Changes in geyser eruption behavior and remotely triggered seismicity in Yellowstone National Park produced by the 2002 M 7.9 Denali fault earthquake, Alaska S. Husen* Department of Geology and Geophysics, University of Utah, Salt Lake City, Utah 84112, USA R. Taylor National Park Service, Yellowstone Center for Resources, Yellowstone National Park, Wyoming 82190, USA R.B. Smith Department of Geology and Geophysics, University of Utah, Salt Lake City, Utah 84112, USA H. Healser National Park Service, Yellowstone Center for Resources, Yellowstone National Park, Wyoming 82190, USA ABSTRACT STUDY AREA Following the 2002 M 7.9 Denali fault earthquake, clear changes in geyser activity and The Yellowstone volcanic field, Wyoming, a series of local earthquake swarms were observed in the Yellowstone National Park area, centered in Yellowstone National Park (here- despite the large distance of 3100 km from the epicenter. Several geysers altered their after called ‘‘Yellowstone’’), is one of the larg- eruption frequency within hours after the arrival of large-amplitude surface waves from est silicic volcanic systems in the world the Denali fault earthquake. In addition, earthquake swarms occurred close to major (Christiansen, 2001; Smith and Siegel, 2000). geyser basins. These swarms were unusual compared to past seismicity in that they oc- Three major caldera-forming eruptions oc- curred simultaneously at different geyser basins. We interpret these observations as being curred within the past 2 m.y., the most recent induced by dynamic stresses associated with the arrival of large-amplitude surface waves. 0.6 m.y. ago. The current Yellowstone caldera We suggest that in a hydrothermal system dynamic stresses can locally alter permeability spans 75 km by 45 km (Fig. 1). Yellowstone by unclogging existing fractures, thereby changing geyser activity. Furthermore, we sug- includes Ͼ10,000 geysers, hot springs, and fu- gest that earthquakes were triggered by the redistribution of hydrothermal fluids and maroles, which are considered the result of hot locally increased pore pressures. Although changes in geyser activity and earthquake trig- water circulating along fracture systems in the gering have been documented elsewhere, here we present evidence for changes in a hy- upper crust and heated by crystallizing magma drothermal system induced by a large-magnitude event at a great distance, and evidence at a depth of 10–15 km (Fournier, 1989). Most for the important role hydrothermal systems play in remotely triggering seismicity. of Yellowstone’s hydrothermal features are lo- cated inside the Yellowstone caldera close to Keywords: Yellowstone National Park, geysers, hydrothermal processes, earthquake swarms. two resurgent domes and along a fault zone extending north from Norris Geyser Basin INTRODUCTION cleation, since elevated fluid pressure will al- (Fig. 1). Hydrothermal systems, including geysers ter the strength and the local stress regime on and hot springs, are regions in Earth’s crust a fault (Hickman et al., 1995). Evidence for HYDROTHERMAL OBSERVATIONS where hot fluids circulate at shallow depths. elevated fluid pressures was, for example, de- Coincident with the arrival of Denali earth- Numerical simulations indicate that geyser tected at the nucleation zone of the M 7.2 quake surface waves, hydrothermal changes eruption intervals are highly sensitive to the 1995 Kobe earthquake, Japan (Zhao et al., were observed at 100 Spring Plain, a hot- intrinsic permeabilities of the geyser’s conduit 1996). Migration and redistribution of fluids spring system at Norris Geyser Basin (S. Stur- and the surrounding rock matrix (Ingebritsen might also trigger earthquake swarms (Waite tevant, 2003, written commun.). Several small and Rojstaczer, 1993, 1996). It has been sug- and Smith, 2002) and drive aftershocks (Nur hot springs, not known to have geysered be- gested that geyser eruption intervals respond and Booker, 1972). Furthermore, remotely fore, suddenly surged into a heavy boil with to small changes in strain induced by atmo- triggered earthquakes occur preferentially in eruptions as high as ϳ1 m. The temperature spheric loading, Earth tides (Rinehart, 1972), geothermal and volcanic areas, suggesting that at one of these springs increased rapidly from and earthquakes (Silver and Valette-Silver, fluids are important in explaining this phe- ϳ42 to 93 ЊC and the pH value increased from 1992). Hot springs and geysers in the Yellow- nomenon (Hill et al., 1993). 