Recent eruptions on the CoAxial segment of the Juan de Fuca Ridge: Implications for mid-ocean ridge accretion processes

R. W. Embley,¹ W. W. Chadwick,² M. R. Perfit,³ M. C. Smith,³ and J. R. Delaney⁴

¹Pacific Marine Environmental Laboratory, National Oceanic and Atmospheric Administration, Hatfield Marine Science Center, Newport, Oregon 97365
²Cooperative Institute for Marine Resources Studies, Oregon State University, Hatfield Marine Science Center, Newport, Oregon 97365
³Department of Geology, University of Florida, Gainesville, Florida, 32611
⁴School of Oceanography, University of Washington, Seattle, WA 98195

Journal of Geophysical Research, 105(B7), 16,501–16,525 (2000).
Copyright ©2000 by the American Geophysical Union. Further electronic distribution is not allowed.

10. Discussion

10.1. Relationship of T Wave Seismicity to Seafloor Structure and Geologic Observations

Where did the 1993 dike originate and how does the spatial pattern of the 1993 T wave swarm relate to the physical manifestations of the 1993 dike intrusion on the seafloor? A puzzling problem has been the discrepancy between the trend of the 1993 T wave swarm and the trend of the neovolcanic zone at CoAxial as defined by geologic mapping. We think that the 1993 dike intruded from a magma source beneath the southern CoAxial segment and propagated northward along the CoAxial neovolcanic zone and that Axial Volcano was not involved in the 1993 event, for the following reasons:

1. Hydrothermal plumes related to the 1993 dike intrusion were located over the neovolcanic zone of CoAxial segment, not along the trend of T wave epicenters [Baker et al., 1998].
2. Time series measurements of hydrothermal plumes at both the Flow and Floc sites show a power law decline of temperature and rise height, consistent with the expected thermal decline during the cooling of a dike [Baker et al., 1998; Cherkaoui et al., 1997].
3. Venting sites on the seafloor were located directly under the hydrothermal and exclusively along the neovolcanic zone of CoAxial. The evolution of the vent sites on the seafloor at the Flow and Floc sites, including a rapid bacterial bloom followed by rapid cooling of the system, is consistent with a geothermal response to an intruded dike. Geothermal perturbations associated with dike intrusions in volcanic rift zones on land generally appear directly over the axis of the dike [Björnsson et al., 1979].
4. SeaBeam bathymetry of the central JdFR is consistent with Axial Volcano and its rift zones effectively behaving as one of many independent spreading segments along the ridge, with overlapping (but not intersecting) relationships with the adjacent ridge segments (Vance and CoAxial).
5. Side-scan imagery and bottom photography delineate the neovolcanic zones of AVNRZ and CoAxial segment and show that they are separate from one another. The northern limit of the AVNRZ only extends to 46°15–18N, where it merges with the western fault block ridge, an older tectonized feature hypothesized to have formed along CoAxial. There is no evidence from geologic mapping that dikes have propagated in the past from the AVNRZ into and along the neovolcanic zone of CoAxial. Although dikes intrude at variable directions over a period of time on a given rift zone [Gudmundsson, 1990], it is difficult to envision how a dike could propagate across structure from one volcanic rift zone to another and then along structure on the second rift zone.
6. An ocean bottom seismometer deployment made around the Floc site detected small earthquakes tightly clustered at the Floc site along the trend of the neovolcanic zone [Sohn et al., 1998].
7. Geochemistry of lavas from Axial Volcano and its rift zones is distinct from lavas collected from CoAxial segment, particularly with relation to Sr isotopic signatures. Geochemical signatures of the 1993 lava flow and the two 1981–1991 flows indicate that they all came from CoAxial magma sources, not from Axial Volcano.
8. Detailed geologic mapping of the low-temperature hydrothermal vent sites on the seafloor associated with the 1993 intrusion shows that the vents are localized along fissures and small grabens which are interpreted to have formed or were reactivated directly over the 1993 dike while it was intruding. These structures which trace the path of the dike are located within the CoAxial neovolcanic zone.

