ASU-SFI Biosocial Complex Systems Fellow Stefani Crabtree and SFI External Professor Devin White have been awarded one of three HPC Innovation Excellence Awards from Hyperion Research’s High Performance Computing (HPC) User Forum. The award honors researchers who have delivered “HPC-supported, real-world achievements with significant potential for benefitting humanity.”
In a study published in Nature Human Behavior last year, Crabtree, White, and their co-authors used supercomputers to help identify ancient human highways in Sahul, the landmass that consisted of Australia, New Guinea, and Tasmania approximately 70,000 years ago.
To generate their virtual highway, they began by creating a detailed digital map of Sahul that gave them pixelated topography. Then they simulated the possible paths — a whopping 125 billion — that a hypothetical female could take across this landscape. How would she choose her way? To figure this out, the team weighted the pixels of the landscape for features like elevation, proximity to water, and prominent landmarks — features that, they theorized, bear significantly on migration routes. With the help of a supercomputer, and new statistical techniques, they were able to sort optimal paths from possible paths and arrive at a clear set of likely routes.
After they compiled a central database of the oldest archaeological sites in Australia and New Guinea and compared them to their virtual migratory pathways, they found that the fit was stronger than they had anticipated. With this result, they recognized that their new methods could help solve longstanding challenges in archeology. As White says, “our methods allow us to measure how well our models fit with archeological sites, which we know people traveled between, but where we might not be able to see physical routes anymore.” Because of the refined statistical tools they have created, their work may help others find previously undiscovered archeological sites, and illuminate future pathways, too.
“Our work shows that archeology can save the world,” says Crabtree, and she means it: This work can help researchers identify migration paths people might take when they flee from natural disasters or conflict zones, for example. “Think of climate refugees,” says Crabtree. “We are talking about people who are going to be walking away from flooding coastlines, who might be running into dangerous situations—these kinds of approaches can help us predict where people may choose to go, which then can be helpful for offering aid.”
Crabtree and White have made their methods and data available through their paper, and hope others will use them. They’re enthusiastic to set in motion a new era of archeology that uses supercomputers to illuminate not only highways of the past but human migration patterns of the future.