Deep-Tow Studies of the Structure of the Mid-Atlantic Ridge Crest Near Lat 37°N

Deep-tow studies of the structure of the Mid-Atlantic Ridge crest near lat 37°N

KEN C. MACDONALD* Institute of Geophysics and Planetary Physics, Scripps Institution of Oceanography, La Jolla, California 92093 BRUCE P. LUYENDYK Department of Geological Sciences, University of California, Santa Barbara, California 93106

This article is one of a series appearing in the April and May issues of the Geological Society of America Bulletin on the scientific results of Project FAMOUS. These studies were undertaken in the axial area of the Mid-Atlantic Ridge between approximately 36°30' and 37°N latitudes.

ABSTRACT large faulted blocks toward nearby fracture Le Pichon, 1974). We discuss here the de- zones suggests that spreading-center tec- tailed structure and evolution of the rift val- A detailed study of the structure of the tonics is affected by fracture-zone tectonics leys between fracture zones A and B (rift Mid-Atlantic Ridge median valley and rift throughout the length of the rift in the valley 2) and B and C (rift valley 3) and the mountains near lat 37°N (FAMOUS) was FAMOUS area. Both the crustal accretion rift mountains in these areas. Fine-scale conducted using a deep-tow instrument zone and transform fault zone are narrow, deep-tow studies of the inner floor of rift package. The median valley may have either only 1 to 2 km wide, over short periods of valley 2 are discussed by Luyendyk and a very narrow inner floor (1 to 4 km) and time. In the course of millions of years, how- Macdonald (1977). Studies of near-bottom well-developed terraces or a wide inner floor ever, they apparently migrate over a zone 10 magnetic anomalies and a tectonic synthesis (10 to 14 km) and narrow or no terraces. to 20 km wide. of rift valleys 2 and 3 are presented by The terraces appear to be non-steady-state Macdonald (1977). features of the rift valley. The entire depth OBJECTIVES and gross morphology of the median valley SETTING may be accounted for by normal faulting, Recently, several detailed surveys have while volcanic relief contributes to the been made on the Mid-Atlantic Ridge (at lat The FAMOUS area lies south of the short-wavelength topography (<2 km). 45°N, Aumento and others, 1971; at lat Azores triple junction on the North Most faults dip toward the valley axis an 37°N, Needham and Francheteau, 1974; American-African plate boundary (Fig. average of 50°, and the blocks are tilted back and at lat 26°N, McGregor and Rona, 1A). Rotation of magnetic anomalies about 2° to 3°. Fault dip is asymmetric about the 1975). However, the limitations of wide- the North America-Africa pole of opening valley axis. Active crustal extension in the beam echo sounders and other conventional and absence of a distinct median valley in- inner floor and inner walls has the same geophysical tools have made it difficult to dicate that the ridge crest is influenced by sense of asymmetry as the local spreading investigate the fine-scale structure and tec- triple-junction tectonics as far south as lat rates, reaching a maximum of 18 percent. tonics of the median valley and rift moun- 38°N (J. D. Phillips and H. S. Fleming, in Thus, asymmetric spreading appears to be tains. Several fundamental problems have prep.). South of lat 38°N, however, the me- accomplished by asymmetric crustal exten- remained unsolved: (1) the width and mor- dian valley is well defined, and magnetic sion on a fine scale as well as by asymmetric phologic expression of the crustal accretion anomalies 2' prime and 5 overlap properly crustal accretion. Spreading is 17° oblique to zone within the median valley, (2) the roles when rotated back about the pole of open- the transform faults and shows no indication of volcanism and faulting in creating relief ing. The part of the ridge studied consists of of readjusting to an orthogonal system, even in the median valley and in the rift moun- two segments, each 40 km long, trending on a fine scale. Eighty percent of the decay or tains, (3) the morphologic and structural N17°E (Figs. 1A, IB). Fracture zones A, B, transformation of median-valley relief into expression of asymmetric and oblique and C each right-laterally offset the valley rift-mountain topography is accomplished spreading, (4) the degree of horizontal ex- about 20 km and trend east. Spreading is by normal faults that dip away from the tension of crust within the plate-boundary currently 17° oblique, and detailed studies valley axis. Most of the outward-facing zone, (5) the tectonic interaction between of the tectonic and magnetic trends in the faulting occurs near the median-valley- transform faults and spreading centers, (6) inner floor of the rift suggest that the rift-mountain boundary. Tilting of crustal the relationship between microearthquakes spreading is stably oblique—that is, the blocks accounts for only 20 percent of the and the fine-scale structure of ridge crests, ridge is not readjusting to an orthogonal decay of median-valley relief. Most long- and (7) the evolution of the median valley configuration (Macdonald, 1977; Luyen- wavelength topography in the rift moun- into the relief of the rift mountains. dyk and Macdonald, 1977). Oblique tains has a faulted origin. As in the median To address these problems, a near- spreading may have been stable here for at valley, volcanic relief is short wavelength bottom geophysical study of the Mid- least 6 m.y., even through a change in (<2 km) and appears to be fossil, originating Atlantic Ridge near lat 37°N was conducted spreading direction (Macdonald, 1977). In in the median-valley inner floor. Bending of from the R/V Knorr (cruise 31) of the addition, spreading is highly asymmetric in Woods Hole Oceanographic Institution, rift valley 2, with rates of 7.0 mm/yr to the using the deep-tow instrument package of west and 13.4 mm/yr to the east. At 1.7 the Marine Physical Laboratory (Scripps m.y. B.P., the sense of asymmetry reversed, * Present address: Marine Physical Laboratory and with rates of 13.3 mm/yr to the west and Geological Research Division, Scripps Institution of Institution of Oceanography). This effort is Oceanography, La Jolla, California 92093. part of the FAMOUS project (Heirtzler and 10.8 mm/yr to the east (Macdonald, 1977).

Geological Society of America Bulletin, v. 88, p. 621-636, 16 figs., May 1977, Doc. no. 70502.

621

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 622 MACDONALD AND LUYENDYK

38°W

Figure 1A. Regional setting and plate boundaries in FAMOUS area (unpub. U. S. Navy bathymetric chart, courtesy of J. Phillips and W. Perry). E-E'" is long deep-tow traverse into eastern rift mountains. FZ=Fracture zone.

Spreading is only slightly asymmetric in rift tains is discussed here first. Rift valley 3, looking sonar has a range of 400 to 700 m valley 3, with rates of 9.8 mm/yr to the west which has a much wider inner floor than rift to either side, resulting in an effective search and 10.6 mm/yr to the east (Macdonald, valley 2, is discussed for comparison. path approximately 1 km wide. The side- 1977). looking sonar is particularly effective in Most of the FAMOUS investigations, in- DATA INTERPRETATION mapping linear scarps on a small scale and cluding use of manned submersibles, have is useful in determining the linearity of fea- focused on rift valley 2 (Fig. IB) (Reid and The structural studies are based on in- tures. We (Luyendyk and Macdonald, Macdonald, 1973; Needham and Fran- terpretation of three types of deep-tow 1977) discuss photographic data, and cheteau, 1974; Macdonald and others, sonar records: high-frequency (40 kHz) (Macdonald, 1977) discusses near-bottom 1975; Laughton and Rusby, 1975; Bel- narrow-beam sonar for depth, low- magnetic data. laiche and others, 1974; Moore and others, frequency (4.0 kHz) sonar for sediment Within the outer walls of the median val- 1974; ARCYANA, 1975; Ballard and penetration, and left and right side-looking ley the fish was navigated using bottom- others, 1975; Ballard and van Andel, 1977; sonar (110 kHz) (Spiess and Tyce, 1973). moored acoustic transponders (Spiess and Bryan and Moore, 1977; Luyendyk and When the fish is towed at normal heights Tyce, 1973). Relative accuracy is 10 to 50 Macdonald, 1977; Macdonald, 1977). Rift above bottom (50 to 150 m), the 40-kHz m. Outside the transponder net the fish was valley 3 has also been studied but in some- sonar can track slopes greater than 70°, as tracked using satellite navigation and a what less detail (Laughton and Rusby, verified by submersible observations of the cable trajectory program by Ivers and 1975; Macdonald, 1977; Ramberg and van same scarps (FAMOUS dive team, 1975, Mudie (1973), for which accuracy is 500 to Andel, 1977; J. D. Phillips and H. S. Flem- personal commun.). The 4.0-kHz sonar re- 1,000 m. Location relative to latitude and ing, in prep.). The deep-tow studies concen- liably penetrates and detects sediment longitude was determined by comparing trating on rift valley 2 and the rift moun- thicknesses from 5 to 100 m. The side- some 40 satellite fixes with transponder-

