U.S. Dept. of Commerce / NOAA / OAR / PMEL / Publications

Rapid dike emplacement leads to eruptions and hydrothermal plume release during seafloor spreading events

R.P. Dziak

Cooperative Institute for Marine Resource Studies, Oregon State University, and Pacific Marine Environmental Laboratory, NOAA, Hatfield Marine Science Center, Newport, Oregon

D.R. Bohnenstiehl

Department of Marine, Earth and Atmospheric Sciences, North Carolina State University, Rayleigh, North Carolina

J.P. Cowen

Department of Oceanography, University of Hawaii, Honolulu, Hawaii

E.T. Baker

Pacific Marine Environment Laboratory, NOAA, Seattle, Washington

K.H. Rubin

Department of Geology and Geophysics, School of Ocean and Earth Science and Technology, University of Hawaii, Honolulu, Hawaii

J.H. Haxel and M.J. Fowler

Cooperative Institute for Marine Resource Studies, Oregon State University, and Pacific Marine Environmental Laboratory, NOAA, Hatfield Marine Science Center, Newport, Oregon

Geology., 35(7), 579-582 (2007)
Copyright 2007 by The Geological Society of America. Further electronic distribution is not allowed.


The creation of ocean crust by rapid injection of magma at mid-ocean ridges can lead to eruptions of lava onto the seafloor and  release of “event plumes,” which are huge volumes of anomalously warm water enriched in reduced chemicals that rise up to 1 km above the seafloor. Here, we use seismic data to show that seafloor eruptions and the release of hydrothermal event plumes correspond to diking episodes with high injection velocities and rapid onset of magma emplacement within the rift zone. These attributes result from high excess magma pressure at the dike source, likely due to a new influx of melt from the mantle. These dynamic magmatic conditions can be detected remotely and may predict the likelihood of event plume release during future seafloor spreading events.

Keywords: dike injection, earthquakes, hydroacoustics, ridge, Juan de Fuca.


In studies of the global mid-ocean-ridge system, one of the most important discoveries during the last 15 yr was the first real-time detection of a dike injection and eruption associated with a seafloor spreading event, and a multidisciplinary effort was rapidly assembled to investigate in situ the many biological, chemical and hydrothermal after-effects on the attendant seafloor and water column ecosystems (Fox et al., 1995; Embley et al., 1995; Baker et al., 1995). The key to rapid in situ investigations is the accurate evaluation of the real-time, remotely detected seismic data indicating that a large-scale seafloor eruption and hydrothermal plume release has occurred. Nevertheless, after nearly two decades and the real-time detection of seven mid-ocean ridge seafloor spreading events, evaluation of the likelihood of a seafloor eruption and release of a hydrothermal plume based on remote seismicity has had a success rate of only ~60%. Moreover, verification has required the application of substantial resources to muster oceanographic vessels and personnel to investigate the eruption sites.  We propose that rapid dike injection followed by high propagation rates during seafloor spreading events leads to the eruption of lavas onto the ocean floor and the release of hydrothermal plumes.  Thus, our findings indicate the rate and timing of earthquake migration may be the most reliable parameters in assessing the likelihood of a seafloor eruption and event plume release during future earthquake swarms.

Hydroacoustic Detection Methods

The first studies of mid-ocean-ridge diking and eruptive events and megaplume release in the Pacific were serendipitous, occurring during ongoing studies on the Juan de Fuca Ridge and East Pacific Rise (Baker et al., 1995; Chadwick et al., 1991; Haymon et al., 1993). Direct, organized responses to mid-ocean-ridge magmatic events, however, require a method for remote real-time detection; the unpredictable and episodic nature of these events also requires broad-area monitoring in order to increase the probability of detecting an event. Since seismicity associated with the intrusion of magma through ocean crust produces low-magnitude earthquakes, the relatively high detection threshold of land-based seismic networks (M > 4) causes magmatic events to be incompletely recorded or missed altogether. In 1991, Pacific Marine Environmental Laboratory, National Oceanic and Atmospheric Administration (NOAA), gained access to the U.S. Navy’s Sound Surveillance System (SOSUS) hydrophone arrays throughout the northeast Pacific (Fig. 1); these arrays enable real-time detection of acoustic tertiary (T) waves generated by seafloor earthquakes and reduce the detection threshold for volcanic seismicity at the Juan de Fuca and Gorda Ridges by almost two orders of magnitude, while they improve location accuracy (Fox et al., 1995).