2.44 to 5.30 before returning to 3.72 ϳ18 min stone National Park have displayed significant On 3 November 2002, a M 7.9 earthquake later. In the same area, another hot spring that was usually clear showed muddy, turbid water. changes in their eruption patterns following ruptured the Denali fault in central Alaska We monitored eruption intervals of 22 gey- large earthquakes at regional distances (Ͻ200 (Fig. 1). The Denali fault earthquake (hereaf- sers during the winter of 2002–2003 by plac- km) and following intense local earthquake ter Denali earthquake) was one of the largest swarms (Hutchinson, 1985; Marler, 1964; Pitt ing temperature sensors in the runoff chan- strike-slip earthquakes in North America in and Hutchinson, 1982; Watson, 1961). A re- nels.1 Of these 22 instrumented geysers, 8 the past 150 yr (Eberhart-Phillips et al., 2003). cent study (Rojstaczer et al., 2003), however, displayed notable changes in their eruption in- Within hours after the Denali earthquake, we suggested that geysers might be less sensitive tervals, i.e., the changes were larger than the observed significant changes in the eruption to elastic deformation than previously as- intervals of several geysers throughout Yel- sumed. In addition, variations in geyser erup- 1 lowstone, despite the large distance, 3100 km, GSA Data Repository item 2004087, table and tion intervals may also be controlled by inter- graphs of changes in eruption intervals at monitored nal dynamic processes of the geysers from its epicenter. In addition, intense swarms geysers, and cumulative number of earthquakes (Rojstaczer et al., 2003). of local earthquakes occurred close to hydro- close to major geyser basins prior to and after the 2002 Denali fault earthquake, is available at Fluids play a critical role in earthquake nu- thermal systems that experienced the changes in geyser activity. In this paper we report on www.geosociety.org/pubs/ft2004.htm, or on request from [email protected] or Document Secre- *Now at Swiss Seismological Service, ETH these observations and discuss mechanisms tary, GSA, P.O. Box 9140, Boulder, CO 80301- Hoenggerberg, CH 8093 Zurich, Switzerland. that might explain the observations. 9140, USA. ᭧ 2004 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or [email protected]. Geology; June 2004; v. 32; no. 6; p. 537–540; doi: 10.1130/G20381.1; 3 figures; Data Repository item 2004087. 537 Figure 1. Location of re- motely triggered earth- quakes (colored circles) within first 6 h following passage of M 7.9 2002 Denali earthquake sur- face waves. Color of cir- cles is scaled by focal depth and size of circles is scaled with magnitude. Bold red line marks out- line of 0.60 Ma Yellow- stone caldera. Resurgent domes are shown by thin red lines. White line marks outline of Yellow- stone National Park. Ma- jor hydrothermal systems are shown by black stars. Major geyser basins are labeled: YL—Yellowstone Lake; WT—West Thumb Geyser Basin; UP— Upper Geyser Basin; LB—Lower Geyser Ba- sin; NS—Norris Geyser Basin. Inset map shows locations of Denali earth- quake relative to Yellow- stone volcanic field. Sol- id and dashed lines mark -great circle path ؎10؇ tra versed by surface waves directed along strike of Denali fault earthquake. standard error of the intervals prior to the earthquakes clustered close to major hydro- gering of local earthquakes. The latter is not Denali earthquake. Four geysers were too er- thermal systems (Fig. 1). The triggered earth- an unusual phenomenon; remote triggering of ratic to show any effects, and 10 geysers quakes showed swarm-like behavior, cluster- local earthquakes was observed following the showed no significant changes. There was no ing in time and space (Figs. 1 and 3). Many 1992 M 7.4 Landers (Hill et al., 1993) and the common pattern in geyser response to the of the triggered events had magnitudes of Mc 1999 M 7.1 Hector mine (Gomberg et al., Denali earthquake. The most obvious exam- Ͼ 2.0 (Mc denotes a duration magnitude 2001) earthquakes. Also, the 1988 M 7.6 Gulf ple, Daisy Geyser (Fig. 2) in Upper Geyser scaled to the local Richter magnitude scale), of Alaska earthquake triggered swarms of lo- Basin, showed a rapid decrease in the eruption and some were felt by National Park Service cal earthquakes in the Geysers, California, interval following the Denali earthquake, re- employees. Seismicity in the geyser basins geothermal field at a distance similar to that covering close to pre-Denali eruption intervals had returned to pre-Denali earthquake back- from the Denali earthquake to Yellowstone over subsequent weeks. Other geysers such as ground levels 24 h following the passage of (Stark and Davis, 1996). Castle, Plate, and Plume Geysers in Upper the surface waves (Fig. 3). Most of the trig- Previous studies of remote earthquake trig- Geyser Basin, and Pink Geyser in Lower Gey- gered earthquakes close to the hydrothermal gering in volcanic areas favor models where ser Basin, displayed short-term irregularities areas were relatively shallow; their focal magmatic or hydrothermal fluids interact with in their eruption behavior lasting for a few depths were Ͻ5 km (Fig. 1). dynamic stress transients produced by large- days. Lone Pine Geyser in West Thumb Gey- Although earthquake swarms are common amplitude surface waves (Hill et al., 1995, ser Basin, on the other hand, showed a gradual in Yellowstone, the seismicity following the 1993; Johnston et al., 1995; Linde et al., 1994; increase in eruption intervals that peaked three Denali earthquake is unusual for two reasons.