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 STRUCTURE OF MID-ATLANTIC RIDGE CREST 623

Inner Walls

The inner floor is bounded by the inner walls (Figs. 2, 3). The walls strike approximately N17°E and are continuous throughout the length of the valley, right up to the transform-fault intersections. The northern end of the east inner wall actually forms a corner with the major east-west scarp, which marks the southern boundary of fracture zone A (Fig. 3). Such "corner cliffs" have also been observed at the intersection of the west inner wall with fracture zone B (Laughton and Rusby, 1975). The structure of the inner walls is very asymmetric. East of the valley axis, the floor rises at a slope of 9° for 2.2 km before meet- ing the first major scarp marking the inner wall (Figs. 2, 4). This slope is composed of 13 steps, which are aligned in a N17°E direction (Fig. 4A). The fault scarps have an average dip of at least 60°, and some are essentially vertical. Throws range from 5 to 30 m. This intense faulting commences within 400 m of the inner-floor axis. Scarp dimensions increase abruptly at the first major block fault of the inner wall, with throws of 150 to 350 m and block widths of 500 to 1,500 m (Figs. 2, 3, 4B). The wall looks like a staircase in cross section, consisting of three to six major blocks (Fig. 4B). Faulting is not as intense on the west side of the inner floor, and there is no gradual slope composed of small step faults leading up to the wall. Instead, the floor is abruptly truncated by the west inner wall (Figs. 2, 5). To a first approximation, the west wall is a Figure IB. Bathymétrie chart of FAMOUS area; 100-fm contour interval (after J. D. Phillips and single massive block fault with a throw of H. S. Fleming, in prep.). Solid lines show location of geophysical profiles. Strike-parallel lines in medi- as much as 600 m. In detail it is composed an valley not shown (see Fig. 3). Dashed lines show location of supplementary surface-ship bathy- of a series of narrow slivers of faulted crust métrie profiles shown in Figure 20. (Fig. 5A, B). Microearthquakes recorded in the me- navigated ship positions. However, so that (2) The center of the inner floor is marked dian valley cluster at the first and second the deep-tow work could be compared to by a narrow (1.0 to 1.5 km) lineation of al- steps of the inner walls and not along the submersible work directly, the coordinates ternating central highs and depressions valley axis (Fig. 6; Reid and Macdonald, were shifted 0.7' west to match the submer- which appears to be the locus of youngest 1973; Spindel and others, 1974). This sible base map. crust. (3) The central highs have elongation suggests that the quakes are associated with ratios of about 4:1 and appear to be created the incipient and ongoing uplift of fault MORPHOLOGY AND STRUCTURE mostly by linear fissure eruptions that paral- blocks forming the inner wall rather than OF RIFT VALLEY 2 lel the trend of the valley. (4) The central with dynamics of intrusion in the inner lows may be either grabens or collapse floor. Of 20 events located on the inner Macdonald and others (1975) divided rift structures. (5) Volcanism dominates the re- walls, 18 were on the east inner wall (Fig. valley 2 into four physiographic provinces: lief of the inner floor. Faults are numerous 6). In a later microearthquake study of the (1) the inner floor, (2) the inner walls, (3) but have small throws (2 to 10 m, generally) rift valley north of fracture zone A, the west the terraces, relatively flat regions between and make up very little of the inner-floor re- wall there was found to have almost an the inner and outer walls, and (4) the outer lief. (6) Intense faulting and fracturing of equal level of seismicity (Francis and others, walls, which form the boundary between the crust is observed beyond 500 m from 1977). The level of earthquake activity was the median valley and the rift mountains the center of the floor, with fault spacings as high, averaging 10 to 30 events per day (Fig. 2). close as 20 m. (7) The hundreds of small (Reid and Macdonald, 1973; Spindel and faults and tensional cracks mapped in the others, 1974). Inner Floor inner floor have an average trend parallel to Fault scarps were analyzed to determine that of the median valley. The scatter in dip. Scarps exceeding 100 m in throw and Deep-tow studies of the inner floor of rift trends forms a normal distribution about with sharp rectilinear crosssections were valley 2 are discussed by Luyendyk and N17°E, with a standard deviation of only considered. Talus piles at the bases of Macdonald (1977). Here we summarize 6°. Thus, there is no indication of reorienta- scarps commonly appear in bottom photos some major findings pertinent to this paper: tion to a north-south, orthogonal spreading (Luyendyk and Macdonald, 1977), so dip (1) The inner floor is only 2 to 3 km wide. direction, even on a fine scale. was measured near the top of the scarp,

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 624 MACDONALD AND LUYENDYK

generally on its steepest slope. The average is no significant difference in tilt between major faults, with a gradual slope at the base dip toward the valley axis is 50°. However, the east and west inner walls. That the out- of the wall consisting of a series of small the west inner wall is somewhat steeper ward slopes of the blocks are indeed caused steps (Fig. 2, F-F', G-G', H-H'; Fig. 4A). A than the east wall (Tables 1,2). The steeper by tilting has been verified by observation similar type of relay faulting occurs on the dip of the west wall is largely caused by the of elongate pillows directed upslope on top west inner wall, so that to the south, the wall prominent scarp responsible for most of the of some of the tilted blocks (U.S. FAMOUS is wider, consisting of at least eight slivers throw of the west wall, which has dips ex- team, 1975, personal commun.). with considerable backward tilt (Fig. 5B). ceeding 75° (Fig. 2, D-D'; Fig. 5). The same vertical uplift occurs all along To the north, the wall becomes narrower Most of the blocks face the valley axis, the east inner wall but is accomplished and crustal slivers show less tilt (Fig. 5A), and some are tilted back. Of the blocks through an overlapping and bifurcating and farther north the wall looks like a single showing a measurable backward tilt, the series of faults of varying throw. This has massive block (Fig. 2, D-D'). At the north average tilt is 6° [standard deviation (SD) been observed in the Afar rift and is termed end, the throw of this scarp decreases to 200 = 2°; standard error (SE) = 0.7°]. The av- "relay faulting" (H. D. Needham, 1974, m, the remainder of the uplift appearing on erage for all the inner-wall blocks, however, personal commun.). In places, the east inner the block carrying Mount Mercury (Fig. 2, including those with zero or forward (nega- wall consists of five to six major steps (Fig. A-A', B-B'; Fig. 3). This fault is 4.5 km tive) tilt, is 2° (SD = 2.0; SE = 0.3°). There 4B). Along strike it changes to two or three long, has as much as 400 m of throw, and is

_l -15 -10 -5 0 KM 10 15 20 25 OUTER TERRACE INNER I TERRACE OUTER WALL FLOOR I WALL \ INNER ^ V.E. = 2X WALL

Figure 2. Near-bottom bathymétrie profiles across rift valley 2. Labels indicate physiographic provinces. Numbers with sediment ponds indicate maximum thickness in metres. L = lip (volcanic construction perched at edge of faulted block); V = other volcanic construction outside inner floor. Dashed lines = major faults discerned from near-bottom echo-sounder and side-looking sonar data. Vertical exaggeration is 2X to reduce distortion.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 STRUCTURE OF MID-ATLANTIC RIDGE CREST 625

