UMN key player in new NSF Data Science Institute on polar regions
CS&E researchers will lead the design of a new generation of spatial data science methods to understand atmospheric drivers and ice dynamics in polar regions
Tens of millions of people live in areas that are at risk for flooding due to climate change, sea-level rise, and melting of glaciers. In fact, the Intergovernmental Panel on Climate Change (IPCC) estimates that sea levels could increase by 26–98 cm by the end of this century. This large range can be partly attributed to an incomplete understanding of fast-flowing regions of the Greenland and Antarctic ice sheets.
To help populations prepare for and respond to these risks, the National Science Foundation (NSF) is funding a new $13 million data science center through their Harnessing the Data Revolution (HDR) Big Idea program. The University of Minnesota is one of the key collaborators on the grant.
Led by the University of Maryland Baltimore County (UMBC), the new Institute for Harnessing Data and Model Revolution in the Polar Regions (iHARP) will serve as a research hub where experts in data science, Arctic and Antarctic science, and cyberinfrastructure—from academia, government, and private sectors—come together to address national priorities and challenges related to climate change, sea-level rise, and the rapidly changing Arctic.
“It is so exciting to be selected as one of the five HDR institutes in the nation, however, this comes with huge responsibility,” said UMBC's Maryam Rahnemoonfar, the principal investigator of the five-year grant. “We are the first data science and machine learning institute in the world that is dedicated to research in polar regions.”
The institute will carry out about a dozen research projects grouped into four focus areas: (a) integrate heterogeneous, noisy, and discontinuous data in space and time; (b) assimilate data into numerical and geo-physical models; (c) develop spatial-temporal methods to forecast the changes in the Arctic and Antarctic, and (d) build scalable algorithms to apply the solution methods at planetary scale.
Furthermore, the institute will design, develop curricula, and offer hands-on community workshops, lecture series, conference tutorials, and training, in partnership with related communities like EarthCube, IS-GEO Research Coordination Network and South Big Data Hub and their international counterparts such as the University of the Arctic consortium. It will also contribute to K-12 education and informal learning via existing local and national platforms such as the Earthrise Media for citizen science education.
In addition to the University of Minnesota and UMBC, other participating academic institutions include the University of Alaska Fairbanks, the University of Colorado Boulder, the University of Texas at Austin, Amherst College, Bowie State University, and Dartmouth College. The NASA Universities Space Research Association, the NASA Jet Propulsion Laboratory, NVIDIA, IBM, and Amazon will also be involved in the research.
Driven to discover at Minnesota
The University of Minnesota researchers will lead the design of a new generation of spatial data science methods for century-scale ice-sheet forecasts and spatial-temporal pattern mining to characterize relationships between ice-surface melts and sub-surface and atmospheric drivers.
Shashi Shekhar, a Distinguished McKnight University Professor and a Distinguished University Teaching Professor, is one of the co-principal investigators and part of the institute's leadership team. He will co-lead one of the four research focus areas spanning multiple projects as well as coordinate with other HDR institutes.
Shekhar will also lead the project on spatial-temporal pattern mining for data-driven understanding of atmospheric drivers and ice dynamics. This project will develop novel spatial data science methods for identifying patterns such as hotspots of recurring surface melts along with their relationships with events and processes in the ice-sheet and atmosphere.
Professor Mohamed Mokbel will lead a project on combining and learning from multimodal measurements from radar, laser altimetry, and gravimetry to produce high-resolution spatio-temporal data needed for the ice-sheet models.
This project builds on ongoing NSF-sponsored spatial data science research projects at the University of Minnesota, including:
- Polar Geospatial Center
- Midwest Big Data Hub: Building Communities to Harness the Data Revolution
- Investigating Spatio-temporal Informatics for Transportation Science
- Spatiotemporal Big Data Analysis to Understand COVID-19 Effects
- Adopting Machine Learning Techniques for Big Spatial and Spatio-temporal Data and Applications
- Research Infrastructure for Big Spatial and Temporal Data
- Indexing, Querying, and Visualizing Big Spatial and Spatio-temporal Data