Therefore all of the geologic and geochemical evidence leads to the conclusion that the 1993 dike originated within the CoAxial segment and intruded northward along its neovolcanic zone and did not follow the path suggested by the T wave epicenters from AVNRZ to CoAxial. What could explain the oblique trend of the T wave swarm, particularly since the T wave epicenters cluster perfectly above the 1993 eruption site at the northern end of the swarm? There are two hypotheses that could explain this discrepancy. One hypothesis, suggested by Fox et al. [1995], is that the first half of the T wave swarm consisted of deeper earthquakes [Schreiner et al., 1995] and the relatively shallow seafloor to the west of the central CoAxial segment may have radiated more of the acoustic energy and so added a systematic location error to the epicenters in the southern part of the swarm. A second hypothesis is that the early T wave events might have been associated with faulting along the boundary of the western fault block ridge rather than directly associated with the dike itself, located on-axis. Intruding dikes have triggered normal faulting in volcanic rift zones on land [Rubin, 1992; Sigurdsson, 1980], and several normal fault earthquakes located on the bounding faults of the summit caldera of Axial Volcano were triggered during the early stages of the 1998 dike injection there [Dziak and Fox, 1998, 1999].

10.2. Implications for Stress Release During Accretion Events on the Mid-Ocean Ridge

Three separate eruptions have occurred on the CoAxial segment within the 12-year period from 1981 until 1993 (Plate 1). The two younger eruptions occurred near the northern end of the segment, and the oldest event occurred at the midpoint of the segment at 46°15–21N.

How much strain has been released during this time period and what are the implications for the space-time pattern of strain and volcanism for the rest of the ridge system? It seems reasonable to assume that the two 1981–1991 eruptions represent diking events similar to the 1993 event because since (1) all the lava flows are similar in volume and are elongated along the axis of the neovolcanic zone, and (2) all of the other known historical eruptions documented to date along the northeast Pacific spreading centers [Chadwick et al., 1998; Dziak and Fox, 1998; Dziak et al., 1995; Embley and Chadwick, 1994] have been associated with dikes that have intruded over tens of kilometers along axis. The T wave epicenters suggest that the 1993 dike was injected from a latitude between the Floc and Source sites (~46°12N), so presumably a magma reservoir exists beneath the axis of CoAxial around this location. If we assume that all three recent eruptions were fed by dikes that intruded northward from this area, this may have interesting implications for the pattern of stress accumulation and release from this section of the ridge. At a spreading rate of 5.5 cm/yr, each meter thickness of intruded dike relieves about 18 years of accumulated stress. Although we cannot directly estimate the thickness of any of the dikes intruded during the CoAxial events, estimates of dike widths from sections of oceanic crust exposed in ophiolite sections are typically 0.5–1.5 m [Gillis, 1995]. The three documented eruptions occurring over a 12-year period represent only a minimum number of diking events because additional events may have taken place without producing eruptions in the 1981–1991 period prior to SOSUS monitoring and after the initial SeaBeam survey. Also, the lack of teleseismic events (M  4) along this portion of the ridge (R. Dziak, personal commununication, 1999) in the 1966–93 time period suggests that faulting was not a significant factor in stress release for this period. Nevertheless, the record of recent eruptions at the CoAxial segment represents the first direct evidence of the style of stress release over the timescale of a about a decade along a portion of the MOR. The CoAxial diking events of 1981–1993 must have relieved a significant amount of stress along the segment. No diking events have been detected by SOSUS in the 1993–1999 period.

Although the period of observation is very short, this pattern could be analogous to the behavior of Icelandic central volcanos. The long-term record of volcanism in Iceland shows that the central volcanos release accumulated stress in a series of major rifting/volcanic events about every 100–150 years [Björnsson et al., 1979]. During the latest of these episodes, the Krafla Volcano produced 20 diking/rifting events between 1975 and 1984, only about half of which erupted lava at the surface [Björnsson, 1985]. Whether the CoAxial "rifting episode" has ended remains to be seen, although at the time of this publication, the segment has been seismically quiet for 6 years. Geodetic monitoring of the Krafla episode suggests that as each diking event relieves some of the regional stress, it becomes more difficult to inject additional dikes into an increasingly compressive stress field [Ewert et al., 1991].