only 0.5 km from the valley axis, suggesting the faults are either active or originated near steeper on the west terrace than on the east that very large scale block faulting can occur their present location. terrace (Tables 1,2). As with the inner walls, within hundreds of metres of the center of The asymmetry in extension is largely this results in greater extension associated the floor. caused by the asymmetry in fault density and with faulting on the east terrace than on the Horizontal extension was calculated from fault dip between east and west. It is also in west. This is in keeping with the higher fault dips and throws on the traverse cros- keeping with the highly asymmetric spread- spreading rate on the east side and with the sing Mount Venus (Fig. 2, D-D'). The prin- ing rates that prevail out to the outer wall greater width of the east terrace. The average cipal assumption is that the faults considered (anomaly 2). The rates are 7.0 mm/yr to the dip of scarps on both terraces is less than that are either still active or originated near their west and 13.4 mm/yr to the east (Mac- of the corresponding inner wall. However, present location. Extension is highly asym- donald, 1977). The ratio of east to west the significance level of the difference is low, metric (Fig. 7). The west wall and inner floor spreading rate is close to that of east to west and there is no measurable change in tilting represent only 230 m of extension, com- horizontal extension (11 to 18 percent). of the blocks (Tables 1, 2). pared to 870 m for the east. Normalized Thus, the extreme asymmetry in spreading The fault pattern on the terraces is quite against distance from the floor axis, only 11 rate may be due to a greater rate of horizon- different from that on the inner walls. The percent of the horizontal movement is due to tal extension to the east as well as a higher lateral spacing between faults with throws the extension on the west, as opposed to 18 rate of crustal accretion (spreading rate exceeding 50 m increases from 0.8 km for percent on the east. Extension rates within equals rate of crustal accretion plus rate of the inner walls to 2.4 km on the terraces the floor are lower—4 percent on the west horizontal extension). (Fig. 2). The change in fault density suggests side and 6 percent on the east. Most of the either that coalescing of faults by reverse extension occurs in the inner walls. Exten- Terraces faulting occurs in going from the inner sion decreases abruptly on the terraces. walls to the terraces or that the terraces are However, some horizontal extension must The median-valley terraces between the not steady state. Reverse faulting requires occur in the terraces and even in the rift outer and inner walls are characterized by compressional stresses within the median mountains, because there is evidence of ac- relatively horizontal, flat topography. The valley. This contradicts focal-mechanism tive normal faulting there (discussed below). terrace is 14 km wide on the east side and 8 solutions (Sykes, 1967) and the observation The amount of extension there is indeter- km on the west. Block-fault scarps form of normal faulting and graben formation in minant because it is not known how many of much of the relief. Fault dips are significantly the median valley. There are also regions as

37°00"

Figure 3. Tectonic map of rift valley 2 and its intersection with fracture zone A. Height of scarps is indicated by thickness of lines, horizontal extent of scarps by hachures. Fracture zone A physiography is after De- trick (1974). Notice asymmetry in structure of inner walls and terraces, large graben on west terrace, and sharp intersection between west inner wall and fracture zone A, forming "corner cliffs."

33°26' 24' 22' 20' 18' 16' 14' 12' 10' 08' 06' 04' 02'

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 626 MACDONALD AND LUYENDYK

1600

5 o 2000

DISTANCE FROM AXIS/KM) Figure 4. A. Tracing of near-bottom bathy- métrie profile, showing intense small-throw nor- 2400 VERTICAL EXAGGERATION 2.3X mal faulting on east side of inner floor leading up to east inner wall. Short dashed lines = faults inferred from bathymétrie and side-looking sonar data. B. Original near-bottom echo-sounding record showing stair-step block-faulted profile of east inner wall (along D-D'). Notice tilting of 1.5 20 3.0 4.0 5.0 6.0 blocks away from rift axis. DISTANCE EASTWARD FROM AXIS OF MEDIAN VALLEY (KM)

wide as 10 km in the terraces that are apparently totally unfaulted, suggesting that large parts of the inner floor were uplifted as single units (for example, 5 to 15 km, Fig. 2). Thus, it appears that the terraces are not steady state but are transient features of the median-valley structure. Supporting this observation is the fact that, at present, terraces are nonexistent or barely identifiable in rift valley 3 (discussed below). On the west side, between 6.5 and 8.0 km from the valley axis, there is a graben 200 to 300 m deep (Fig. 3). It extends at least 20 km parallel to the valley and is bounded by scarps with slopes of 45° to 55°. The only earthquake located on the terraces was at the south end of this graben (Fig. 6), indi- -2.0 -10 DISTANCE FROM AXIS IKMI cating that it is still active. This may be sig- -2.0 -1.8 -1.6 nificant, for it suggests that the zone of ac- DISTANCE FROM AXIS (KM) tive crustal extension is at least 16 km wide, Figure 5. A. Tracing of near-bottom record, showing narrow slivers of crust composing steep west even though most of the extension occurs at inner wall (along G-G'). Notice prominent volcanic lip at top of scarp. Short dashed lines = faults in- the inner walls in a zone 7 km wide. ferred from bathymetric and side-looking sonar data. B. West inner wall (along H-H'), again showing The outward-facing scarps of the graben narrow slivers of crust composing this massive block. Notice that here blocks are tilted back away may be caused by conjugate normal faulting from valley axis. Again, there is a lip at top of fault scarp. or by antithetic faulting, both requiring extension. Antithetic faulting could occur if many of the blocks (Fig. 2). They are gener- inner floor. If the volcano is dormant, block the inward dip of the main fault decreased ally symmetric in cross section and range faulting is likely to be concentrated along its with depth. Further uplift of the west block from being equidimensional to having elon- edges, where thé crust is thinnest, resulting would result in a gap, filled with antithetic gation ratios of 6:1. The morphology and in a volcanic lip perched at the edge of the blocks, at the surface (Fig. 8A). Under in- dimensions suggest a volcanic origin. They block fault (For example, Mount Mercury, creased lithostatic pressure, the internal are very similar to the central highs and Fig. 2, A-A'). If the volcano has been re- friction of rock decreases, resulting in shal- other volcanic constructions in the inner cently active, the crust may be thinnest lower fault dip at depth (de Sitter, 1964). A floor. along its axis, and it may be split in two by shallowing of fault dip with depth in the Most of the highs are not randomly block faulting. The lip at —1 km on D—D' crust has been observed at different ero- situated but form "lips" at the edges of fault (Fig. 2) has coherent flows on the west side sional levels in the Rhine Graben (H. lilies, blocks. Of 82 highs mapped on the terraces, and truncated pillows on the east, suggest- 1974, personal commun.). It provides a 51 (62 percent) were lips (72 out of 107 for ing such splitting (French FAMOUS dive mechanism for outward tilting of fault the entire median valley). If highs of 500-m team, 1975, personal commun.). If lips blocks, as well as for the formation of gra- average width were randomly distributed were created by volcanism outside the inner bens and outward-facing faults through along the fish path on the terrace, only 30 floor, their location relative to the blocks antithetic faulting (Fig. 8B). percent would be lips. It seems likely that might be random. If there were any sys- Topographic highs 100 to 800 m across volcanic highs—like Mount Pluto or Mount tematic relationship between volcanism and and 40 to 200 m high are superimposed on Venus—particularly lips, are created in the block faulting, one would expect eruptions

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 STRUCTURE OF MID-ATLANTIC RIDGE CREST 627

TABLE 1. AVERAGE DIPS OF FAULT SCARPS IN RIFT VALLEYS 2 AND 3

Dip SD SE No. of (°) (°) n samples

Rift valley 2 West inner wall 56 11.2 2.8 17 East inner wall 47 6.1 1.2 27 West terrace 52.5 7.6 1.6 22 East terrace 44 5.3 1.1 23 West outer wall 48.6 8.1 2.4 11 East outer wall 46.4 6.9 2.4 7 Rift valley 3 50 1.3 1.3 8

Note: SD = standard deviation; SE = standard error.

