Variations in structure and tectonics along the Mid-Atlantic Ridge, 23⁰N and 26⁰N

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy at the Massachusetts Institute of Technology and the Woods Hole Oceanographic Institution June 1990

The variation in the depth and width of the median valley along the Mid-AtlanticRidge (MAR) suggests that the formation of ocean crust at slow spreading centers is not asimple two-dimensional process in which crustal accretion occurs uniformly both along theridge axis and with time. Rather, it has been proposed that the ridge axis can be dividedinto a number of distinct segments or spreading cells. This thesis investigates thesegmentation model by studying the variability in the structure and tectonics withinspreading cells at 23°N and 26°N along the MAR. The results support the segmentationmodel in which accretion varies along the ridge, evolving as independent spreading cells orsegments, with different portions of the ridge system being in different stages of volcanicand tectonic evolution.Chapter 2 presents an overview of morphologic and tectonic variations along a 100-km-length of the MAR south of the Kane Fracture Zone (MARK area). Sea MARC.I sidescan sonar data and multi-beam Sea Beam bathymetry are used to document the distributionof crustal magmatism and extensional tectonism near 23°N. The data indicate a complexmedian valley composed by two distinct en echelon spreading cells which overlap in adiscordant zone that lacks a well-developed rift valley or neovolcanic zone. The northerncell, immediately south of the fracture zone, is dominated by a large constructional volcanicridge and is associated with active high-temperature hydrothermal activity. In contrast, thesouthern cell is characterized by a NNE-trending band of small fissured and faultedvolcanos that are built upon relatively old, fissured and sediment-covered lavas; this cell isinferred to be in a predominantly extensional phase with only small, isolated volcaniceruptions. Despite the complexity of the MARK area, volcanic and tectonic activity appearsto be confined to the 10-17 km wide inner rift valley. Small-offset normal faulting alongnear-vertical planes begins within a few kilometers of the ridge axis and appears to belargely completed by the time the crust moves out of the median valley. Mass-wasting andgullying of scarp faces, and sedimentation which buries low-relief seafloor features, are themajor geological processes occurring outside the rift valley. In Chapters 3 and 4, the microearthquake characteristics and P wave velocitystructure beneath the median valley of the Mid-Atlantic Ridge near 26°N are studied; thisridge segment is characterized by a large high-temperature hydrothermal field situatedwithin the inner floor at the along-axis high. Chapter 3 explores the tectonic variationswithin the crust as evidenced from the distribution and source mechanisms ofmicroearthquakes observed by a network of seven ocean bottom hydrophones and twoocean bottom seismometers over a three week period in 1985. Hypocenters weredetermined for 189 earthquakes, with good resolution of focal depth obtained for 105events. Almost all events occurred at depths between 3 and 7 km beneath the seafloor,with earthquakes occurring at shallower depths beneath the along-axis high (<4 km). Thedistribution of hypocenters and the diversity of faulting associated with earthquakesbeneath the inner floor and walls suggests a spatially variable tectonic state for the ridgesegment at 26°N. These variations are presumably a signature of lateral heterogeneity in thedepth region over which brittle failure occurs, and are a consequence of along-axis changesin the thermal structure and state of stress.We suggest that at present the hydrothermal activity and deposition of massivesulfides is being sustained by heat generated by a recent magmatic intrusion. Aconsequence of this scenario is that thermal stresses play a dominant role in controlling thedistribution of earthquakes and nature of faulting. Such a hypothesis is consistent with anapparent lack of seismicity beneath the hydrothermal field, the location of hypocentersaround the low velocity zone (Chapter 4), attenuation of P wave energy to instruments atopthe high (Chapter 4), the higher b-values associated with the along-axis high region, andthe occurrence of high-angle (or very low angle) normal faulting and reverse faulting, aswell as the variability in nodal plane orientations, associated with inner floor events beneaththe along-axis high and the volcano. In Chapter 4, we report results from the explosive refraction line and from thetomographic inversion of P wave travel time residuals for seismic velocity structure in thevicinity of the hydrothermal field. The twcrdimensional