My research focuses on imaging the details of earthquake ruptures and the structure of magmatic systems using seismic waves generated by natural events and man-made sources.  I’m particularly interested in how this information can be used to assess, mitigate, and predict natural hazards.  Below are brief descriptions of my past and present research.  The earthquake source work shown below was done in collaboration with my graduate school advisor Miaki Ishii.  You can find additional earthquake results on her website.

Imaging the magmatic system under Mount St. Helens (iMUSH)

The iMUSH experiment is a multi-disciplinary study that will utilize active- and passive-source seismic, magnetotelluric, and petrologic data to gain a better understanding of the details of the magmatic system beneath Mount St. Helens.  Alan Levander and I organized the active-source seismic portion of this project, which took place in July-August 2014 and involved the deployment of ~2500 “Texan” Reftek instruments to 5000 locations (red and blue dots in the figure below) over the course of 3 weeks by ~60 volunteers.  This instrument deployment was accompanied by 23 1000-2000 lb. shots (red and blue stars in the figure below) distributed around Mount St. Helens.  Concurrent with the Texan deployment,  a Nodal Seismic deployment organized by Brandon Schmandt from the University of New Mexico set out 900 instruments (magenta dots) along hiking trails near the edifice of the volcano.  Though the data analysis has just begun, the signal to noise ratio of the data seems to be very high and several of the shots were recorded across the entire Texan array.  Over the next few years we hope to use this data to obtain a 3-D image of this magmatic system from the edifice of the volcano down to the Moho.

Hidden aftershocks of the March 11, 2011 Mw 9.1 Tohoku, Japan earthquake

Following the March 11, 2011 Mw 9.1 Tohoku, Japan earthquake, a vigorous aftershock sequence began that included several large events both on the plate interface and in the outer-rise adjacent to the mainshock slip area.  These events were recorded at local stations on Japan, resulting in a high-quality earthquake catalogue of the aftershock sequence.  Many of these events were also recorded at global seismic stations, including the USArray Transportable Array.  Applying a method known as back-projection to this teleseismic array data, we were able to enhance and identify the origins of signals that were produced by events during the first 25 hours of this aftershock sequence.  During this time period, we imaged around 600 events covering an approximate area of 150,000 km².  Surprisingly, when we compare this array earthquake catalogue with the catalogue constructed by the Japan Meteorological Agency (JMA) using local data (black dots in the figure to the right), about half of the events that we detect are not in the JMA catalogue (red dots in the figure to the right).  In addition, most of these events occurred near the Japan trench, and had magnitudes as large as 6.8.  It is possible that the velocity structure of the subduction zone in this region causes defocusing of the seismic waves from near-treach events at Japan.  When combined with the high noise levels associated with aftershock sequences, these low amplitude waves may have been difficult to identify using standard processing methods.

March 11, 2011 Mw 9.1 Tohoku, Japan earthquake

Using high-frequency (0.8-2 Hz) North America seismic data, we image the rupture properties of the 2011 Mw 9.1 Tohoku, Japan earthquake.  The results show that rupture lasted about 240 seconds, and involved large variations in propagation direction and rupture speed (See animation below).  Similar to the 2010 Mw 8.8 Chile earthquake, this event shows considerable frequency-dependent rupture properties.  The figure below shows the locations of energy release during the mainshock using data bandpass-filtered at four different frequency ranges (black dots: 0.8-2 Hz, red dots: 0.25-0.5 Hz, green dots: 0.1-0.2 Hz, blue dots: 0.05-0.1 Hz).  The along-dip frequency dependence of the rupture suggests that the slip rates involved with the near-trench rupture that likely generated the large tsunami associated with this event differ from the slip rates of the deeper rupture which is likely the source of strong ground shaking along the coast of Honshu.  The ability to identify the sources of specific hazards associated with a large earthquake directly following the event can significantly improve hazard mitigation efforts.

February 27, 2010 Mw 8.8 Maule, Chile earthquake

The February 27, 2010 Mw 8.8 Maule, Chile earthquake occurred just north of the massive 1960 Mw 9.5 Chile earthquake and very close to a known seismic gap where the plate interface between the Nazca and South American plates had not ruptured in a large event since 1835.  We investigated the rupture properties of this event using USArray Transportable Array data filtered between 1-5 Hz and 0.5-1 Hz.  The animation below and to the left shows the rupture imaged using high-frequency data (1-5 Hz).  This rupture was bilateral to the north and south, though the strongest source of high-frequency energy came from the dominant northward propagating rupture.  Additional insights into the earthquake source can be gained by comparing rupture properties obtained using data filtered at different frequency ranges (below to the right).  These results show that high-frequency energy precedes, and is down dip of, the lower-frequency energy that follows.  We interpret the high-frequency sources as coming from the rupture front, while lower-frequency sources represent the smoother slip that follows the initiation of the fault.

Intermediate-depth earthquakes

Using data from the High Sensitivity Seismograph Network (Hi-net) in Japan we investigated the rupture properties of intermediate-depth earthquakes from five different regions of deep seismicity (Hindu Kush, Indonesia, Tonga-Kermadec, Vanuatu, and Alaska).  Our results show that these events typically involve sub-horizontal rupture planes, and for the larger events the earthquake source is actually composed of multiple rupture planes.  One example of this composite behavior is shown in the figure to the right where this Mw 7.4 intermediate-depth earthquake in Hindu Kush involves two rupture planes that are separated by 75 km in depth and 8.5 seconds in time.  These results suggest that dynamic triggering of sub-events is common for intermediate-depth earthquakes, and may provide some insight into the mechanism that allows these events to occur at such high pressures and temperatures where brittle failure is not expected.