Figure 1. Juan de Fuca and Gorda Ridge map showing location and date of significant earthquake swarms (seafloor spreading events) detected by SOSUS system. All swarms had either a response cruise that investigated the site or preswarm in situ instrumentation. Inset diagram shows cartoon of event and chronic plumes released from a mid-ocean ridge. F.Z.—fracture zone.

Because SOSUS hydrophones are placed at fixed locations on the seafloor, their area of optimal coverage is predetermined. Fortuitously, the SOSUS geometry is well positioned for excellent monitoring of north Pacific spreading systems. Monte Carlo error simulations (Slack et al., 1999) have been used to estimate earthquake location uncertainties based on T-wave arrivals on the 4–10 azimuthally discrete SOSUS hydrophones. Error estimates range from 0.5 to 3.5 km in latitude and 0.8–2 km in longitude along the Explorer, Juan de Fuca, and Gorda Ridges (at the 68% confidence interval). This uncertainty level is sufficient to lead field parties to sites of active seafloor volcanism (e.g., Embley et al., 1995; Cowen and Baker, 1998).

Juan de Fuca Ridge Dike Injection Events

The use of SOSUS met with immediate success with the first real-time observation of seismicity characteristic of a lateral dike injection at the CoAxial segment of the Juan de Fuca Ridge in June 1993 (Fox et al., 1995; Dziak et al., 1995). Multidisciplinary response cruises over the next several months found and sampled three event plumes and a still-cooling lava flow (e.g., Embley et al., 1995; Baker et al., 1995) and included the discovery of microbial communities living in a subseafloor ecosystem (Holden et al., 1998). Since the CoAxial event, six other uniquely large earthquake swarms have been identified on the Juan de Fuca and Gorda Ridges (Table 1; Fig. 1). Observations of the intensity and spatial patterns of the thousands of earthquakes occurring during these seafloor spreading events (Table 1; Fig. 2) allowed identification of several distinctive characteristics, including: (1) the sequences are swarms, lacking a dominant event, with a temporal history that cannot be described by an Omori Law aftershock decay rate, ~t-p, (2) the total number of earthquakes during swarms exceeds 10% (>350 events/wk) of the variance of long-term (12 yr) background Juan de Fuca-Gorda Ridge seismicity, (3) swarms have several episodes of intense activity that reach 50–100 earthquakes/hr, (4) swarms last from several (>5) days to several weeks, (5) earthquakes at the ridge crest migrate >10 km along axis following the lateral injection of magma through the crust, and (6) swarms may be accompanied by continuous, broadband energy (3–30 Hz) interpreted as “intrusion tremors,” resulting from magma breaking through the crust. Other first-order factors such as geologic setting (ridge versus transform), bathymetric gradient, proximity to chronic venting, or a known magma body also are incorporated into the decision-making process to assess the probability of a magmatic seafloor spreading event. Earthquake swarms meeting these criteria are some of the largest swarms ever recorded on SOSUS and are easily identifiable since they are rare (2–4 yr intervals).


Figure 2. Diagrams of earthquake distribution (distance-time) for seven major Juan de Fuca Ridge (JdFR) and Gorda Ridge swarms recorded on SOSUS. Green arrows highlight seismicity migration associated with dike emplacement during each event. Red arrows show arrival date of research cruises for on-site investigations; Endeavor 1999 is the only event without response cruise (in situ sensors). NA on diagrams indicates periods when SOSUS arrays were offline. Onset time of earthquake migration and migration rate for each event were taken from references of each eruption (Table 1). Migration is assumed to begin with first earthquake to occur at a point greater than maximum location error (3.5 km in latitude and 2 km in longitude) away from beginning earthquake of swarm. Evaluation of migration is then derived from regression line fit to earthquake locations. Migration lines shown in Figure 2 are only examples shown to highlight migration directions and rates estimated in original references.