along-axis P wave structurebeneath the inner floor indicates that young oceanic crust cannot be adequately characterizedby a simple, laterally homogeneous velocity structure, but that one-dimensional StruGturesare at least locally valid (at 5-10 km length scales). The shallowmost crust (upper 1-2 km)beneath an axial volcano and the along-axis high is characterized by significantly highervelocities (by more than 1 km/s) than are associated with the upper crust in the deepestportions of the median valley. The variation is inferred to be a consequence of more recentmagmatic and volcanic activity in the along-axis high region, as compared with the alongaxisdeep where tectonic fissuring has created a highly porous crust characterized by lowerseafloor velocities. The crust beneath the along-axis deep appears to be typical of normalyoung oceanic crust, with a mantle velocity of 8.25 krn/s observed at 5 k:m depth.A low velocity zone centered beneath the along-axis high and extending under anaxial volcano is imaged from 3 to 5 km depth (7.2 km/s to 6.0 km/s); the velocity decreaseis required to satisfy the travel time residual data and to explain the severe attenuation incompressional wave energy to instruments atop the along-axis high. The presence of anactive high-temperature hydrothermal field atop the along-axis high, together with theobservations of lower P wave velocities, the absence of microearthquake activity greaterthan 4 km in depth, and the propagation of S waves through the crust beneath the volcanoand along-axis high (Chapter 3), suggest that the volume corresponds to a region of hot rock with no seismically-resolvable pockets of partial melt. The shallow velocity gradientsdescribing the low velocity volume(<0.6 s-l) appear to be a corrunon characteristic ofinferred zones of magmatic intrusion on the MAR. Comparison of the depth to the velocityinversion with the depths determined in other seismic studies at locally high regions alongthe MAR, the Juan de Fuca Ridge, and the East Pacific Rise reveals a correlation betweenlid thickness and spreading rate, suggesting that the amount of magma available at eachlocation is spatially variable, or that the differences in lid thickness are describing thetemporal evolution of magmatic intrusions beneath mid-ocean ridges.In Chapter 5, the first direct measurement of upper mantle P- and S-wave delaytimes beneath an oceanic spreading center is presented. Two independent estimates of theepicenters and origin times are made for each of two earthquakes in a 1985 earthquakeswarm near 25°50'N on the Mid-Atlantic Ridge using local and teleseismic arrival timedata. Comparison indicates a 14-20 km northward bias in the epicenters teleseismicallylocated using a Herrin [1968] Earth model. The bias is due to departures of the actualvelocity structure from that implicit in the travel time tables used for the locations,combined with unbalanced station distribution. The comparison of origin times for thebest-located event, after correction for the epicentral bias and for an oceanic crustalthickness, shows there to be only slightly lower velocities than a Herrin [1968] uppermantle; the P-wave delay is +0.3 ± 0.9 s (+0.2 ± 0.9 sand -2.4 ± 0.9 s relative to theisotropic Preliminary Earth Reference Model (PREM) and the Jeffreys-Bullen [1940] (JB)travel time tables, respectively). The lack of a resolvable P-wave delay suggests that theHerrin [1968] model is a good approximation to the average upper mantle velocity beneaththis segment of the MAR.Measurement of the S-wave delay for the same MAR swarm event shows there tobe a positive delay (+3.1 ± 2.0 s), or larger travel times and slower velocities compared tothe JB S-wave tables (+ 3.9 ± 2.0 s relative to the isotropic PREM S-wave model). Incontrast to the larger P-wave delays found in other MAR studies, the lack of a significantseismic anomaly near 26°N indicates that sizeable regions of low velocity material do notpresently exist in the upper few hundred kilometers of mantle beneath this section of theridge. This evidence argues for substantial along-axis variations in the active upwel~ng ofmantle material along the slowly-spreading Mid-Atlantic Ridge. In order to explain theobservation of a smaller than expected P wave delay in a region where the S delay suggestssignificant temperature anomalies (low velocities), we propose a model for mantleupwelling in which the decrease in travel time is due to an anisotropic P wave structure(fast direction vertical); the anisotropy results from the reorientation of olivine crystalsparallel to the ascending flow and balances the travel time delay due to a region of lowvelocities.

This thesis was funded in part by grants EAR-8407798, EAR-8407745, and EAR-8817173 from the National Science Foundation.