UNIVERSITY PARK, Pa. — In 2018, the side of the Anak Krakatau volcano collapsed in a powerful eruption and produced a tsunami that killed hundreds and injured thousands on nearby Java and Sumatra in Indonesia. A new analysis of satellite data showed the mountainside was slipping for years and accelerated before the eruption — information that could have potentially offered a warning of the collapse.
The team, led by researchers at Penn State, recently published their findings in the journal Geophysical Research Letters.
"Ocean volcanoes, when they are unstable, can collapse catastrophically and generate a tsunami," said Christelle Wauthier, associate professor of geosciences at Penn State and co-author of the study. "And when that happened in 2018, more than 400 people died because nobody had instruments on the ground to know potentially if there was acceleration or change in the deformation behavior. Nobody knew the collapse was imminent. This study shows, unfortunately retrospectively, that we could forecast it if we had this remote sensing data set to get the surface deformation."
Scientists can track ground movements — or surface deformation — using radar satellites sensitive enough to spot changes of just a few inches.
In this work, the team used a remote sensing technique called Interferometric Synthetic Aperture Radar (InSAR), to create highly accurate maps of those changes over time. The researchers analyzed more than a decade of data from three satellites — ALOS-1, COSMO-SkyMED and Sentinel 1 — and used InSAR techniques to map deformation in the lead up to the 2018 Anak Krakatau eruption.
"Overall, the detachment fault experienced approximately 15 meters — roughly 50 feet — of slip from 2006 to 2018 with acceleration and deceleration periods, and a notable acceleration prior to the 2018 collapse," said Young Cheol Kim, a doctoral candidate at Penn State and lead author of the study.
While the InSAR technique is not novel, it is rare to analyze such a large amount of data, the scientists said. The work required access to the high-performance Roar computer cluster managed by the Penn State Institute for Computational and Data Sciences.
"Integrating hundreds of radar images requires a great deal of computational power," Wauthier said. "It's a lot of data storage and data processing and it takes some time and resources."
Still, the technique may show promise for near-real-time monitoring of active ocean volcanoes, especially in locations where other monitoring is not available, the scientists said.
Slip occurs when there is a weakness — or fault — under a volcano. The volcano grows larger as it erupts over time, and eventually it reaches a threshold where there is too much weight for the fault to support, leading to a collapse.
This process may start like a "slow landslide" over years, Wauthier said, but when it begins to accelerate it can be a sign that collapse is imminent.
"The whole chunk of the volcano that collapsed was already moving — like a slow landslide," said Wauthier, who is Kim's adviser. "And, so, it's very important to be able to look at the temporal evolution of that deformation, because if you have an acceleration, it can lead to a collapse. Our data shows that, basically, there was a precursor to the collapse."
The researchers said that other ways to track deformation, like ground-based GPS instruments are often lacking in locations like Anak Krakatau. Because it is an active volcano, there are safety and permitting issues. Ground-based equipment is costly to deploy and maintain and funding is not always available.
"If you have a sudden acceleration of slip, it might be the sign that you will have a collapse happening," Wauthier said. "Whether it's this volcano or others susceptible to collapse worldwide, if you don't have ground-based data in real time, maybe having near-real-time InSAR processing can help researchers be on the lookout for any significant acceleration in slip."
Thomas R. Walter, professor at the GFZ German Research Centre for Geosciences, also contributed to this work.
NASA, the U.S. National Science Foundation and the Federal Ministry of Education and Research of Germany supported researchers involved in this project.
