Journal offhe Geological Society, London, Vol. 153, 1996, pp. 185-189, 4 figs, 1 table. Printed in Northern Ireland Models of fracture orientation at oblique spreading centres G. W.TUCKWELL, J. M. BULL & D. J. SANDERSON Department of Geology, University of Southampton, Southampton Oceanography Centre, Southampton S014 3ZH. UK Abstract: Three models are proposed for the orientation of extensional faults and dykes at mid-ocean ridges based on their relationship to the ridge axis and the relative plate separation vector. These models predict four ridge geometries allowing them to be tested by orientation data from 17 different sites within the Earth's oceans. A transtensional model is shown to he generally applicable to the structure of mid-ocean ridges, with special conditions required for a departure from it. Spreading rate influences ridge geometry with intermediate and fast spreading ridges more likely to have an orthogonal spreading geometry, which is an end member case of the transtensional model. Keywords: Mid-ocean ridges, structural geology, transtension, spreading centres, faults. The observed structural style of constructive plate margins relative plate motion vector (Fig. 2b), see also Teyssier et al. varies greatly. Morphological variations with spreading rate (1995) who apply the sameconcept of transpression to are widely documented, e.g Carbotte & Macdonald (1994). convergentboundaries. Extensional fractures and normal Previous investigations into deformation at spreading faults (F), orientated normal to S,,, would form at an angle 8 centres have concentratedon fault facing direction (e.g. (Fig. 2b), oblique to both the relative plate motion vector Carbotte & Macdonald 1990; Shaw & Lin 1993). Fault (V) and the plate margin (P), where 8 =A/2.The orientation on the oblique spreading Reykjanes Ridge has orientation of these fractures may also be determined by a been studied (Murton & Parson 1993: Applegate & Shor simple graphical construction (McCoss 1986). Thusexten- 1994: McAllister et al. 1995) butthe causes of, andthe sional faults and dykes would be predicted to be oblique to constraints on, the orientation of the fracture distributions both the ridge and the plate vector. These predictions are are not well understood. Constructive plate margins display supported by experimental (clay) models of oblique rifting three important structuralelements: the relativeplate (Withjack & Jamison 1986). Transtensionaldeformation motion vector, the orientation of the plate margin;and the involves an area increase in the horizontal plane. If this is fabricproduced by extensionalfaults and fractures. The achieved by dykeintrusion no thinning of the crust need geometry can be described by four angles (Fig. 1): A, the occur, whereas if accompanied by extensional faulting there angle between the relativeplate motion vector andthe will be thinning (Sanderson & Marchini 1984). Extensional normal to the ridge; a = 90 -A, the angle between the plate fractures may be linked by transform faults (parallel to the vector andthe ridge axis: 8, the angle between the plate vector) or may form as en echelon segments along the extensional fracture (fault or dyke) and the ridge axis: and 4 ridge axis (geometry b). the angle between the platevector and thefracture The third model (Fig. 2c) assumes that the ridge axis acts orientation,as specified by Taylor et al. (1994) such that as a plane of weakness with intrusion of dykes parallel to the 4 = a + 8. Clearly only two angles are necessary: either A axis (8 = OO), and hence oblique to the plate vector (4 = a). or a, and either 8 or 4, but it is convenient to use various When the magma pressure (P,,,) exceeds the normal stress combinations of these in discussing each geometry. (a,) on a pre-existing fracture, intrusion will occur (Delaney et al. 1986). Since the minimum horizontal principal Models compressive stress (S,) is at (Y to the ridge We present three models for extensional faults and dykes at mid-ocean ridges based ontheir relationship tothe ridge axis and the direction of plate separation. The first model (Fig. 2a)comprises extensional structures orientated Clearly for a given stress system the smaller a, the greater orthogonal to the vector of plate separation (4 = 90'). In the magma pressure P, required to open a fracture parallel this model,oblique opening (that is when a <90) is to the ridge. accommodated by linking these extensional structures with Geometry (d) (Fig. 2d) is a special case predicted as an transform faults (geometry a). endmember byall three models, andrepresents the Sanderson & Marchini (1984) presenta model for 'classical' interpretation of ridge-transform geometry, where transtension, which involves oblique divergence across a spreading is normal tothe axis (a = 90", A = Oo) and zone of deformation, accompanying area increase in the extensional faults and dykes are parallel to the axis (8 = 0). horizontal plane being accommodated by vertical thinning of Thus it is only at obliquespreading axes thatthe three the crust. They show that the orientation of