TABLE 2. STUDENT'S T-TEST OF SIGNIFICANCE OF DIFFERENCES IN FAULT DIPS IN RIFT VALLEY 2

Provinces compared Significance (%) East and west inner walls 96 East and west terraces 99 East terrace and east inner wall 90 West terrace and west inner wall 96

Note: Normal distribution assumed.

the valley axis. They are linear and continuous at least 18 km on the west and 10 km on the east. Correlation with the surface-ship bathymetry suggests in fact that the outer walls are continuous over the 30- to 40-km distance between fracture zones A and B (Fig. IB). Figure 6. Seismicity of FAMOUS area. Solid circles = microearthquakes located by Reid and Macdonald (1973), with approximate location accuracies of 2km. Squares = microearthquakes located by Spindel and others (1974), with a 500-m location accuracy. Large open circles = teleseismi- Role of Block Faulting in Creating cally located events for 1961-1972 reported by Environmental Science Services Administration Median-Valley Relief (ESSA), with location accuracies of about 20 km. Approximate plate boundaries are indicated superimposed on same bathymetric map as in Figure 18. To what extent can block faulting account for the depth and relief of the median valley? To quantify the contribution of to occur at the bases of scarps along frac- Spreading rates of 7.0 mm/yr to the west faulting, we tabulated the throws of major tured fault planes, not at the top of fault and 13.4 mm/yr to the east determined from block faults (inward and outward facing blocks on their outer edges. Detailed near-bottom magnetic data are consistent relative to the rift axis) as well as the change magnetic studies of the topography concur with the asymmetric position of the walls in elevation due to tilting of blocks (Fig. 9). with an inner-floor origin for the lips and and show that the outer walls are essentially Inward-facing faults increase the depth of volcanoes. Only 7 out of 170 volcanic fea- isochrons about 1.5 m.y. old. the median valley, whereas outward-facing tures in the FAMOUS area either have very The outer walls stand 800 ± 100 m faults and outward tilting of blocks de- high magnetizations or polarities opposite above the terraces and 1,600 ± 100 m crease the elevation; thus, throw of inward to the surrounding topography, indicative above the valley floor. The outer walls ap- facing faults minus throw of outward facing of younger crust that must have originated pear to have been constructed by block faults minus tilt equals the contribution of outside the inner floor (Macdonald, 1975). faults, the faults dipping toward the valley faulting in creating median-valley depth. (Tables 1, 2) and the block tops tilted back We find that faulting and tilting account for Outer Walls 3° to 8° (Fig. 2, F-F', G-G', J—J'). In places, nearly all the gross relief of the valley and, the tilt is replaced by faults along planes in fact, account for more than 95 percent of The outer walls are located asymmetri- that dip away from the valley axis the depth of the median valley (Fig. 9). Vol- cally about the valley axis, giving a (outward-facing faults). In such cases a canic relief contributes to topographic median-valley half-width of 11 ± 1 km to horst marks the outer edge of the valley roughness but very little to large-scale relief the west and 20 ± 1km to the east. Despite (see, for example, Fig. 2, B—B', H—H'). The (>2-km wavelength). Outward-facing the highly asymmetric location of the outer walls are composed of one to three major faulting and tilting contribute almost walls, their depth is essentially equal on scarps from 100 to 550 m in throw. The equally in decreasing median-valley depth both sides of the median valley (Fig. 2). major scarps forming the outer wall parallel (discussed below).

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 628 MACDONALD AND LUYENDYK

APPROX. AGE (my.) APPROX. AGE (my)

1.0 2.0 3.0 DISTANCE FROM AXIS OF INNER FLOOR (WEST)

B INNER FLOOR W. INNER WALL

Figure 7. A. Cumulative horizontal extension of crust east of rift axis at latitude of Mount Venus, calculated from the dips and throws of faults. B. Horizontal extension west of inner- floor axis at latitude of Mount Venus (similar to Fig. 13A). Notice that horizontal extension is asymmetric: 4 and 6 percent for west and east sides, respectively, of inner floor, and 11 and 18 percent for west and east inner walls, respectively. This is 1.0 2.0 3.0 4.0 50 consistent with sense of asymmetric spreading and suggests DISTANCE FROM AXIS OF INNER FLOOR (EAST) that asymmetric spreading is accomplished through asymmet- INNER FLOOR E. INNER WALL E. TERRACE ric extension as well as asymmetric crustal accretion.

RIFT MOUNTAINS cate earthquakes in the rift mountains be- that active normal faulting is occurring in cause their arrays were deployed right over the rift mountains in crust at least 1.5 m.y. The rift-mountain province begins out- the plate boundaries and their location old. This is consistent with studies of intra- side the median valley, just past the outer- range was 20 km at best. Francis and Por- plate earthquake focal mechanisms by wall boundary. Topography is still very ter's (1973) deployment of a single seis- Sykes and Sbar (1973) which suggest that rugged and is characterized on a large scale mometer in the rift mountains at lat 45°N the ocean crust is still under uniaxial ten- by rolling relief with a 6- to 12-km lasted only three days, and this seismometer sion out to about 20-m.y. age. In older wavelength (Fig. 10). As in the median val- recorded little or no activity in the moun- crust, they find that stresses appear to be ley, topography is dominated by block tains. However, the dramatic increase in compressional. faults with throws of as much as 300 m. density of outward-facing faults suggests The cumulative throw of inward- and Large block faults and series of faults are linear and continuous over the 4- to 8-km line spacing (Fig. 10). Smaller individual faults of 50- to 100-m throw are linear for at least 1 km, according to side-looking sonar data, but they do not appear on adja- 2000 cent traverses. Lineation direction is generally parallel to the N17°E strike of the valley. Major outward-facing blocks, which are g rare in the median valley, are common in the rift mountains. The average lateral density of outward-dipping faults with at least 1 75 m of throw increases from 0.5/10 km in 1000 the median valley to 2.5/10 km in the rift mountains. If the median valley is a I steady-state feature, then outward-facing ï faulting must be occurring outside the valley. Teleseismically located earthquakes do occur in the rift mountains in the FAMOUS area (Fig. 6). Reid and Macdonald (1973) and Spindel and others (1974) did not lo- 20 E Km DISTANCE: FROM AXIS Figure 9. East side of rift valley 2 (profile D-D"). Cumulative throw of inward-facing normal faults (dipping toward valley axis), outward-facing normal faults (dipping away from valley axis), and decrease in elevation due Figure 8. A. Graben formation to tilting of blocks. Solid black line indicates total tectonic contribution to through antithetic faulting due to relief [inward-facing fault throw minus outward-facing fault throw minus curved shape of main fault. B. Tilt- tilt (I - O - T)]. Faulting accounts for entire depth of median valley as well ing of faulted blocks due to shal- as for most large-scale relief (>2-km wavelength). If gross relief of median lowing of fault dip with depth. valley were largely of volcanic origin, then I — O — T curve would fall well (After de Sitter, 1964.) below observed topography. Volcanism contributes to short-wavelength roughness but contributes little to long-wavelength changes in elevation. Vertical exaggeration is 10X.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 STRUCTURE OF MID-ATLANTIC RIDGE CREST 629

outward-facing faults and the effect of tilt ternatively, the intensity of faulting and, the median valley and have similar dimen- were tabulated along traverses I—I' and perhaps, depth of the median valley may be sions and morphology. As in the rift valley, E-E' in the rift mountains (Fig. 11), similar greater now than it was 1 to 6 m.y. B.P. volcanic features contribute to small-scale to the analysis for the median valley (Fig. However, the apparent reduction in throw relief but are secondary to faulting in creat- 9). Once again we find that faulting and tilt- per kilometre in the rift mountains may ing large-scale relief. The only very large ing of crustal blocks accounts for most of very well be a problem of resolution. With volcanic feature is at —36 km (Fig. 10). the gross relief and nearly all of the regional inward- and outward-facing normal faults Magnetic studies indicate that this feature change in depth. In contrast to the median superimposed in the rift mountains, much and nearly all other volcanic topography valley, the cumulative throw of outward- of the tectonic relief appears symmetrical in were created in the median valley (Mac- facing faults is as great as that for inward- cross section and is difficult to distinguish donald, 1977). facing faults (Fig. 11). This contrast again from volcanic relief. Thus, in the rift moun- indicates active faulting in the rift mountains, the intensity of both inward- and SEDIMENT DISTRIBUTION tains. outward-facing faulting may be higher than Both outward-facing faulting and tilting shown. Photographic and side-looking sonar of the crust result in die decay of median- Some of the long-wavelength topography data suggest that most of the sea floor, even valley relief and increase in depth in the rift (6 to 12 km) appears to correlate with al- in the inner floor, is covered with a thin mountains. However, outward-facing fault- ternating groups in inward- and outward- veneer of sediment (Luyendyk and Mac- ing accounts for nearly 80 percent of the facing scarps, resulting in a large-scale un- donald, 1977). Outside the inner floor we decay of median-valley elevation, while tilt- dulating horst and graben terrain (Fig. 10). have tabulated sediment thickness using the ing of fault blocks accounts for only about Superimposed on the longer wavelength 4-kHz sediment-penetration system on the 20 percent (Fig. 11). This is surprising, be- faulted relief are equidimensional and fish (reliable for sediment thicknesses of 5 cause a tilt of only 6° to 7° could account slightly elongate volcanic highs. They occur to 100 m) and airgun reflection profiles for the cancellation of median-valley depth. almost as often in the rift mountains as in where sediment thickness exceeded 100 m. However, the average outward tilt of blocks in the median valley is 2.5° (SD=2.7° for OUTER WALL 126 samples) while that for the rift moun- WEST EAST tains is 3.8° (SD=2.5° for 64 samples). Clearly, the change in tilt is statistically in- significant and in any case is five times too small. Locally, however, tilt may be an important mechanism for the decay of topographic relief. For example, on several traverses across the outer walls, subsidence of crust is accommodated by tilting of 8° to 12° (Fig. 17, E-E', G-G'). However, outward-facing faulting is by far the most important mechanism for the decay of rift- valley elevation in the rift mountains. The slope of the line indicating decay of relief through outward-facing faulting in the rift mountains is essentially constant (Fig. 11). This may be important, for it suggests that most of the faulting occurs right at the median-valley—rift-mountain boundary. If faulting continued significantly farther out in the rift mountains, the slope of the line representing outward- facing faulting (that is, the density of 30 20 outward-facing faults) should continue to DISTANCE FROM AXIS (km) increase with distance from the axis, but this is not observed. This suggests that most of the median-valley relief is cancelled right near the edges of the valley in the first 10 to 20 km of the rift mountains. The throw per kilometre of inward- facing faults appears to be 30 to 40 percent less in the rift mountains than in the rift valley. This may indicate some reduction of throw of normal faults by reverse faulting along the original normal fault planes. As mentioned earlier, however, the compressional stresses required by reverse faulting Figure 10. A. Deep-tow geophysical traverse from rift valley 2 outer walls going into east rift would contradict available focal- mountains (top 3 profiles) and west rift mountains (bottom 2 profiles). Dashed lines = major faults; mechanism solutions (Sykes and Sbar, L = lip (volcanic feature perched at edge of block fault); V = other volcanic features. Numbers with 1973). Furthermore, there is no particular sediment ponds indicate maximum sediment thickness in metres. Vertical exaggeration is 2X to mini- reason why reverse faults should occur pre- mize distortion. B. Deep-tow geophysical profile 50 to 130 km into east rift mountains (see Fig. 1A for cisely along relict normal fault planes. Al- track location); symbols same as in A.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 630 MACDONALD AND LUYENDYK

