GHub

GHub's crevasse detection workflow uses high performance computing resources at CCR to run a detection algorithm on NASA IceBridge Airborne Topographic Mapper (ATM) data with user-selected parameters. The top pane shows the raw ATM data, the middle shows gridded anomalies, and the bottom pane shows the crevasse features identified by the algorithm.

GHub's crevasse detection workflow uses high performance computing resources at CCR to run a detection algorithm on NASA IceBridge Airborne Topographic Mapper (ATM) data. The top pane shows the raw ATM data, the middle shows gridded anomalies, and the bottom pane shows the crevasse features identified by the algorithm.

Credit: Renette Jones-Ivey, Kristin Poinar, Alek Petty

Published June 3, 2021

GHub: Collaboration and analysis space for ice sheet scientists

Print

Supplemental Material

The GHub science gateway hosts datasets, workflows, and tools to unify ice sheet observations and modeling and improve estimates of future sea level rise. “Sea level rise is a grave concern, making ice melt rates an important area of study,” said GHub Co-Principal Investigator (PI) Kristin Poinar, an assistant professor in Geology at University at Buffalo (UB). “The Greenland Ice Sheet in particular is melting and calving ice at an alarming rate -- the equivalent of all of the water in Lake Erie every two years. This has raised the global sea level by more than one centimeter over the past twenty years.”

Predicting future ice-sheet change requires the efforts of a range of disciplines in ice-sheet science, including those versed in observational data, paleoglaciology ("paleo") data, numerical ice sheet modeling, and emerging methodologies such as machine learning. The GHub project aims to unify researchers from these diverse disciplines and enable them to bring multiple different technologies to bear on this problem. "A significant bottleneck is slowing progress in understanding ice sheets and sea level rise...it relates to a lack of open communication and knowledge accessibility between the ... scientific communities involved,” stated GHub PI Jason Briner, a geology professor at UB. “Ghub is designed to reduce this bottleneck.”

To do so, the GHub science gateway uses High Performance Computing and data storage provided by UB's Center for Computational Research (CCR). GHub's software platform is based on the San Diego Supercomputer Center's HUBzero®, which enables gateway users to develop computational tools using their own scientific codes. To date, the GHub team has developed eight computational tools accessible from the GHub gateway, and hosts the Ice Science Modeling Intercomparison Projects (ISMIP6) dataset, consisting of more than seven terabytes of data.

With over 85 researchers already utilizing GHub to conduct their studies, the team is now working with these users to integrate additional tools with crucial datasets stored at locations like the National Snow and Ice Data Center (NSIDC). These tools will allow for rapid data analysis and ice-sheet model validation, helping scientists catalog the ice sheets of Greenland and Antarctica and how they are changing.

GHub is funded by the US National Science Foundation. The GHub community ran more than 10,000 jobs on the CCR cluster in 2020.

With thanks to Kimberly Mann Bruch at University of California San Diego