Scientists analyzing the data from seismic networks are becoming better at detecting volcanic activity and at depicting the source and structure of the "plumbing" beneath the world's volcanoes. But a critical question remains: Can these data help predict when a volcano is close to erupting?

In a session at the 2016 Annual Meeting of the Seismological Society of America (SSA) held April 20-22 in Reno, Nevada, researchers will describe how they are using new and repurposed tools to zero in on the sequence of events that precedes a volcanic eruption. The seismologists are looking for patterns of seismic activity to compare with past eruptions to determine when one particular volcano might erupt, as well as larger patterns that could be used to predict when volcanoes of a certain type might erupt.

"In the last ten years, there have been a lot more seismometers placed on volcanoes," said Weston Thelen, a geophysicist at the U.S. Geological Survey. "We're now looking for eruption signals from earthquakes that others might cast off as too small to bother with, but we want to use all the different signals that are out there."

At the SSA meeting, USGS scientist Randall White will present information on a "progression of seismicity" before an eruption that he and others have gleaned from studying more than 35 eruptions at 24 dormant volcanoes over the past 20 years. More than 90% of the eruptions at these dormant volcanoes are preceded by significant (magnitude 3 or larger) volcano-tectonic earthquakes on faults near but not under the volcanoes, they note. Other waves of low-frequency seismicity follow, as magma intrudes into rock and interacts with different parts of the earth's crust.

At Washington State's Mount St. Helens volcano and at Little Sitkin Volcano in Alaska's Western Aleutians, researchers are taking a closer look at the swarms of repeating small earthquakes that appear to precede many eruptions. These swarms can come and go underneath a volcano without being connected to an eruption, however, so seismologists would like to learn more about what each kind of swarm might indicate about the direction and speed of moving magma.

At Mount St. Helens, University of Washington researchers are developing an open-source tool called REDPy (Repeating Earthquake Detector in Python) to look at swarms at the volcano in near real-time, to calculate how the number and size of the earthquake clusters might relate to eruption timing. At Little Sitkin Volcano, Alaska Volcano Observatory seismology Matthew Haney and colleagues are analyzing the seismicity surrounding a 2012 swarm there to learn more about how magma is moving between reservoirs under the volcano.

A 2014 swarm beneath Mammoth Mountain in California allowed USGS scientist David Shelly and his colleagues to trace the complex relationship between fluid movement and fault activation at the heavily monitored lava dome. Their analysis yielded a set of more than 6000 precisely located earthquakes that helped to trace a wave of earthquake propagation along multiple faults during the swarm sequence.

Another presentation in the SSA session will discuss how seismic data were used in 2015 to detect a massive magma chamber under the Yellowstone supervolcano, connecting an earlier-known upper crust magma reservoir with the mantle plume that fuels the supervolcano.

Seismic studies such as the Yellowstone report that help to define the structure of magma chambers and movement are key to connecting seismic activity with the timing of a volcanic eruption, said Thelen. "For many volcanoes, we're still trying to figure out where the magma lies, and where the hydrothermal systems are," he said. "When we understand that better, we can interpret the seismicity better when it comes up."