loading page

Insights into the collapse of Kīlauea caldera using seismicity and infrasound
  • +2
  • Weston Thelen,
  • David Shelly,
  • Gregory Waite,
  • Aaron Wech,
  • Brian Shiro
Weston Thelen
USGS Cascades Volcano Observatory

Corresponding Author:wthelen@usgs.gov

Author Profile
David Shelly
USGS National Earthquake Information Center
Author Profile
Gregory Waite
Michigan Technological University
Author Profile
Aaron Wech
Author Profile
Brian Shiro
USGS Hawaii Volcano Observatory
Author Profile


The 2018 eruption of Kīlauea Volcano included caldera collapse at the summit of the volcano that was well recorded by a network of both permanent and temporary seismometers and infrasound microphones. Volcanic activity prior to the start of the eruption was elevated, including high lava lake levels and increased seismicity. Deflation of the summit began shortly after the eruption in the lower East Rift Zone commenced and was accompanied by a drop in the summit lava lake, which eventually disappeared from view after lowering by several hundred meters. Continued volume loss from beneath the summit eventually led to caldera collapse, which was accompanied by increases in earthquake rates and tremor amplitudes. Infrasonic tremor that originally rose from spattering at the surface of the lava lake was replaced by discrete infrasonic events arising from rockfalls and slumps. On May 16, the first of tens of collapse events occurred, beginning a cycle of increasing earthquake rates and energy release in the presence of deflation. After an initial increase in rate, seismicity rates remained constant for several hours prior to the next collapse event. Earthquakes during these events were typically part of repeating earthquake families and occurred on one of several circumferential cracks that marked active collapsing blocks. Through time, earthquake activity has mirrored the morphology of the collapse, migrating primarily north and east as caldera down drop extended in those directions. Collapse-related infrasound arrivals were initially down, suggesting downdropping of the surface, followed several seconds later by higher frequency infrasound, which we interpret to reflect the explosive or expulsion phase reaching the surface. Infrasound signals were 2-3 times more energetic below 0.5 Hz than above. Especially after May 29, when caldera collapse became much larger in surface area, infrasound signals were highly repetitive with a strong downward first motion. The collapse events were also deficient at seismic frequencies > 0.5 Hz, compared to typical tectonic earthquake sources. Despite similarities in waveforms at low frequencies (<0.1 Hz) and in infrasound, the seismic waveforms at high frequencies were not similar, reflecting either a unique source in each collapse event or a change in location.