2000 30 40 50 60 20 30 DISTANCE FROM VALLEY AXIS (Km, EAST) DISTANCE FROM AXIS (Km) Figure 11. A. Contribution of normal faulting and tilting of blocks to relief in east rift mountains on profile E, starting at rift outer wall (20 km). Crustal tilt is approximately same as in median valley, while outward-facing normal faulting increases dramatically and accounts for more than 80 percent of decay in median-valley elevation in rift mountains. Less than 20 percent is due to tilting. Again, faulting and tilting account for nearly all large-scale relief (I — O — T), while volcanism contributes to short-wavelength roughness in observed topography (bottom). B. Contribution of normal faulting and tilting of blocks in west rift mountains on profile I' —I", starting at west outer wall (12 km). Results are same as in A. Again, outward-facing faulting accounts for more than 80 percent of decay of median-valley elevation as opposed to less than 20 percent due to tilting.

Sediment thickness generally increases with distance from the valley axis (Fig. 12). Close to the axis, the increase in thickness with distance shows the same asymmetry as the spreading rates, the sediments being thinner on the east side because of the faster spreading half-rate. On both sides of the ridge, the increase in sediment thickness is highly erratic, unlike in the Pacific (Larson, 1971; Klitgord and Mudie, 1974). The sediment accumulates in ponds, mostly in faulted depressions, and on the back sides of tilted fault blocks (Figs. 2, 10). Appar- ently, sediment distribution is dominated by downslope transport. Ponding of sediments has also been observed at lat 22°N on the Mid-Atlantic Ridge (van Andel and Komar, 1969). Bottom currents of as much as 20 cm/sec (Keller and others, 1975), and a high -50 0 50 100 150 level of microearthquake activity (Reid and DISTANCE FROM AXIS (km) Macdonald, 1973; Spindel and others, 1974) on a rugged block-faulted terrain re- Figure 12. Sediment thickness as function of distance from valley axis sult in a thorough redistribution of sedi- (fromprofiles I—I", G—G', E-E"'). Note how sediment cover is very localized and spotty, yet still shows expected increase in thickness away from axis. ments downslope into ponds. For example, Within 30 km of axis, sediment cover is asymmetric, reflecting asymmetric on the west terrace the graben at —8 km spreading half-rates. (Fig. 2) has as much as 120 m of sediment, whereas the average sediment thickness of RIFT VALLEY 3 fracture zone B. However, the two valleys the terrace is about 20 m. This thorough re- are morphologically very different. Rift val- distribution of the sediments helps to ex- Rift valley 3, south of fracture zone B, ex- ley 2 is a "double valley" with a narrow plain the absence of internal reflectors on tends 30 km before being offset right later- inner floor and well-defined terraces. Rift the fish and airgun records and makes it dif- ally 12 km by fracture zone C (Fig. 1). Like valley 3 is essentially a single valley with a ficult to use displacement of sediment hori- rift valley 2, it strikes approximately wide floor and poorly developed terraces. A zons as detectors of tectonic movements. N17°E, oblique to the east-west trend of narrow terrace (5 km) can be traced on the

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 STRUCTURE OF MID-ATLANTIC RIDGE CREST 631

east side between the blocks at 10 and 15 1000 km, but there is no terrace on the west side. 1500 The median valley is approximately 24 km wide, extending asymmetrically from —9 to + 15 km (Figs. 13, 14; the center of the valley is taken as midway between the inner walls). The floor of the valley averages 11 km in width, five times the average width of the inner floor of rift valley 2. There is no well-defined valley axis marked by a distinct central high or central low. In cross section, the floor has four to six topographic highs with more than 100 m of relief (Figs. 13, 14). There are more than 35 30 INNER FLOOR I _ RIFT MOUNTAINS topographic highs in rift valley 3, some -INNER WALLS - more than 10 km long (Laughton and Rusby, 1975). Side-looking sonar records Figure 13. Bathymétrie profiles across rift valley 3, picked from a U.S. Navy bathymétrie chart using on our only deep-tow traverse indicate that multi-narrow-beam data (J. D. Phillips and H. S. Fleming, in prep.). Deep-tow profile S—S' is almost at these highs have lobate boundaries along same place as profile 3. Physiographic provinces are as shown. Notice very wide inner floor and poorly developed terraces, in contrast to rift valley 2. Vertical exaggeration is 2X. strike (Fig. 14). Their symmetrical profiles and lobate edges suggest that the highs are volcanoes. No single eruptive center or group of volcanoes dominates the morphology. Instead, the volcanoes seem to be randomly situated throughout the 11-km- wide floor. Mount Saturn (at +3.7 km, Fig. 14) appears to be the most recent site of extrusive volcanism in the floor. It is as high as 300 m and at least 3 km long (sonograph II—JJ, Laughton and Rusby, 1975), about the same size and shape as Mount Venus in rift valley 2. The 4.0-kHz sediment-penetration 0 4 system and side-looking sonar system indicate sediment cover as much as 12 m thick DISTANCE FROM AXIS (KM) in the valley floor, except between +2 and 5 SEDIMENT THICKNESS (M.) SCARP,STRIKE,AND EXTENT, +4 km in the vicinity of Mount Saturn. The ARROWHEAD MEANS UNEATION SEDIMENT COVER INFERRED EXTENDS BEYOND SONAR RANGE thin veneer of sediments (less than 1 to 2 m) FROM SIDE-LOOKING SONAR detected on the east flank of Mount Saturn STRIKE AND EXTENT OF A VOLCANIC FEATURE is consistent with downslope transport of i sediment from the nearby east walls. If STRIKE AND EXENT OF FISSURES Mount Saturn were an older feature created Figure 14. Deep-tow geophysical profile S-S' across rift valley 3. S Mount Saturn, most recent near the valley axis, the sediments should be site of volcanism. Vertical exaggeration is 2X. more symmetrically distributed on its flanks and thicker. Mount Saturn is also the most narrow. Reid and Macdonald (1973) lo- linear band through the transform valley highly magnetized feature in the valley cated a narrow band of microearthquakes (Fig. 6). A detailed chart of fracture zone B floor, suggesting that it is the freshest ac- in fracture zone A that ended abruptly at shows that the transform valley is divided cumulation of pillow basalt (Macdonald, the intersection of this fracture zone and rift by a north-south ridge between long 33°24' 1977). Thus, sediment and magnetic data valley 2 (Fig. 6). A subsequent tectonic map and 33°30'W. On the north side of the both suggest that the most recent site of ex- based on deep-tow data showed that these transform valley (lat 36°37'N), this north- trusive volcanism is well over on the east earthquakes occurred on two linear scarps south ridge is transected by a 1-km-wide side of the valley floor, almost on the east that also mark a discontinuity in sediment east-west trough. Within the 2-km accuracy wall. This suggests that the numerous vol- thickness. This single scarp system appears of location, the microearthquakes fall right canoes in the floor were not necessarily to define the current transform fault, a zone on this narrow east-west cleft, continuing erupted along the valley center line and that of active faulting less than 1 km wide within into the rift valley 2 intersection (Fig. 15B). volcanism may occur anywhere within the the 4- to 10-km-wide transform valley (Fig. Thus, in both fracture zones A and B, the 11-km-wide valley floor. 15A; Detrick and others, 1973). A swarm active transform fault is a narrow, approx- of seven microearthquakes was located at imately 1-km-wide boundary within a DISCUSSION the west end of the active fault (Fig. 15A). broad transform valley. Furthermore, The scarp and the microearthquake activity N17°E lineations associated with the Microearthquakes and Plate Boundaries abruptly stop at a point well into the inter- spreading plate boundary continue to section of rift valley 2 and fracture zone A. within 1 km of the active fault. This Although the transform fault valleys of The fracture zone B transform valley is suggests that the transform fault is a narrow fracture zones A and B are wide, the active from 10 to more than 15 km wide. Again, plate boundary zone, perhaps a single fault, transform plate boundaries are linear and the microearthquakes define a narrow, over short intervals of time. However, the