the horizontal models can be tested. principal compressive stress axes (S, > S,) will be oblique to These models raise three basic questions. (1) Are the both the displacement vector and the trend of the zone. To a resulting geometries supported by recent investigations of first order, a divergent plate boundary may be modeled as a oblique spreading axes? (2) Is there a relationship between transtensional zone, whose walls are displaced parallel to the ridge geometry andthe contributions of magmatization 185 Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/153/2/185/4889002/gsjgs.153.2.0185.pdf by guest on 26 September 2021 186 G. W. TUCKWELL ET AL. normal to plate margin Short segment oblique opening Relative plate motion Geometry (a) @=W vector (V) ,v transform offset non-transform oftset 'Extcnaional fracture8 (F) (parallel to SH) Fig. 1. The geometrical relationship between the main structural elements on constructive plate boundaries, may be described by four angles. A =the angle between the relative plate motion vector (v) and the normal to the ridge; a =the angle between the plate vector (V) and the ridge axis (P): 0 = the angle between the extensional fracture (F, fault or dyke) and the ridge axis (P): and 4 the angle between the plate vector and the fracture orientation. (dykes) and tectonics(fractures and faults) tothe overall extension? (3) Does spreading rate influence the ridge geometries? The validity of these models may be tested using orientation data from sites within the world oceans. i LP&F Data Data for the orientation and spreading rate from 17 ridge (d) axis areas (Fig. 3) within the Atlantic, Pacific and Indian Fig. 2. Three models for extensional faults and dykes at obliquely oceans are summarized in Table 1. Data froma previous diverging plate margins predict four possible plate margin compilation by Taylor et al. (1994) beenhas geometries. (a) Orthogonal opening on segments perpendicular to checked from the original sources and recalculatedwhere the relative plate motion vector predicts geometry (a) with short necessary. These data have been plotted in Fig. 4, which spreading segments connected by long transform faults. shows 4 against a with symbols indicating spreading rate. (b) Transtensional deformation with extensional fractures oriented The NUVEL-1 plate motion model of DeMets et al. (1990) parallel to the maximum principal compressive stress S,, as hasbeen used as the source of the relativeplate motion predicted by the model of Sanderson & Marchini (1984) predicts geometry (b). Segments may be connected by transform faults, or vector unless otherwise specified. may form en echelon arrays. (c) Oblique opening on fractures Within theNorth Atlantic the slow-spreading ridge oriented parallel to the ridge trend produces geometry (c). of system can be characterized by two the model geometries. (d) Geometry (d) is predicted as a special case by all three models The central North Atlantic sites (Ml, 27"N; M2, 37"N; M3, when the relative plate motion vector is orthogonal to the trend of 40"N) have approximate normalspreading (a= 4 = 90", the plate margin. geometry d) while sites furthernorth (ReykjanesRidge, ReykjanesPeninsula, NE Iceland,Tjornes Fracture Zone, and the Mohn's Ridge) where spreading is oblique to the between the Cocos and Nazca plates (Allmendinger & Riis ridge, all obeythe transtensionalmodel (4 = a/2 + 45", 1979). The relative plate motion vector may be calculated geometry b). from the NUVEL-1 plate model to be oriented at 005". The In the Indian Ocean, the Southwest Indian Ridge, with a ridge trends at 093", faulting is reported to strike parallel to half-rate of about 7mma-', follows the transtensional the axis and displays strong symmetry across it. Immediately model. Between Somalia and Arabia, the Gulf of Aden is a to the east of the Galapagos Ridge is the Ecuador Rift. The small ocean basin runningENE-WSW arising from the vector of plateseparation predicted from the NUVEL-1 separation of Arabia from Africa in the early Miocene plate model is 003", the ridge trends at 093", and Carbotte & (Laughton 1966). The spreading centre around 45"E has a MacDonald (1994) examined fault azimuth as a function of low relief, between 0.5 and 1.0 km, and small en echelon sea-floor age. Boththese areas also display normal ridges and valleys nested within it. Between 43.O"E and spreading (geometry d). 46.5"E (GA) the median valley is orientated oblique to the A right-stepping fault system connects the tectonics of spreadingdirection (Tamsett & Searle 1988). The Gulf of the Gulf of California with that of the San Andreas Fault Aden closely follows the transtensional model and has a system (Lominitz et al. 1970; Elders et al. 1972). The slow-spreading rate (10 mm a-'). Imperial Valley (IV) areacontains three right lateral The East Pacific Rise at 8"30'-10"N has a half-spreading strike-slip faults which bound two zones of transtension (Fig. rate of 58 mm a-'and faultstrike dataare taken from 3) with faults orientated at a low-angle to the strike of the Carbotte & MacDonald (1994).