Migrating Activity of the 2010 Madison Plateau, Yellowstone National Park, Earthquake Swarm: Evidence for Fluid Triggering?

Yellowstone National Park, USA, includes one of Earth's largest volcanic-hydrothermal fields and a youthful caldera. More than half of the seismicity in the volcanic field occurs as earthquake swarms. The 2010 Madison Plateau earthquake swarm, near the northwest rim of Yellowstone caldera, was one of the three largest swarms recorded in the park since monitoring began in the 1970s. The swarm consisted of ~2200 cataloged events (minimum magnitude -0.6), and included 17 events of magnitude 3.0 or larger. The largest event was M 3.9. Low-level seismicity in the area began on January 15, 2010 and abruptly accelerated late on January 17. Seismicity continued at an elevated rate through February 6 before decaying to background levels over the following days. To fully explore the activity and evolution of the swarm, we combine event detection based on waveform cross-correlation (to detect low signal-to-noise events missing from the catalog) with precise double-difference relative relocation of hypocenters. Detection and location goals are accomplished in tandem with cross-correlation, using catalog events as waveform templates that are systematically scanned through the continuous seismic data. Using this procedure, we detected ~8700 events that can be precisely located, a factor of ~4 more events than included in the Yellowstone Seismic Network standard catalog. These newly detected and relocated hypocenters reveal a complex spatial-temporal evolution. The majority of activity is concentrated on a NNW-striking, ENE-dipping structure, with dimensions of approximately 3 by 3 km, between 8 and 11 km depth. Activity began abruptly at about 10 km depth, and expanded episodically and systematically outward (both shallower and deeper) along this structure over the next few days. As time progressed, seismicity began to appear on nearby linear and planar structures of varying orientations, most of which were shallower than the initial activity. We hypothesize that the swarm activity may have been triggered by the rupture of a confined high-pressure fluid system into neighboring pre-existing crustal fractures. Besides the outward expansion of hypocenters in a manner consistent with diffusional fluid flow, we observe that the swarm activity front is usually led by small earthquakes rather than large ones. The front sometimes propagates outwardly in linear "finger-like" structures consistent with pressure increases along highly permeable linear pathways. Interestingly, nodal planes from double-couple-constrained fault solutions are dominantly strike-slip and are oblique to the general trend of hypocenters. This may suggest en-echelon faulting and/or a component of fault opening, perhaps in a fracture mesh geometry similar to that envisioned by Hill [1977]. The primary dipping structure is likely an inherited Basin and Range normal fault, which could provide a permeable pathway of this orientation. This swarm exhibits many similarities with the 1985 Yellowstone swarm just 5 km to the NW, including its migration patterns and dominant orientation (roughly perpendicular to both the minimum regional compressive stress and the caldera boundary). It is also only ~20 km SE from the closest rupture of the deadly M 7.3 1959 Hebgen Lake normal-faulting earthquake.