10.3. Implications for Magma Delivery Systems at Intermediate Spreading Rates

How magma is delivered along segments of the MOR is poorly known because of the lack of direct measurements of 3-D seismicity patterns and deformation during diking events. Models of magma delivery systems have been mostly inferred from surface morphology, lava geochemistry, and geophysical measurements [Batiza, 1996]. Even in Iceland, where a rifting episode was monitored in 1975–1984 at the Krafla Volcano and where there has been extensive geologic mapping, there remains some controversy as to how magma is delivered to the upper crust. The seismicity patterns during the Krafla rifting events showed that at least some of the dikes were propagated laterally from beneath Krafla caldera [Einarsson and Brandsdottir, 1980] and this led Sigurdsson [1987] to propose that lateral propagation of dikes is an important process on Icelandic central volcanos. On the other hand, Gudmundsson [1995] concludes that many of the long dikes generated during episodes on Icelandic central volcanos are propagated primarily vertically from deeper magma sources. Another source of information about magma delivery systems comes from ophiolites. Interestingly, a study of flow directions in the dikes of the Troodos ophiolite [Varga et al., 1998] showed a significant lateral component of magma transport during diking.

Embley et al. [1994] proposed that the dike(s) that produced the 1980s eruption(s) at the north Cleft segment had a significant lateral component. This was based primarily on the observation that hydrothermal cooling proceeded more rapidly at the end of the dike even though that is where the largest eruption occurred. Their preferred interpretation of this observation was that the distal end of the dike is "rootless"; that is, it did not erupt vertically from a local magma source but was fed laterally from a more robust magma source. The longer-lasting high-temperature venting associated with a sheet flow eruption to the south was assumed to be over the location of the primary magma source.

What was the mode of dike injection during the 1993 event? The northward propagation of the T waves during the 1993 CoAxial event is certainly consistent with a lateral dike injection [Dziak et al., 1995]. Also, Butterfield et al. [1997] concluded that the chemistry of the 1993 pillow mound site vent fluids could be explained by the cooling of the dike and lava flow alone (i.e., "rootless portion"), whereas the chemistry of the fluids at the Floc site constrained them to a larger and hotter source. Although there were no observations of active venting on the older pillow lava eruptive mounds at the CoAxial site, the presence of the yellowish sediments on the 1982–1991 mound is consistent with a low-temperature reaction zone of a rapidly cooling dike and lava flow. Smith [1999] explained a decrease in MgO content in the lavas at the Floc site 1981–1991 eruption mounds as crystal fractionation within a northward propagating dike. Finally, two of the CoAxial eruptions and the north Cleft eruption [Chadwick and Embley, 1994] were characterized by 80–100% of the eruption by volume being emplaced at the deep end of the segments. This pattern is also consistent with a lateral dike injection in that it is [Fialko and Rubin, 1998] hydraulically easier to erupt lava as the seafloor deepens.

Thus the spatial and temporal patterns of eruption and hydrothermal cooling and the lava geochemistry of the 1993 and older events are consistent with a dike that had a significant lateral component of injection. Admittedly, this conjecture is largely circumstantial, unfortunately; there are very few data (e.g., seismic reflection) on the along-axis melt supply of the Juan de Fuca Ridge. However, some new and more direct evidence for lateral dike injection on a segment of the Juan de Fuca Ridge was the 3-m drop in the caldera of Axial Volcano (Figure 1) measured by a seafloor pressure gauge coupled with eruption of lava on the rift zone during the seismicity episode in 1998 [Fox, 1999; Embley et al., 1999]. Clearly, monitoring of diking events (at a range of spreading rates) with arrays of seismometers and instruments to measure vertical and horizontal strain is needed to provide more definitive data on the geometry of MOR crustal level magma plumbing systems.

Return to previous section or go to next section

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage