U.S. Dept. of Commerce / NOAA / OAR / PMEL / Publications
Discussion
Sequence of Events
The detailed observations presented above offer some important clues about the recent sequence of events at the north Cleft segment.
Pre-YSF eruptions. Little is known of specific events before the eruption of the YSF, but the axis of intrusion and extrusion was apparently 100200 m east of the YSF eruptive vents, judging from the locations of Age 2 pillow mounds and associated fissure swarms (Figures 3 and 5). The age of the Age 2 pillow mounds is unknown, but judging from the amount of sediment cover, some of these mounds probably erupted within the past few hundred years. The older chimneys probably formed after the eruption of these mounds and the older sheet and lobate flows on their west side. One of these chimneys has been dated at >100 years by Pb210 [Koski et al., this issue].
YSF and NPM eruptions. Both the YSF and NPM eruptions were fed from long line sources; dikes intruded along different parts of the same fissure system [Embley et al., 1991; Chadwick and Embley, this issue]. However, there is evidence that the YSF and NPM erupted during two separate events probably separated by only a decade or less. The lava flows produced by these eruptions differ dramatically in their morphology, indicating that the YSF was erupted at a much higher rate than the NPM [Griffiths and Fink, 1992].
The north-to-south topographic gradient and the NE-SW trend of the YSF lineations indicate that lava flowed from north to south, and the evidence for drainback into the northern part of the fissure system strongly points to the area between Cavern and Monolith as a primary eruptive vent area. However, submersible observations indicate that lava was also erupted from the fissure system as far south as 44°57'N.
The sheet flow covers about 3.5 km2, but its thickness is not directly measurable. The volume of the sheet flow is 0.0035 km3 for each meter of thickness, which means that it would have to be 14 m thick to equal the estimated total volume of the NPM (0.05 km3 [Chadwick and Embley, this issue]). Submersible observations and topographic profiles (Figure 6) suggest that the YSF is probably no more than 5 m in thickness, making it of considerably less volume than the NPM.
The origin of lava pillars has been hypothesized to be from the trapping and superheating of water under a lava flow [Francheteau et al., 1979], analogous to the formation of spiracles on marshy areas on land [Waters, 1960]. Observations of sheet flows along the Galapagos Rift [Ballard et al., 1979] and the EPR [Francheteau and Ballard, 1983] suggested that pillars form within the margins of lava ponds or lakes, whereas the north Cleft pillars formed along the boundary of a moving flow, although it may have ponded temporarily as it advanced. Presumably, the YSF moved faster in the middle than at the edges, so the pillars only formed where water pockets were trapped and could rise up through the relatively stagnant edges of the flow. The lineated flows in the floors of collapsed areas mark the top of the subsided interior of the lava flow after the bulk of the flow had drained out to the south or back into the fissure.
Magma supply system: Clues from the pattern of hydrothermal activity. The eruptive fissure on the eastern side of the YSF is the locus of most of the hydrothermal venting along the northern Cleft segment. Water column data [Embley et al., 1991; Baker, this issue] show that the northern limit of the hydrothermal plumes is at about 45°03'N. Towed camera and submersible data show only sporadic venting north of about 45°00'N and no active venting north of 45°04'N. The YPM are, in places, covered with thick pockets of yellow/orange sediments with some sulfide minerals [Chadwick and Embley, this issue], indicating some short-lived venting probably occurred immediately following the eruption. The most important point is that the NPM are characterized by a lack of hydrothermal venting relative to the YSF area even though the volume of extruded lava in the NPM is probably greater than the YSF.
One model to explain this pattern is that the YSF overlies a magma reservoir and that the NPM were fed by a lateral dike injection from the reservoir (Figure 7). The proposed magma body beneath the YSF is the heat source for the long-lived, high-temperature vents there, but only short-lived venting was associated with the dike intrusion beneath the NPM. Dikes of several meters width (typical of ophiolitic, and by inference, oceanic dikes) intruded into cold host rocks will "freeze" (cool to solidus temperature) within a few months [Wilson and Head, 1988]. Alternatively, the dike that fed the NPM eruption could have been injected vertically from below, but the lack of a long-lived (more than a few years) hydrothermal system implies that its magma source was either completely depleted by the eruption or is sealed off from transferring heat to the seafloor. We prefer the lateral dike injection model because it is supported by the body of evidence from the mid-ocean ridge system and from terrestrial volcanic systems for long-distance lateral dike injection within rift zones.
Figure 7. Cartoon of preferred model for Cleft 1980s intrusion/eruption episode(s) based on geologic and hydrothermal observations discussed in this paper. Sketch covers along-axis section from approximately 44°53'N to 45°10'N.
