Penn State Models Predict Volcano Flank Collapse — A Tool to Prevent Deadly Tsunamis

Rising magma beneath a recently erupted volcano can destabilize its slopes and trigger flank collapse — a process that can in turn generate catastrophic tsunamis. Researchers at Penn State have developed predictive stability models that estimate how volcano flanks respond to rising magma and identify where collapse is most likely.

Christelle Wauthier, an associate professor in Penn State's Department of Geosciences and hub director for Computational and Data Sciences, led the study published in the Journal of Geophysical Research: Solid Earth. Using analyses of past slope failures, the team created a method that simulates a range of subsurface conditions and forecasts which parts of a volcano are most susceptible to collapse.

The models indicate that flanks resting above shallowly dipping faults in the subsurface are at higher risk of failure. These model outputs can guide targeted monitoring and preparedness efforts so authorities know where to focus limited resources.

"If you have an idea of which area of the volcano is more susceptible to collapse, you could place ground-based sensors such as seismometers or GPS to monitor a risky flank on a minute-to-minute or hour-to-hour basis well before a collapse happens," Wauthier said.

A flank collapse can devastate nearby coastal communities directly through landsliding and rockfall, but the most deadly consequence is often the tsunami generated when large volumes of material enter the ocean. A historical example is the 1883 eruption and collapse at Anak Krakatau in Indonesia: a flank failure triggered a tsunami that accounted for the vast majority of the estimated fatalities from that event, according to the U.S. National Oceanic and Atmospheric Administration.

Beyond acute hazards, scientists are investigating how long-term climate shifts might influence volcanic behavior and flank stability. Some analyses suggest that changes in precipitation, glacier retreat, and sea-level variations could alter stress conditions on volcano flanks, potentially affecting collapse frequency in certain regions.

Wauthier and collaborators plan further work to refine the models, incorporate more field data, and validate predictions with instrument networks. When combined with strategically placed seismometers, GPS stations, and other sensors, these models could become part of operational early-warning systems that give coastal communities more time to evacuate and prepare.

Implications for communities: Local authorities can use the model outputs to prioritize monitoring, run targeted emergency drills, and define evacuation zones. Early identification of a high-risk flank allows continuous monitoring and faster, more informed decisions that could save lives.