Despite these attempts at quantifying the characteristics of eruption- and plume-producing earthquake swarms, in situ observations have confirmed that only 4 out of 7 earthquake swarms (Table 1; Fig. 1) designated as volcanogenic were actually associated with seafloor eruptions and the release of a hydrothermal plume. Consequently, an analysis of the swarm seismicity was undertaken to identify relationships that may better predict the likelihood of magmatic activity during future earthquake swarms (see Table 1). All parameters of the event seismicity were compared, and two characteristics were identified as important for recognizing earthquake swarms associated with eruptions and hydrothermal discharge: rapid earthquake migration rate and a relatively short onset time of down rift earthquake migration (Fig. 3). Migration rate is the mean rate at which earthquake locations systematically migrate parallel to the ridge axis, tracing the movement of magma, from the initial swarm location to the point and time where the events cease to move further down rift (Fig. 2). Additionally, events that produce eruptions and plumes typically exhibit a relatively shorter time interval between the onset of earthquake activity at the magma source and the time when the dike is injected and begins to propagate down the rift zone. The onset of rapid propagation of the dike down the rift during all eruption- and hydrothermal plume-producing earthquake swarms resulted in earthquake migration within the first 15 h of the swarm (Fig. 3). The events that initially exhibited high rates of seismic activity, but where the earthquakes did not move out from the initial epicentral area or had exceptionally low earthquake migration rates (Gorda 2001, Middle Valley 2001, and Endeavor 2005), did not produce seafloor eruptions or hydrothermal plumes. Thus the dynamic, migratory behavior of the magmatic activity, as reflected by the earthquake swarm, seems to be a key factor in determining the eventual hydrothermal and volcanic expression of the event.


Table 1. Juan de Fuca Ridge seafloor spreading event parameters.


Figure 3. Relationship of earthquake migration rate (m s-1) to onset of migration (hours after beginning of swarm) for swarms shown in Figure 1. Earthquake swarms associated with seafloor eruptions, fluid-temperature changes, or megaplume events are shown as stars; swarms with no clear magmatic activity or megaplumes are circles. Logarithmic decay curve (solid line) and attendant error were derived from nonlinear regression fit of data; dashed line shows 95% confidence interval (Cl) derived from model. Function derived from analysis is shown in upper right; y is migration, and x is time of migration. Subaerial 1979 Krafla (Iceland) dike injection event (Einarsson and Brandsdottir, 1980) is added for comparison.


Our interpretation of these observations is that higher propagation rates and shorter onset times to migration are both related to a higher excess magma pressure at the dike source, which results from a new influx of melt from the mantle (Rubin, 1995; Jaupart and Tait, 1995). Events with these high-pressure diking characteristics are more likely than others to have manifestations at the seafloor, and likely occur when the shallow-crust magma reservoirs, which are the dike’s source, are fed by melts ascending from a deeper chamber or a mantle reservoir through periodic influxes of melt (Jaupart and Tait, 1995). This new, hotter liquid is accommodated into the finite volume of the preexisting magma reservoir, where it forms a buoyant plume that rises into the shallow region of the reservoir (Jaupart and Tait, 1995). The character of the mixing between the plume and resident liquid will depend on the Reynolds number of the flow (Rubin, 1995). If this plume has a low Reynolds number (<103), it will become laminar, rise to the top of the chamber, and spread out to form a discrete layer, and mixing with the resident melt is limited. Lavas sampled from seafloor eruptions at the Juan de Fuca Ridge do indeed show a compositional heterogeneity reflecting incomplete mixing of magmas (Rubin et al., 2001) consistent with this interpretation. Moreover, flow in Juan de Fuca Ridge basalt dikes is likely to be laminar over the range of injection speeds observed, implying that this compositional heterogeneity will be preserved even as the dike propagates tens of kilometers down the rift zone. For basaltic magmas, flow will be laminar if the Reynolds number (Re) is <103, where Re ≡ [(ρm ū w) / η] where w is the dike thickness and ū is the magma flow velocity. Since Juan de Fuca basalt dikes are relatively thin (w ≤ 2 m) (Chadwick et al., 1991) with magma viscosities (η) in the range of 101–102 kg m–1 s–1 and magma densities (ρm) on the order of 1000 kg m3, the Reynolds numbers for dikes with even the highest propagation rates (Table 1) are ≤101. Geochemical heterogeneity of mid-ocean-ridge lavas also can be due to inefficient shallow mixing of either multiple parental magmas (from a heterogeneous mantle source or slightly different melting conditions at a single source) or of magmas that have experienced variable shallow cooling and differentiation histories, resulting from low thermal stability of the shallow crust due to hydrothermal cooling (Rubin et al., 2001). Nevertheless, once new liquid is accommodated into the finite volume of the preexisting reservoir, the excess magma pressure is increased, ultimately leading to ejection of the liquids from the chamber via a dike injection and possible seafloor eruption (Rubin, 1995). The amount of overpressure at the reservoir, caused by the crustal thickness and the flexural load over the magma chamber also will add to the injection and eruption driving force (Buck et al., 1997).  Indeed, a common eruption-triggering scenario at basaltic volcanoes on land is via the rapid addition of magma into a shallow reservoir, leading to ground deformation and chamber over-pressurization. Although gradual input of basalt magmas and volatile exsolution can lead to increased pressure and diking, the rapid hydrothermal cooling of the crust and relatively small magma chambers should create mid-ocean ridge basalts that are evolved with complex phenocryst zonations, which typically are not observed in submarine lava flows. The regional crustal stress state may also play a role in dictating the distance a dike can propagate and may make it easier for the dike to initiate, although overpressurization of the magma reservoir is still needed for magma to fill the propagating crack as it moves down the rift.  While other scenarios are possible, rapid magma chamber overpressurization is consistent with the petrologic and geophysical observations at mid-ocean-ridge eruptions, and this is a plausible mechanism for eruptive dike initiation from mid-ocean-ridge melt reservoirs. 