Iceland is one of the few portions of the mid-ocean ridge above sea level, and while it may be anomalous in some ways, it has helped illustrate important plate boundary processes. A well-monitored rifting episode occurred in northern Iceland between 1975 and 1984, and consisted of repeated lateral dike injections, basaltic fissure eruptions, and associated tectonic extension along the plate boundary [Bjornsson et al., 1979; Sigurdsson, 1987]. During each of 20 discrete rifting events, magma from a reservoir beneath Krafla Caldera was injected into the rift zone and dikes were propagated for tens of kilometers. It is noteworthy that the along-axis gradients of the Krafla Rift Zone and the Cleft segment do not dramatically differ (Figure 2). Evidence for lateral dike injection is also commonly observed in Hawaii [Rubin and Pollard, 1987] and has recently been found within the Troodos ophiolite [Staudigel et al., 1992].
In addition to the pattern of hydrothermal activity, the relationship between faults and fissures and volcanic vents supports the lateral dike injection model for the north Cleft segment. The NPM lie along an apparently continuous fissure/graben system (Figures 1 and 3 [Chadwick and Embley, this issue]) that is also along the same strike as the eruptive fissure of the YSF. Continuity of structures and their association with volcanic vents are all consistent with the lateral passage of dikes close to the seafloor [Rubin, 1992].
A detailed analysis of the Cleft segment lava chemistry data led Smith et al. [this issue] to conclude that the NPM and YSF lavas have slight, but significant chemical differences. The YSF lavas are relatively evolved and remarkably chemically homogeneous, whereas the NPM are more mafic and show more chemical diversity [Smith et al., this issue]. The geochemical data are consistent with the idea that the NPM lavas were erupted after the YSF. The more primitive NPM magma apparently rose from a deeper level, mixed with the more differentiated and shallower YPM magma and then was intruded into the shallow oceanic crust.
Geophysical data are equivocal on the presence or absence of a small magma lens beneath the north Cleft segment. Morton et al. [1987] collected multichannel seismic reflection profiles over the Cleft segment and interpreted a faint reflector at a subbottom depth of 2.32.5 km to be the top of a partly solidified magma chamber. This reflector appears on cross sections beneath the Cleft axial valley at 44°40'N and the Vance axial valley at 45°05'N, but the only data from the YSF area (part of an along-axis profile) were not reprocessed to verify the presence of the reflector. Shallow seismic refraction data for the northern Cleft segment [McDonald et al., this issue] are explained in terms of variations in layer 2A thickness, but a magma lens could be present below the limit (~500 m) of those measurements. Analysis of bottom gravity measurements reveals a possible low-density anomaly in the overlap region between the Cleft and Vance segments, which Stevenson et al. [this issue] interpret as a magma lens. In any case, the resolution of the available data is insufficient to precisely relate surface morphology and depth of magma lens (or absence thereof) to the hydrothermal vent distribution, such as Haymon et al. [1991] were able to do at the 9.5°N East Pacific Rise site.
The most likely cause of megaplume formation over the southern Juan de Fuca Ridge was the rifting associated with the NPM intrusion and eruption [Baker et al., 1989; Baker and Lupton, 1990; Embley et al., 1991]. Less clear is the extent of the rifting. No new lava has been observed along the southern Cleft segment during submersible dives mad e in the late 1980s and early 1990s, and Baker [this issue] does not report any major perturbations at the southern site during the annual along-axis CTD tows made since 1986. Therefore, even though the south Cleft segment is the shoalest part of the segment, it seems unlikely that the recent northern lavas can be attributed to along-strike dike injection originating on the southern end of the Cleft segment or that the two sites are underlain by a single magma chamber. Also, variations in lava chemistry along the Cleft segment [Smith et al., this issue] strongly imply that separate melt lenses have fed the youngest flows at the northern and southern ends of the segment. Therefore the most recent spreading event may have been confined to the northern portion of the segment.
Baker [this issue] shows that the level of hydrothermal activity as measured in the integrated neutrally buoyant plume has been steadily decreasing since 1990. If this trend continues, one may conclude that the magma lens beneath the YSF area is relatively small and is cooling faster than it is being replenished or is being sealed off from the seafloor plumbing system. Continued monitoring of the plumes and vent fluids over the next several years will test this hypothesis. As this paper was in its final stages of revision in June 1993, a submarine rifting event was detected by a newly developed real-time monitoring system using the U.S. Navy's Sound Surveillance System (SOSUS). Swarms of seismic events occurring over several weeks were observed between 46°10'N and 46°35'N on the northern Juan de Fuca Ridge [Fox, 1993; Dziak and Fox, 1993]. This episode also generated megaplumes (albeit somewhat smaller than the 1986 megaplume) that formed over the site of a small volcanic eruption that took place during the seismic swarm period [Embley et al., 1993]. This event appears to be analogous to the Cleft episode, and may provide an important step forward to a better understanding of the mechanism of crustal accretion at intermediate-rate spreading centers.
Return to previous section or go to next section
