South SFV Today

As the planet experiences warming, Antarctica's ice sheet is melting and contributing to global sea-level rise. This vast ice reserve holds enough frozen water to raise sea levels by 190 feet worldwide, making accurate predictions about its future movement crucial for coastal protection. Traditional climate models have struggled with this task due to limited data and the complex interplay of oceanic, atmospheric, and surface factors.

In a recent paper published in Science on March 13, researchers from Stanford University employed machine learning to analyze high-resolution remote-sensing data of Antarctic ice movements for the first time. This study provides insights into the fundamental physics governing large-scale movements of the Antarctic ice sheet, potentially improving future predictions.

Ching-Yao Lai, an assistant professor of geophysics at Stanford Doerr School of Sustainability and senior author of the paper, explained: “A vast amount of observational data has become widely available in the satellite age. We combined that extensive observational dataset with physics-informed deep learning to gain new insights about the behavior of ice in its natural environment.”

The research focused on five Antarctic ice shelves—floating platforms extending over oceans from land-based glaciers. Findings revealed that parts closest to the continent are compressed while farther areas are pulled seaward. This causes anisotropy—ice having different physical properties depending on direction.

“Our study uncovers that most of the ice shelf is anisotropic,” said Yongji Wang, first author and postdoctoral researcher at New York University. “The compression zone – the part near the grounded ice – only accounts for less than 5% of the ice shelf. The other 95% is the extension zone and doesn’t follow the same law.”

This work challenges existing assumptions that Antarctic ice shares uniform physical properties across all directions. It suggests current models fail to capture observed satellite movements accurately.

“People thought about this before, but it had never been validated,” Wang stated. “Now, based on this new method and rigorous mathematical thinking behind it, we know that models predicting future evolution should be anisotropic.”

While exact causes for extension zone anisotropy remain unknown, further analysis with additional data is planned. These findings may help understand stresses leading to rifts or calving events when large chunks break away from shelves.

Lai emphasized their methods' broader applicability: “We are trying to show that you can actually use AI to learn something new...this combined approach allowed us to uncover ice physics beyond what was previously known."

Co-authors hail from institutions including University of Otago and MIT; funding came from various sources such as Stanford Doerr Discovery Grant and NASA.