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 632 MACDONALD AND LUYENDYK

very existence of the broad transform val- tive appendage. It is relict from a period ence of neighboring fracture zones extends leys suggests that over millions of years, the when transform fault B was farther south throughout the entire length of the 45-km- transform fault may migrate perpendicular and was severed from rift valley 2 by the long rift. North or south of the midpoint of to its strike over a zone at least 10 to 15 km northward migration of fracture zone B. the median rift, the blocks immediately wide. Currently, the fracture zone A active plunge toward the nearest fracture zone. fault lies near the axis of the transform val- Tectonic Influence of Neighboring Small faults in the inner floor also cut across ley. The fracture zone B transform fault is Fracture Zones on Rift Valley 2 bathymetric contours, plunging toward the on the north side of the transform valley in nearest fracture zone. The faults do not a narrow trough. Where major block faults extend for tens bend around, following the contours, as the The transform-fault earthquake distribu- of kilometres, they plunge toward the frac- transform-spreading intersection is ap- tion has important implications for the me- ture zones. On the west side of the valley, proached. dian valley. The sharp intersections of the single major blocks in the inner and outer One possible mechanism for the bending active transform faults with rift valley 2 at walls can be traced over 20 km (Fig. 3). is lateral heat conduction across the frac- both ends suggest that the median-valley (This was not possible on the east side of ture zone between crust of different ages crustal accretion zone must currently be a the valley because of extensive relay fault- and a resulting anomaly in elevation of the well-defined zone less than a few kilometres ing in the inner wall and poor coverage of block edges. Lateral heat conduction and wide. The earthquakes do not extend very the outer wall.) The blocks plunge toward resulting elevation changes due to thermal far into the rift valley 3 intersection, where the fracture zones about a bending point contraction were computed at various the floor is five times wider than in rift val- midway between fracture zones A and B points across the fracture zones assuming a ley 2. This suggests that the crustal accre- (Fig. 16). This suggests that much of the vertical age boundary. A maximum of 1° of tion zone is not narrowly defined where the vertical relief of the fracture-zone valley is plunge can be accounted for this way, com- floor is wide, and the present spreading accommodated by north-south bending of pared to the 2.8° to 3.8° of plunge observed center may be on the east side of rift valley the rift-valley walls, as well as by normal (Fig. 16). Another possible mechanism is 3. Furthermore, the current position of the faulting along east-west faults bordering the mechanical coupling of lithosphere of fracture zone B transform fault and the fracture-zone valley (for example, at lat different ages, the older cooler lithosphere overlapping of rift valleys 2 and 3 suggest 36°54'N). More important, the plunging of holding down the edge of the adjacent that the south end of rift valley 2 is an inac- the blocks suggests that the tectonic influ- younger lithosphere.

33°16W 33°12' 33° 08' 33°04' 33°00' 32°56'w Figure 15A. Microearthquake epicenters on fracture zone A and its intersection with rift valley 2. Tectonic map from deep-tow data (after Detrick, 1974); epicenters from Reid and Macdonald (1973). Star = swarm of at least seven events located on west end of active transform-fault scarp in fracture zone A. Note that active seismic zone continues well into fracture zone A-rift valley 2 intersection, suggesting that spreading plate boundary in rift valley 2 is itself very narrow and is located near center of inner floor.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 STRUCTURE OF MID-ATLANTIC RIDGE CREST 633

Roles and Interaction of Volcanism and samples document the existence of recent too common to be accounted for by ran- Faulting in Rift Valley 2 flank eruptions (FAMOUS dive team, 1975, dom volcanism away from the inner floor. personal commun.); however, near-bottom The central highs are transported to either The greatest concentration of recent vol- magnetic studies suggest that they are of side by spreading and then uplifted by block canism lies near the inner-floor axis and is small volume and thickness relative to axial faulting, creating a lip at the edge of the represented by features such as Mount eruptions (Macdonald, 1977). scarp. Such lateral transport of the vol- Venus and Mount Pluto. Most of the recent The central volcanic highs are trans- canoes away from the median valley pro- volcanoes occur in a narrow band 1.0 to 1.5 ported out of the inner floor on block faults. vides a mechanism for extremely asymme- km wide which is relatively unfractured and These fossil central highs form prominent tric spreading on a small scale. Repeated unfaulted. Some volcanism occurs along the volcanic lips at the edges of fault blocks. As fracturing along one side of the volcanoes edges of the floor, with lava flows directed discussed earlier, the occurrence of lips in preferentially would create an asymmetry in toward the inner-floor axis. Submersible the valley and in the rift mountains is far spreading rate detectable in magnetic

FRACTURE ZONE B 33° 32' 30' 28' 26' 24'

36°38

Figure 15B. Fifty- fathom contours from U.S. Navy multibeam bathymét- rie chart of fracture zone B (J. D. Phillips and H. S. Fleming, in prep.), with epicenters from Reid and Macdonald (1973) superimposed. This map shows narrow east-west cleft of fracture zone B cutting across north-south ridge that bisects fracture zone B valley (Fig. IB). Note that epicenters fall right on this narrow cleft within 2-km location accuracy, suggesting that this is active transform fault trace. Cleft is only 2 km wide, compared to 20-km width of fracture zone B transform valley.

