UW Information Technology

Groundbreaking cloud computing at UW measures how glaciers respond to climate change — and how to prepare the millions who rely on their life-supporting waters

David Shean, an assistant professor in Civil & Environmental Engineering, has become a leader in his field, developing novel cloud computing tools to build better 3-D maps to show how Earth’s landscape, including glaciers, are changing over time.

By Ignacio Lobos

David Shean could have spent his entire career happily studying Mars’ cryosphere — a portion of the red planet locked in ice.

But when it came time to determine a future direction, Shean knew there were greater, more immediate challenges right here on Earth, where the lives of millions of people are dependent on glaciers. Ice took him millions of miles away, and ice brought him to the UW.

David Shean

David Shean
Assistant Professor, Department of Civil & Environmental Engineering

Today, Shean, an assistant professor in the UW Department of Civil & Environmental Engineering, uses cloud computing to understand how climate change is reshaping the Earth’s cryosphere — polar ice sheets, glaciers and seasonal snow. Glacial retreat is already having profound consequences, and Shean hopes his work will help communities plan better.

That work is already having an impact. Shean pioneered an entirely new way to quantify glacial retreat more accurately, deepening our understanding of the dramatic changes underway across the globe. By relying on data from modern high-resolution imaging satellites and historical aerial and spy satellite photos from the 1960s through the 1990s, he’s letting the world see glaciers in a whole new way.

A satellite captures two images of the same area from two slightly different angles. Later, David Shean uses software to create a highly accurate elevation map to help measure glacial loss over time.

Here in the Pacific Northwest, glaciers cover vast areas of Mt. Rainier, Mt. Baker and the North Cascades to name a few — fresh-water reservoirs that provide up to 10 percent of total late summer river flow that people rely on for drinking water, farming and hydropower. In China and India, glaciers sustain hundreds of millions of people. As they retreat, glaciers reshape the landscape, and create potentially dangerous landslides or floods that could affect people downstream.

“Billions of people across the world could be impacted by changes in the cryosphere,” Shean said. “Our work and the work of other researchers seek to answer how best to prepare for these changes.”

Cloud computing plays key role in understanding glacier change

But studying global glacier change involves processing and analyzing massive amounts of data, which require robust computational resources. So Shean turned to cloud computing, building cloud computing tools and applying new data science methods that are now used by colleagues across the world.

“His work is a rising tide that benefits the vast community of earth scientists whose research depends on precision models of terrain elevation,” said Rob Fatland, director of Cloud & Data Solutions in UW-IT’s Research Computing & Strategy division. Fatland collaborates with Shean to bring cloud computing to his research and to the classroom, where Shean trains the latest generation of scientists and engineers.

In one of his most novel approaches, Shean helped develop automated methods to extract high-resolution terrain data from satellite photos to create extremely accurate 3-D maps that precisely capture surface elevation changes over time.

Mt. Rainier’s glaciers serve as a benchmark to measure ice losses. David Shean put this map together using USGS maps and recent satellite data. It shows that glacial surface elevations have dropped more than 130 feet in some places between 1970 and 2016.

“A modern Earth observation satellite produces a fire hose of information,” Shean said. “You can’t just download these images to a desktop computer. We are talking about hundreds of petabytes of digital information, with a typical photo about 42,000 pixels by 330,000 pixels. That’s enormous.”

One of his earlier projects at the UW involved using satellites to take thousands of images to measure change for all 1,000 glaciers across the Western U.S., starting in 2012.

The result is the first comprehensive view of glaciers across the region, published last year by Shean and collaborators. They concluded that glaciers lost 117 gigatons of ice over the past two decades, enough to cover the entire state of Washington in two feet of water.

It could have been worse. The study found that a relatively snowy period from 2009-2018 replenished many of the glaciers, dampening their overall losses. At Mt. Rainier, large weather systems off the Pacific curbed glacial losses, with some gaining mass.

Creating a digital glacier time machine

Glaciers have a long response time. Decades can pass before a glacier can adjust to a change in climate. Traditionally, researchers have driven stakes into ice and snow to measure benchmark glaciers since the late 1950s. There have been aerial and satellite surveys, but no single method could measure glacial mass loss on a regional scale.

But Shean and colleagues from the Department of Earth & Space Sciences may have found a way. They are turning to previously untapped sources of data — recently declassified spy satellite images from the 1960s through the 1990s and high-quality aerial photographs from the United States Geological Survey (USGS) from the 1960s-1980s.

Arial view of mountain that shows elevation of terrain.

David Shean has found a new life for formerly classified and highly-detailed spy satellite images: Use cloud computing tools to build extremely accurate elevation maps to measure terrain changes over time. This photo shows Mount St. Helens about two years after its May 18, 1980 eruption.

The images, coupled to modern data, could lead to precise terrain models across many decades. When they applied for a 2018 UW Innovation Award, they called their project a “digital glacier time machine.”

“It is not enough to look at data that only goes back 18 years if you want to understand glacial changes and develop strategic, long-term water management solutions,” Shean said.

The team, which includes a USGS collaborator, got the award and $300,000 for two years, to develop automated, open-source, cloud-based workflows capable of processing these historical aerial and declassified satellite images and turning them into usable terrain models. The project is now fully under way.

“This is fantastic data, but it’s going to take a lot of work — and computing power — to make it usable,” said Shean, who wants to expand the project and produce long-term records for glaciers around the world, including in Asia, where hundreds of millions of people depend on their waters.

“When you combine all of these photos and terrain models together, they allow us to see the land in a whole new way. That’s why we call it a time machine,” Shean said. “We can see changes in the recent past, and that can help us see the future.”

Learn more: Visit IT Connect to learn more about cloud computing at the UW; watch a 2018 Geohackweek presentation on raster tools by David Shean.