Presentation
An Embarrassment of Riches: Thanks to HPC, We Now Have Better Topography for the Ice on Earth than the Land
Presenter
Event Type
Invited Talk
TP
Applications
Computational Physics
Cosmology
Earth Systems
Scientific Computing
Simulation
TimeTuesday, 19 November 201911:15am - 12pm
LocationMile High Ballroom
DescriptionFor years, those of us that made maps of the poles apologized. We apologized for the blank spaces on the maps, we apologized for mountains being in the wrong place and out-of-date information. Over the past 10 years the situation improved. An image mosaic of Antarctica was built, and a constellation of satellites started to stream data at ever higher resolution, at an increasing tempo and even during the long polar winters.
Now a diverse collaboration of US science and intelligence agencies with universities has produced REMA, the Reference Elevation Model of Antarctica, and ArcticDEM, using open source software to extract Digital Elevation Models (DEM) on Blue Waters at a resolution of 2m. The data have an accuracy of a foot and repeat coverage of 90% of the poles an average of 10 times over 6 years. This project was too large for any one agency, university, or company. It required an enormous allocation on Blue Waters, 4 satellites that continuously collected sub-meter optical imagery for 5 years, two satellites that produced ground truth, 100Gbit networking, and petabytes of storage.
We never thought that we would see a time when the science community has better topography for ice than land and better topography for the Transantarctic Mountains than the Rocky Mountains. Even we are having a difficult time understanding what we have made.
We now apologize to the polar science community for a different reason. They have to keep up. And the current DEMs are only the beginning. We now face an avalanche of imagery and derived products in an increasingly complex landscape of small-sats launched by the dozen. It is a complex, exciting time.
Now a diverse collaboration of US science and intelligence agencies with universities has produced REMA, the Reference Elevation Model of Antarctica, and ArcticDEM, using open source software to extract Digital Elevation Models (DEM) on Blue Waters at a resolution of 2m. The data have an accuracy of a foot and repeat coverage of 90% of the poles an average of 10 times over 6 years. This project was too large for any one agency, university, or company. It required an enormous allocation on Blue Waters, 4 satellites that continuously collected sub-meter optical imagery for 5 years, two satellites that produced ground truth, 100Gbit networking, and petabytes of storage.
We never thought that we would see a time when the science community has better topography for ice than land and better topography for the Transantarctic Mountains than the Rocky Mountains. Even we are having a difficult time understanding what we have made.
We now apologize to the polar science community for a different reason. They have to keep up. And the current DEMs are only the beginning. We now face an avalanche of imagery and derived products in an increasingly complex landscape of small-sats launched by the dozen. It is a complex, exciting time.
Presenter