Insights on the nature of the geophysical crust at a nearly amagmatic section of the ultra-slow-spreading Southwest Indian Ridge

So-called "normal oceanic crusts" from fast-spreading mid-ocean ridges (spreading velocity >80 mm/year) have been interpreted, hitherto, in terms of a three-layered magmatic crust - the Penrose model. As early as the 1960s, it was proposed that the mantle rocks, which are altered up to 70% may account for a wide range of lower crustal seismic velocities. Seafloor sampling and seismic tomography studies in parts of the slow-spreading Mid-Atlantic Ridge, especially at the end of spreading segments and the ultra-slow-spreading Southwest Indian Ridge, indicate that the magmatic crustal model may not adequately describe the observations in these domains, especially when the melt supply is relatively low. As early as the 1980s into the 1990s, the possibility of a crust forming when the magma supply is moderate began to receive intense attention. Various crustal models were proposed and were refined and are still being refined as new data becomes available. The Southwest Indian Ridge (spreading velocity 20 km along axis. Bathymetry and side-scan sonar images of the seafloor revealed the paucity of volcanic hummocks: features which typically characterize slow spreading centers. Earthquake hypocenters indicate that the dominantly amagmatic portions of the ridge are seismically quiet in the upper 15 km below the seafloor, compared to the more magmatic terrains. Hand in hand with this, tomographic images also reveal crustal thicknesses are less than 5 km, yet non-zero. These observations in this unusual ridge setting makes it a natural laboratory to study the crust forming when the melt supply is critically low. To explain the emplacement of mantle rocks over large distances on the seafloor, the conveyor was proposed to be large-offset normal faults (“detachment faults”) which may flip polarity consecutively. These faults have been recognized on the basis of surface geology, and are typically associated with oceanic core complexes, interpreted as the exhumed footwall of the detachment faults. The importance of these kind of faults at the Mid-Atlantic Ridge suggests that they contribute to shaping up to 50% of the seafloor morphology between 12°N and 35°N. But the nature of the faults at depth are not well constrained. Observations based on the locus of earthquake hypocenters at the Mid-Atlantic Ridge suggest that the faults may steepen with depth. Studying the nature of the faults at the ultraslow-spreading Southwest Indian Ridge and investigating the nature of the oceanic crust are at the heart of this thesis. Using 2-D multi-channel seismic datasets, I apply a 3-D processing scheme on a narrow cluster of 18 profiles dedicated to the crustal structure at the axial domain and longer lines extending off-axis to map possibly older detachment faults. Taking advantage of shooting with a powerful seismic source on three co-located lines, I merged these lines into one line by 3-D binning and it revealed significantly improved results compared to using a less powerful seismic source. From the data analyses, the detachment fault in the axial domain is revealed to consist of clusters of, sometimes, sub-parallel reflectors over a thick domain that is 2 km. I propose that this corresponds to the damage zone of the active detachment fault. The cluster of reflectors defining the damage zone is proposed to have formed either by strain localizing on a series of sub-parallel portions of the footwall simultaneously or strain localizing on individual portions consecutively over a thick domain. Individual portions of the reflectors are found to have dips between 45° and 60° and can be observed down to 5 km below the seafloor. Supporting this interpretation is a coincident ocean-bottom seismometer dataset, which indicates that the velocity structure and the velocity gradient in the proximity of the proposed damage zone is lower compared to other parts of the model (thicker geophysically-defined crust). This suggests that the hanging wall close to the fault is probably fractured. In the hanging wall of the active axial detachment fault (i.e., the axial valley basement), reflectors, dipping towards the active detachment fault, are projected to the seafloor and some of them correspond to mild elevations (<200 m), which can be interpreted as faults and linear volcanic ridges, probably serving as pathways for lava during eruptions. In the Antarctic plate, several reflectors are observed which I propose may correspond to damage zone of older detachment faults. I discuss a mechanism that may produce these older fault systems and draw a comparison with a conceptual model proposed for our study area, i.e., alternating north and south-facing faults. Several sub-horizontal reflectors are also observed in the Antarctic plate which may either correspond to magmatic intrusions in serpentinized mantle peridotites, or as interfaces between more serpentinized peridotite and less serpentinized peridotite. Whichever is likely, their long-term geometry may be influenced by footwall exhumation of successive detachment faults. From the results of the seismic data analysed in this thesis, I propose that the crust at the easternmost Southwest Indian Ridge (64 E) is likely (as proposed by Sauter et al. (2013); Cannat et al. (2006)) formed by pervasive detachment faulting. Volumetrically small episodes of melting may occur in the hanging wall and are transported off axis as the fault system develops offsets during exhumation. Finally, I made an effort towards improving the results of the ray-based travel-time tomography by applying a more robust waveform tomography which attempts to utilize the extra information contained in the waveform. Overall, the study reported in this thesis can be used both as an end-member case of melt-poor and therefore tectonically-dominated mid-ocean ridge accretion, and as a reference for the study of other divergent margins contexts where the melt supply was (or is) poor, such as magma-poor ocean-continent transition zones (e.g., Iberia-Newfoundland Margin) or melt-starved regions of the ultra-slow spreading Gakkel Ridge.