expertise on the social science aspect of risk communication, creating scientific products that are not only valuable research and forecast tools, but are accessible and meaningful to the public as well.
The ultimate aim is to develop not only better scientific models but also greater understanding and trust in the community.
"We are excited to work on these goals as part of a large national institute that includes researchers from computer science, environmental science, and social science as it takes a large interdisciplinary effort to tackle these complex issues and achieve societal impact,” said Ebert-Uphoff.