J L 0 KM

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 634 MACDONALD AND LUYENDYK

DISTANCE FROM FRACTURE ZONE B cal depressions. Thus, the walls should then 30 ~I— 20 have lobate outlines, which contradicts the striking linearity observed (Fig. 3). Alterna- tively, collapse of a linear caldera extending ki the length of the valley may occur. How- -WEST OUTER WALL ever, rift valley 2 consists of a number of volcanic centers and alternating central £

DISTANCE FROM FRACTURE ZONE A Expression of Asymmetry in Median Valley Figure 16. Longitudinal dip (or bending) of west inner and outer walls toward fracture zones A and B. Data from deep-tow profiles, with some supplementary data from U.S. Navy narrow-beam On a gross scale, one is impressed by the bathymétrie charts (J. D. Phillips and H. S. Fleming, in prep.). Dashed line shows distance midway symmetry of the Mid-Atlantic Ridge — its between fracture zones. Triangles show computed effect on elevation of lateral heat conduction across "mid"-Atlantic position between the conti- fracture zone near intersection of west inner wall and fracture zone A. nents, its symmetrical increase in depth, and its symmetrical spreading rate over long anomalies. Magnetic studies also indicate high to cause shear failure under uniaxial periods of time (Pitman and Talwani, that volcanism is extremely rare outside the tension, resulting in nonvertical fault dips. 1972). However, on a scale of kilometres or floor. Thus, morphologic, tectonic, and Alternatives to block faulting have been hundreds of thousands of years, it is dif- magnetic data (Macdonald, 1977) all proposed for formation of the valley walls; ficult to find any parameter that is symme- suggest that nearly all the volcanic topog- they include thrusting (Osmaston, 1971), trical. This asymmetry can be seen in (1) raphy is created in the inner floor of the val- construction by flow fronts, and caldera density of faulting in the inner floor, (2) the ley. collapse. Focal-mechanism solutions (for stair-step versus the massive-block nature Throughout the inner floor, topography example, Sykes, 1967) and the very steep of the east and west inner walls, (3) position is dominated by volcanism, while faulting is fault dips facing the valley axis make thrust- of the inner and outer walls relative to the secondary, acting primarily to fracture the ing unlikely. Formation of the walls by inner-floor axis, (4) width of the terraces, crust. At the inner-wall boundaries and stacking flow fronts would require tre- (5) fault dips, (6) sediment distribution, (7) beyond, however, block faulting dominates mendous volcanic flows originating from crustal extension rates, and (8) short-term the relief. Fault throws abruptly increase outside the inner floor. This is unlikely be- sea-floor—spreading rates. All these asym- from metres to hundreds of metres. Almost cause the most recent zone of large-scale metries are consistent with a skewness of the entire depth of the median valley is volcanism lies near the inner-floor axis. The sea-floor spreading and related tectonism created by block faulting (Fig. 9). Volcanic linearity and angular shape of the scarps and volcanism toward the east. The asym- topography is generally of smaller would also require unrealistic order in metries must from time to time reverse so amplitude and rides on top of the fault stacking the flow fronts. Francis (1974) has that the system is symmetric when averaged blocks. The dip of the faults, averaging ap- drawn an analogy between the 1968 Fer- over long periods of time. proximately 50°, is close to the average dip nandina caldera collapse in the Galápagos of large normal faults on land (de Sitter, Islands and the formation of the median- Oblique Spreading 1964). The dips are consistent with shear valley floor and walls. However, formation failure along the fault planes, whereas the of the inner walls through caldera collapse Precise mapping of rift valley 2 and the vertically sided gja in the inner floor indi- contradicts several observations. There is a adjoining transform faults shows that the cates failure under tension. The difference fine-scale, uniform increase in the density of rift is spreading obliquely (Fig. 3). The in dip may be due to depth of faulting. The faulting and in fault throw (Ballard and van trends of active transform faults are N90°E crust is intensely fractured, and thus the ef- Andel, 1977) with distance from the inner- for both fracture zones A and B, forming an fective pressure equals the lithostatic pres- floor axis. In addition, microearthquakes of oblique angle of 17° with the N17°E trend sure minus the hydrostatic pressure. (If the magnitudes —1 to +1 occur on a day-by- of the valley (Figs. 3, 15). A tightly con- crust were totally impermeable, the effec- day basis associated with apparently con- strained fault-plane solution on the tive pressure would equal the lithostatic tinuous faulting at the base of the inner Oceanographer Fracture Zone also shows plus the hydrostatic pressures.) For very walls. In contrast, the Fernandina caldera current east-west (N86°E) transform mo- shallow cracks and faults in the floor, the collapse was a highly episodic phenomenon tion (Sykes, 1967), even though the gross effective pressure is only a few bars, and involving an intense swarm of 295 earth- trend of the fracture zone is S75°E (Fig. 1). failure occurs under tension, resulting in quakes of magnitude 4 and greater. The col- The tectonic grain of the inner floor is vertical cracks and small steep step faults. lapse occurred over a period of only 11 days overwhelmingly N17°E even at a scale of The large faults of the inner walls may ex- and involved a 300-m subsidence of the metres (Luyendyk and Macdonald, 1977). tend 2 to 3 km into the crust (Weidner and floor, equivalent to half the relief of the Thus, even on a fine scale there is no indica- Aki, 1973). At such depths, the effective inner walls. Furthermore, this and other tion of readjustment within the rift to an or- pressure is nearly 1 kb. This is sufficiently caldera collapses result in circular or ellipti- thogonal system.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 STRUCTURE OF MID-ATLANTIC RIDGE CREST 635

