Seismic imaging of crustal structure at mid-ocean ridges : a three-dimensional approach
Over recent years geological, geochemical and geophysical surveys of mid-ocean ridges have revealed a significant degree of along-axis variability not only in seabed morphology, but also in crustal structure, particularly Numerous geophysical surveys of the Valu Fa Ridge, southwest Pacific, have mapped the extent of an axial mid-crustal reflector. This reflector has been interpreted as representing the top of a sill-like melt lens, comprising a high percentage of partial melt, lying at the top of a crustal magma chamber. In 1995, a controlled-source, wide-angle seismic dataset was acquired at the Valu Fa Ridge during RN Maurice Ewing cruise EW9512, to investigate the mid-deep crustal structure at this ridge, and particularly the crustal magma chamber associated with the melt lens beneath the ridge axis. The EW9512 acquisition geometry was primarily two-dimensional in design, and modelling of these 2-D profiles revealed the presence of an axial low velocity zone beneath the melt lens. This low velocity zone is thought to represent a region of crystal mush comprising a much lower percentage of partial melt than is present in the overlying melt lens. Similar structures have been modelled beneath a number of other mid-ocean ridges. The primary aim of this study was to build on this 2-D interpretation by taking advantage of three-dimensional ray coverage in the axial region in order to assess the along-axis continuity of the magmatic system, correlate this to any ridge segmentation apparent in the seabed morphology, and determine if ridge segmentation is related to the magma supply. The 3-D data were analysed using a tomographic inversion technique. The inversion results suggest that the axial low velocity zone may be segmented on a scale of 5-10 km, which correlates with the morphological segmentation of the ridge crest and is believed to reflect episodic magma supply with different ridge segments at different stages of a cycle of magmatic and amagmatic extension. However, three- dimensional ray coverage is not ideal owing to the dominantly 2-D acquisition geometry. Therefore a detailed assessment of data uncertainty and resolution was undertaken to enable a meaningful interpretation of the inversion results in terms of which features have a geological origin and which are artefacts of the inversion process. P-S mode converted arrivals arising from mid-crustal interfaces were also modelled in order to obtain improved geological constraints on the crustal structure than is possible from P-wave studies alone. This modelling indicates that the uppermost crust is pervaded by thin cracks. In addition, techniques were developed for modeling the polarisation of 5-wave arrivals with low signal strength. Application of these methods suggests that the thin cracks have a preferred orientation parallel to the ridge crest on-axis, and oblique to the ridge crest off-axis which is thought to reflect the pattern of southward propagation of the ridge system inferred from regional tectonic and bathymetric studies. Modelling of P-S mode converted arrivals arising from conversion at the top of the melt lens provided additional constraints on the properties of the melt lens. In conjunction with the 3-D tomographic results, this work suggests that the southernmost ridge segment in the study area has recently become magmatically active following a period of amagmatic extension suggested by its morphology, thus providing evidence for episodic melt supply at this ridge. As part of the suggestions for further work, a theoretical investigation of survey resolution was undertaken to test commonly adopted acquisition geometries with a view to optimising the design and cost-effectiveness of future 3-D controlled-source tomographic experiments.