GHub: The new place to be for ice sheet scientists

Glaciologists have teamed up with computer scientists to build an online portal for ice sheet science.

The new science gateway is called GHub, short for “Glaciology Hub,” and it houses datasets, computational tools and educational materials about the science of ice sheets and sea level rise. The site seeks to centralize these resources. It also provides a space for collaboration.

Users can browse datasets; contribute data; run, develop and share computational tools, including ones that leverage supercomputers at the UB Center for Computational Research (CCR); discover resources for teaching and outreach; and connect with one another.

“We want GHub to become the go-to site for resources for anything having to do with the science of ice sheets and sea level rise,” says Jason Briner, an ice sheet researcher and geology professor in the College of Arts and Sciences. “Our goal is for this to become a community space that makes it easier for scientists to collaborate. We want to reduce bottlenecks that are slowing scientific progress on important problems.”

The GHub team includes scientists, engineers and programmers from UB, Tufts University, the National Center for Atmospheric Research and NASA.

The multi-institution effort is funded by over $4.2 million from the U.S. National Science Foundation, with Briner as the project lead.

GHub is maintained by UB CCR and powered by the HUBzero® Platform at the San Diego Supercomputer Center, located at the University of California, San Diego.

For research projects that bring together large groups of ice sheet scientists, “GHub is full of potential,” says Sophie Nowicki, Empire Innovation Professor in the UB Department of Geology and UB RENEW Institute, and a GHub team member. ISMIP6, a major international ice sheet modeling collaboration that Nowicki launched and co-leads, has already leveraged GHub’s capabilities.

“Science gateways are a bit of a democratizer,” says Jeanette Sperhac, a scientific programmer at CCR and GHub team member. “They allow people to access computing and data that they might not otherwise know about or be able to use.”

In a 2020 paper in the journal Concurrency and Computation: Practice and Experience, Sperhac and colleagues at UB, the NASA Goddard Space Flight Center, the NASA Jet Propulsion Laboratory and Tufts outline GHub’s objectives.

“With this new cyberinfrastructure, ice sheet scientists will gain integrated tools to quantify the rate and extent of sea level rise, benefiting human societies around the globe,” the authors write.

They explain that the ice sheet science community includes two broad groups of researchers: those who use fieldwork, satellite observations and other data to learn about the past and current states of ice sheets, and those who develop computational models to forecast the future of ice sheets and sea level rise.

The models used by the second group are validated against data collected by the first. However, “These two groups are not well integrated,” Sperhac and co-authors write. “Better coordination between data collection efforts and modeling efforts is imperative if we are to improve our understanding of ice sheet loss rates.”

And this, says Briner, is the vision for GHub. The team plans to publish what Briner calls “dark data” that’s of interest to modelers. As he explains, “These are data on the Greenland and Antarctic ice sheets that are not archived anywhere public, that are obscured, that are hidden in papers, all formatted differently. We want to put those in a database and make them available for everybody.”

GHub already links out to a variety of datasets and houses others, with the project currently hosting about 15 terabytes of data at CCR for ice sheet research, including for ISMIP6. As Sperhac explains, “Software we run at CCR enables researchers to easily transfer large datasets back and forth. This capability — high-volume, secure data transfer and hosting — is really an important piece of the solution.”

GHub also aims to provide access to computational tools that could serve a range of purposes, such as reformatting large datasets for use in ice sheet models, reformatting model outputs for comparison with observations, and facilitating analysis and visualization of complex datasets. In addition, the GHub team will help users build and share their own computational tools.

Nowicki highlights the value of GHub’s many capabilities, noting that, “We plan to use GHub in ISMIP6 follow-on activities for so many things, from asking participating modelers to check the quality of their submissions prior to upload (thereby catching errors) to developing tools that reproduce figures from community papers (thereby allowing a modeler to see how a new model simulation compares to or alters published results). The possibilities are endless.”

The GHub team includes Briner, Beata Csatho, Nowicki, Kristin Poinar and Anton Schenk in the UB Department of Geology (Nowicki and Poinar are also part of the UB RENEW Institute); Sperhac at UB CCR; Erika Simon at the NASA Goddard Space Flight Center; Abani Patra at Tufts; William Lipscomb and Katherine Thayer-Calder at the National Center for Atmospheric Research; and Eric Larour and Justin Quinn at the NASA Jet Propulsion Laboratory (Quinn is also owner of Tidbit Software).

The GHub steering committee includes Andy Aschwanden at the University of Alaska Fairbanks; Shantenu Jha at Rutgers University; Ian Joughin at the University of Washington; Amanda Leon at the National Snow and Ice Data Center; Bette Otto-Bleisner at the National Center for Atmospheric Research; Joerg Schaefer at the Lamont-Doherty Earth Observatory at Columbia University; Andrew Shepherd at the University of Leeds; and Lev Tarasov at Memorial University.