The past 15 yr of earthquake swarm detection and the subsequent multidisciplinary in situ response efforts have greatly increased our understanding of the seismic character and overall impact of mid-ocean-ridge diking events. It seems apparent that rapid dike propagation events are more likely to be associated with seafloor eruptions and dramatic, transient changes in hydrothermal circulation and discharge. Our results suggest that earthquake migration rate and timing criteria may now be viewed as reliable parameters for assessing the likelihood of a seafloor eruption and hydrothermal plume release during future earthquake swarms.

Acknowledgments. This project would not have been possible without the hard work of our colleagues: D. Butterfield, B. Chadwick, E. Davis, B. Embley, D. Fornari, T. Garfield, B. Glazer, S. Giovanoni, H.P. Johnson, D. Kadko, J. Lupton, B. Lavelle, M. Lilley, S. Merle, and J. Resing. We also thank S. Carbotte and an anonymous reviewer for very helpful comments. The SOSUS project benefited from the dedication of C. Fox, S. Hammond, T.-K. Lau, H. Matsumoto, and J. Klay. We thank the National Oceanic and Atmospheric Administration (NOAA) Vents Program, the National Science Foundation (NSF) Ocean Sciences Program (grant OCE-0623823), the Ridge 2000 Program, and the University of Hawaii NASA Astrobiology Institute for support. This is Pacific Marine Environmental Laboratory contribution 2986 and School of Ocean and Earth Science and Technology contribution 7062.



Baker, E.T., Massoth, G.J., Feely, R.A., Embley, R.W., and Thomson, R.E., 1995, Hydrothermal event plumes from the CoAxial seafloor eruption site, Juan de Fuca Ridge: Geophysical Research Letters, v. 22, p. 147–150, doi: 10.1029/94GL02403.

Baker, E.T., Fox, C.G., and Cowen, J.P., 1999, In situ observations of the onset of hydrothermal discharge during the 1998 submarine eruption of Axial Volcano, Juan de Fuca Ridge: Geophysical Research Letters, v. 26, p. 3445–3448, doi: 10.1029/1999GL002331.

Bohnenstiehl, D.R., Dziak, R.P., Tolstoy, M., and Fox, C.G., 2004, Temporal and spatial history of the 1999–2000 Endeavour segment seismic series, Juan de Fuca Ridge: Geochemistry, Geophysics, Geosystems, v. 5, doi: 10.1029/2004GC000735.

Buck, W.R., Carbotte, S.M., and Mutter, C., 1997, Controls on extrusion at mid-ocean ridges: Geology, v. 25, p. 935–938, doi: 10.1130/0091-7613(1997)025<0935:COEAMO>2.3.CO;2.

Chadwick, W.W., Embley, R.W., and Fox, C.G., 1991, Evidence for volcanic eruption on the southern Juan de Fuca Ridge between 1981 and 1987: Nature, v. 350, p. 416–418, doi: 10.1038/350416a0.

Cowen, J.P., and Baker, E.T., 1998, The 1996 Gorda Ridge event detection and response activities: historical perspective and future implications: Deep Sea Research, v. 45, p. 2503–2511, doi: 10.1016/S0967-0645(98)00080-0.

