The U.S. National Science Foundation has announced funding for 25 projects, totaling over $20 million, through the Collaborations in Artificial Intelligence and Geosciences (CAIG) program. This investment aims to advance the development and implementation of innovative AI techniques in geosciences while increasing technical capacity and expanding access to education and training opportunities for using AI approaches in geosciences research.
"The CAIG program presents an exciting opportunity to address big questions in geosciences research while fostering collaborations between geoscientists and AI experts," said Wendy Graham, the director of the NSF Division of Research, Innovation, Synergies, and Education. "All 25 of the CAIG projects will foster transdisciplinary partnerships that lead to innovation in both AI and geosciences. These projects will provide cross-training for AI and geoscience knowledge, significantly building our capacity to study and analyze extreme weather, solar activity and earthquake hazards, among many topics."
The program promotes transdisciplinary collaborations among geoscientists, computer scientists, mathematicians and others to drive transformative discoveries, innovations and solutions. Research teams will adopt various AI techniques - such as generative AI, surrogate models, causal AI and other AI approaches - to advance understanding of complex Earth systems. Topics include increasing efficiency and enhancing utility and equity of geoscience models; improving forecasting, preparation and mitigation for natural hazards; enhancing understanding of earthquake dynamics; improving natural resource management and decision-making in response to climate change; and elucidating the drivers of physical and biological processes of oceans.
The CAIG projects support key technology areas, including AI, cyberinfrastructure (CI) and advanced computing, highlighted in the "CHIPS and Science Act of 2022." Many projects align with the interagency National AI Research Resource Pilot by making AI resources more accessible and providing education and training for geoscientists at all career stages. Several projects will also contribute to the NSF National Discovery Cloud for Climate initiative, a pilot effort to build an integrated national-scale CI capable of supporting end-to-end climate research and education.
The awards are:
-
A Bayesian Inference Framework for Learning Earthquake Cycle Deformation Processes Across Scales via Novel Neural Operators. The University of Texas at Austin, award 2425922.
-
A Large Foundational Model for Earthquake Understanding. Harvard University and Boston University, awards 2425714 and 2425715.
-
Advancing Subsurface Flow and Transport Modeling with Physics-Informed Causal Deep Learning Models. University of Southern California, award 2425919.
-
Advancing Wildfire Science, Prediction, and Management with Machine Learning. University of California Irvine and Spatial Informatics Group, award 2425932.
-
AI-Driven Generation of Vector Magnetograms in the Chromosphere and Photosphere with Application to Explainable Solar Eruption Predictions. New Jersey Institute of Technology, award 2425602.
-
An AI-based Approach to Quantifying and Explaining Uncertainty and Inequity in Geoscience. University of Maryland, College Park, award 2425735.
-
Data Science Frontiers in Advancing Predictive Understanding of Landscapes and Erosional Extremes under Changing Climatic Scenarios. University of Central Florida and University of California Irvine, awards 2425747 and 2425748.
-
Developing AI Emulator Tools for Extreme Events with Application to Heat Waves and Cold Snaps. University of Chicago and New York University, awards 2425898 and 2425899.
-
Discovering the Law of Stress Transfer and Earthquake Dynamics in a Fault Network using a Computational Graph Discovery Approach. University of Southern California and California Institute of Technology, awards 2425908 and 2425909.
-
Empowering Artificial Intelligence to Reveal Phytoplankton Community Dynamics in Coastal Oceans. University of Louisiana at Lafayette and University of Delaware, awards 2425811 and 2425812.
-
Interpretable, Stable, Mass-Conserving AI for Air Pollution Modeling. University of Illinois Urbana-Champaign and Massachusetts Institute of Technology, awards 2425760 and 2425761.
-
Investigating Convection Initiation Dynamics and Predictability with Physically Explainable Deep Learning and Uncertainty Quantification. Penn State University and National Center for Atmospheric Research, awards 2425658 and 2425659.
-
Investigating Tornadogenesis via Explainable Deep Learning. University of Oklahoma, award 2425732.
-
Leveraging AI to Observe and Predict the Drivers of Mixed Layer Heat Inventory Variability. University of California, Santa Cruz, and University of California, Davis, awards 2425905 and 2425906.
-
Navigating the Climate Science Deluge: Training Language Models to Assist in Comprehensive Assessments. Brown University, award 2425380.
-
Next Generation Machine-Learning Approach to Decode High-Resolution Earthquake Catalogs. Carnegie Mellon University and Georgia Institute of Technology, award 2425888 and 2425889.
-
PaleoPAL: An AI Research Assistant for Paleoclimatology. University of Southern California, award 2425885.
-
Physics Informed Graph Neural Network-based Snow Water Equivalent Forecasting. George Mason University, Earth Science Information Partners and University of Washington, awards 2425687 and 2425688.
-
Stability and Physical Consistency of AI-based Climate Emulators for Estimating Forced Responses. University of California, Santa Cruz, award 2425667.
-
Toward a Deeper Understanding of Cloud Processes and Future Storm Modes using AI. Colorado State University, award 2425923.
-
Toward Next-Generation Global Forest Carbon Monitoring via Integrated Sensing, Modeling and AI to Advance Carbon Cycle Science. University of Maryland, College Park, and University of Pittsburgh, awards 2425844 and 2425845.
-
Unifying Landslide Domain Knowledge and AI to Understand Landslide Causality. CUNY City College, award 2425802.
-
Unraveling the Extreme Near-Global Marine Heatwaves of 2023: Using Artificial Intelligence to Understand the Physics and Implications for the Future. Columbia University, award 2425306.
-
Unsupervised Deep Clustering of Seismic Time Series Data at Volcanoes. The University of Texas at El Paso and Colorado State University, awards 2425852 and 2425853.
-
Using Deep Learning to Learn about the Deep Sea: Application of AI to Elucidate Drivers of Global Biogeochemical Cycles. Oregon State University and University of California, Santa Barbara, awards 2425834 and 2425835.