Relationships between Rift Valleys 2 and 3 Tops of the faulted blocks are tilted, on REFERENCES CITED average, 2 to 3° away from the valley axis. The major difference between rift valleys Tilt as well as graben formation in the ter- 2 and 3 is the presence of wide terraces. races by antithetic faulting may be due to a Arcyana, 1975, Transform fault and rift valley geology from bathyscaph and diving saucer: Where terraces are well developed, the shallowing of fault dip with depth. Science, v. 190, p. 108-116. Approximately 80 percent of the trans- inner floor of the valley is narrow, 1 to 4 km Aumento, F., Loncarevic, B. D., and Ross, D. I., wide, as in rift valley 2. The volcanic zone formation of median-valley relief into rift- 1971, Hudson geotraverse: Geology of the appears to be confined to the narrow inner mountain topography is accomplished by Mid-Atlantic Ridge at 45°N: Royal Soc. floor. Where the terraces are narrow or normal faults that may dip away from the London Philos. Trans., ser. A, v. 268, p. nonexistent, the inner floor is wide, 10 to valley axis. Most of the outward-facing 623-650. 14 km, as in rift valley 3. Volcanic features faulting occurs right near the median- Ballard, R., and van Andel, Tj. H., 1977, Mor- occur throughout the wide inner floor, and, valley—rift-mountain boundary. Tilting of phology and tectonics of the inner rift valley in fact, the most recent volcanism appears crustal blocks accounts for only about 20 at lat 36°50'N on the Mid-Atlantic Ridge: Geol. Soc. America Bull., v. 88, p. 507-530. to have occurred well off to one side of the percent of the decay of median-valley relief. Ballard, R. D., Bryan, W. B., Heirtzler, J. R., Kel- inner floor. The non—steady-state nature of Long-wavelength topography in the rift ler, G., Moore, J. G., and van Andel, Tj. H., the terraces suggests that the median valley mountains is caused by groups of inward- 1975, Manned submersible observations in may vary between these two types of struc- and outward-facing faults. Volcanic relief the FAMOUS area, Mid-Atlantic Ridge: ture, perhaps in a cyclical manner. The var- accounts only for the short-wavelength to- Science, v. 190, p. 103-108. iation in median-valley structure and pography. Bellaiche, G., Cheminée, J. L., Francheteau, J. L., inner-floor width may have important ef- The common occurrence of "lips," vol- Francheteau, J., Hekinian, R., Le Pichon, fects on the width of the zone of crustal canic features at the edges of faulted blocks, X., Needham, H. D., and Ballard, R. D., formation and the recording of the Earth's indicates that much of the volcanic relief in 1974, Rift valley's inner floor: First submer- magnetic field in the crust. This problem is the terraces and rift mountains originated sible study: Nature, v. 250, p. 558-560. Bryan, W. B., and Moore, J. G., 1977, Composi- discussed by Macdonald (1977). within the rift inner floor. tional variations of young basalts in the Bending of large faulted blocks toward Mid-Atlantic Ridge rift valley near lat SUMMARY nearby fracture zones suggests that 36°49'N: Geol. Soc. America Bull., v. 88, p. spreading-center tectonics is affected by the de Sitter, L. U., 1964, Structural geology: New The Mid-Atlantic Ridge median valley fracture zones throughout the length of rift York, McGraw-Hill Book Co., 551 p. has a relatively constant width but may valley 2. Detrick, R. S., 1974, Fracture zone A, Mid- have either a very narrow inner floor (1 to 4 Both the crustal-accretion and Atlantic Ridge 37°N: A near-bottom km) and well-developed terraces or a wide transform-fault zones are narrow, only 1 to geophysical study: Scripps Inst. Oceanog- raphy Ref. 74-26,10 p. inner floor (10 to 14 km) and narrow or no 2 km wide, over short periods of time. Over Detrick, R., Mudie, J. D., Luyendyk, B. P., and terraces. The terraces appear to be non— millions of years, however, they apparently Macdonald, K. C., 1973, Near-bottom ob- migrate over a zone as much as 10 to 20 km steady-state features of the median valley. servations of an active transform fault: The entire depth and gross morphology wide. Mid-Atlantic Ridge at 37°N: Nature, v. 246, of the median valley may be accounted for p. 59-61. by block faulting. Volcanic relief contri- Francis, T.J.G., 1974, A new interpretation of the butes only to the short-wavelength (<2 km) ACKNOWLEDGMENTS 1968 Fernandina caldera collapse and its topography. implications for mid-ocean ridges: Royal As- Nearly every aspect of rift valley 2 is We are grateful to J. D. Mudie and F. N. tron. Soc. Geophys. Jour., v. 39, p. 301-318. asymmetric, consistent with faster spread- Spiess, who were chief scientists for parts of Francis, T.J.G., and Porter, I. T., 1973, Median valley seismology: The Mid-Atlantic Ridge ing to the east: local spreading rates, hori- cruise 31 of the R/V Knorr and who near 45°N: Royal Astron. Soc. Geophys. zontal crustal extension, fault dips, density contributed greatly to the success of the ex- Jour., v. 34, p. 279-311. of crustal fracturing within the inner floor, pedition. Capt. Hiller and his crew and the Francis, T.J.G., Porter, I. T., and McGrath, J. R., and location of the inner and outer walls. deep-tow group, including T. Boegeman, 1977, Ocean-bottom seismograph observa- Asymmetric spreading appears to be man- M. Benson, and J. Donovan were largely re- tions on the Mid-Atlantic Ridge near 37°N: ifested on a very fine scale rather than being sponsible for successful completion of the Geol. Soc. America Bull., v. 88, p. 664-677. accomplished by large ridge jumps. operations under difficult conditions. S. Heirtzler, J. R., and Le Pichon, X., 1974, FA- Morphology of the rift inner floor is Gegg, S. Miller, and D. Richards helped at MOUS: A plate tectonics study of the genesis dominated by volcanism. Intense faulting sea and on shore in reducing the data. S. of the lithosphère: Geology, v. 2, p. 273- 275. and fracturing occur within the inner floor, Miller, J. D. Mudie, and R. Groman as- Ivers, W. D., and Mudie, J. D., 1973, Towing a sisted considerably in computer work. For but fault throws are very small, contribut- long cable at slow speeds: A three- ing little to the relief. Detailed trends of stimulating discussions, we thank T. Atwa- dimensional dynamic model: Marine faulted and volcanic features within the ter, J. D. Mudie, K. Klitgord, K. Louden, Technology Soc. Jour. v. 7, p. 23-28. inner floor do not indicate any readjustment W. B. Bryan, Tj. van Andel, R. Ballard, J. Keller, G. H., Anderson, S. H., Koelsch, D. E., and to a north-south spreading center, ortho- Moore, J. Francheteau, P. Taponier, and P. Lavelle, J. W., 1975, Near-bottom currents gonal to the east-west fracture zones. Molnar. It was particularly useful and ex- in the Mid-Atlantic Ridge rift valley: Cana- Active crustal extension in the inner floor citing to compare our observations with dian Jour. Earth Sci., v. 12, p. 703-710. and inner walls has the same sense of those of the submersible diving teams. This Klitgord, K., and Mudie, J., 1974, The Galapagos spreading center: A near-bottom geophysi- asymmetry as the local spreading rates and research was supported by National Science Foundation Grants GA-36818 and 31377X cal survey: Royal Astron. Soc. Geophys. reaches a maximum of 18 percent in the Jour., v. 38, p. 563-586. and the Office of Naval Research. Mac- east inner wall. Asymmetric spreading ap- Larson, Roger L., 1971, Near-bottom geologic pears to be accomplished by asymmetric donald thanks the MIT/Woods Hole Joint studies of the East Pacific rise crest: Geol. crustal extension as well as by asymmetric Program in Oceanography for financial Soc. America Bull., v. 82, p. 823-841. crustal accretion. support during part of the program. Laughton, A. S., and Rusby, J. S., 1975, Long

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021 636 MACDONALD AND LUYENDYK

range sonar and photographic studies of the Geophys. Research, v. 80, p. 3307-3314. Porter, R. P., and Phillips, J. D., 1974, Mi- median valley in the FAMOUS area of the Moore, J. G., Fleming, H. S., and Phillips, J. D., croearthquake survey of median valley of the Mid-Atlantic Ridge near 37°N: Deep-Sea 1974, Preliminary model for extrusion and Mid-Atlantic Ridge at 36°30'N: Nature, v. Research, v. 22, p. 279-298. rifting at the axis of the Mid-Atlantic Ridge, 248, p. 577-579. Luyendyk, Bruce P., and Macdonald, Ken C., 36°48'North: Geology, v. 2, p. 437-440. Sykes, L. R., 1967, Mechanism of earthquakes 1977, Physiography and structure of the Needham, H. D., and Francheateau, J., 1974, and nature of faulting on the mid-ocean inner floor of the FAMOUS rift valley: Ob- Some characteristics of the rift valley in the ridges: Jour. Geophys. Research, v. 72, p. servations with a deep-towed instrument Atlantic Ocean near 36°48'North: Earth 2131-2153. package: Geol. Soc. America Bull., v. 88, and Planetary Sci. Letters, v. 22, p. 29—43. Sykes, L. R., and Sbar, M. L., 1973, Intraplate p. 648-663. Osmaston, M. F., 1971, Genesis of ocean ridge earthquakes, lithosphere stresses and the Macdonald, Ken C., 1975, Detailed studies of the median valleys and continental rift valleys: driving mechanism of plate tectonics: Na- structure, tectonics, near-bottom magnetic Tectonophysics, v. 11, p. 387—405. ture, v. 245, p. 298-302. anomalies and microearthquake seismicity Pitman, W. C., III, andTalwani, M., 1972, Sea- van Andel, Tj. H., and Komar, P. D., 1969, of the Mid-Atlantic Ridge near 37°N [Ph.D. floor spreading in the North Atlantic: Geol. Ponded sediments of the Mid-Atlantic Ridge thesis]: Cambridge and Woods Hole, Mas- Soc. America, Bull., v. 83, p. 619-646. between 22° and 23° North latitude: Geol. sachusetts Inst. Technology and Woods Ramberg, I., and van Andel, Tj. H., 1977, Mor- Soc. America Bull., v. 80, p. 1163-1190. Hole Oceanog. Inst., 248 p. phology and tectonic evolution of the rift Weidner, D. J., and Aki, K., 1973, Focal depth 1977, Near-bottom magnetic anomalies, valley at lat 36°30'N, Mid-Atlantic Ridge: and mechanism of mid-ocean ridge earth- asymmetric spreading, oblique spreading, Geol. Soc. America Bull., v. 88, p. 577- quakes: Jour. Geophys. Research, v. 78, p. and tectonics of the Mid-Atlantic Ridge near 586. 1818-1831. lat 37°N: Geol. Soc. America Bull., v. 88, p. Reid, I., and Macdonald, K. C., 1973, Micro- p. 541-555. earthquake study of the Mid-Atlantic Ridge Macdonald, Ken C., Luyendyk, B. P., Mudie, near 37°N using sonobuoys: Nature, v. 246, MANUSCRIPT RECEIVED BY THE SOCIETY NO- J. D., and Spiess, F. N., 1975, Near-bottom p. 88-90. VEMBER 26, 1975 geophysical study of the Mid-Atlantic Ridge Spiess, F. N., and Tyce, R. C., 1973, Marine Phys- REVISED MANUSCRIPT RECEIVED AUGUST 25, median valley near lat 37°N: Preliminary ob- ical Laboratory deep-tow instrumentation 1976 servations: Geology, v. 3, p. 211-215. system: Scripps Inst. Oceanog. Ref. 73-4, MANUSCRIPT ACCEPTED SEPTEMBER 13, 1976 McGregor, B. A., and Rona, P. A., 1975, Crest of 37 p. CONTRIBUTION NO. 3655, WOODS HOLE the Mid-Atlantic Ridge at 36°N: Jour. Spindel, R. C., Davis, S. B., Macdonald, K. C., OCEANOGRAPHIC INSTITUTION

Printed in U.S.A.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/88/5/621/3418363/i0016-7606-88-5-621.pdf by guest on 23 September 2021