Davis, E.E., Becker, K., Dziak, R., Cassady, J., and Wang, K., 2004, Hydrological responses to a seafloor spreading episode on the Juan de Fuca Ridge: Nature, v. 430, p. 335–338.

Dziak, R.P., and Fox, C.G., 1999, The January 1998 earthquake swarm at Axial Volcano, Juan de Fuca Ridge: Hydroacoustic evidenceof seafloor volcanic activity: Geophysical Research Letters, v. 26, p. 3429–3432, doi: 10.1029/1999GL002332.

Dziak, R.P., Fox, C.G., and Schreiner, A.E., 1995, The June–July 1993 seismoacoustic event at CoAxial Segmen, Juan de Fuca Ridge: Evidence for a lateral dike injection: Geophysical Research Letters, v. 22, p. 135–139, doi: 10.1029/94GL01857.

Einarsson, P., and Brandsdottir, B., 1980, Seismological evidence for lateral magma intrusion during the July 1978 deflation of the Krafla Volcano in NE Iceland: Journal of Geophysical Research, v. 47, p. 160–165.

Embley, R.W., Chadwick, W.W., Jonasson, I.R., Butterfield, D.A., and Baker, E.T., 1995, Initial results of the rapid response to the 1993 CoAxial event: Relationships between hydrothermal and volcanic processes: Geophysical Research Letters, v. 22, p. 143–146, doi: 10.1029/94GL02281.

Fox, C.G., and Dziak, R.P., 1998, Hydroacoustic detection of volcanic activity on the Gorda Ridge, February-March 1996: Deep Sea Research, v. 45, p. 2513–2530, doi: 10.1016/S0967-0645(98)00081-2.

Fox, C.G., and the Event Detection and Response Team, 2001, Detection and Response to a seafloor spreading episode on the central Gorda Ridge, April 2001, Eos (Transactions, American Geophysical Union) v. 82, no. 47, Fall Meeting Supplement, abstract S22C–05.

Fox, C.G., Radford, E., Dziak, R., Lau, T.-K., Matsumoto, H., and Schreiner, A., 1995, Acoustic detection of a seafloor spreading episode on the Juan de Fuca Ridge using military hydrophone arrays: Geophysical Research Letters, v. 22, p. 131–134, doi: 10.1029/94GL02059.

Haymon, R., Fornari, D., Von Damm, K., Lilley, M., Perfit, M., Edmond, J., Shanks, W., Lutz, R., Grebmeier, J., Carbotte, S., Wright, D., McLaughlin, E., Smith, M., Beedle, N., and Olson, E., 1993, Volcanic eruption of the mid-ocean ridge along the East Pacific Rise at 9º45–52’N: Direct submersible observation of seafloor phenomena associated with an eruption event in April: Earth and Planetary Science Letters, v. 119, p. 85–101, doi: 10.1016/0012-821X(93)90008-W.

Holden, J.F., Summit, M., and Baross, J.A., 1998, Thermophylic and hyperthermophylic microorganisms in 3–30ºC hydrothermal fluids following a deep-sea volcanic eruption: FEMS Microbiology Ecology, v. 25, p. 33–41.

Jaupart, C., and Tait, S., 1995, Dynamics of differentiation in magma reservoirs: Journal of Geophysical Research, v. 100, p. 17,615–17,636, doi: 10.1029/95JB01239.

Rubin, A.M., 1995, Propagation of magma-filled cracks: Annual Review of Earth and Planetary Sciences, v. 23, p. 287–336, doi: 10.1146/annurev.ea.23.050195.001443.

Rubin, K.H., Smith, M.C., Bergmanis, E.C., Perfit, M.R., Sinton, J.M., and Batiza, R., 2001, Magmatic history and volcanological insights from individual lava flows erupted on the sea floor: Earth and Planetary Science Letters, v. 188, p. 349–367, doi: 10.1016/S0012-821X(01)00339-9.

Slack, P.S., Fox, C.G., and Dziak, R.P., 1999, P wave detection thresholds, Pn velocity estimates, and T wave location uncertainty from oceanic hydrophones: Journal of Geophysical Research, v. 104, p. 13,061–13,072, doi: 10.1029/1999JB900112.

Return to Abstract

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage