* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Do fracture zones define continental margin segmentation
Survey
Document related concepts
Post-glacial rebound wikipedia , lookup
Anoxic event wikipedia , lookup
History of geology wikipedia , lookup
History of geomagnetism wikipedia , lookup
Hotspot Ecosystem Research and Man's Impact On European Seas wikipedia , lookup
Oceanic trench wikipedia , lookup
Seismic inversion wikipedia , lookup
Great Lakes tectonic zone wikipedia , lookup
Magnetotellurics wikipedia , lookup
TaskForceMajella wikipedia , lookup
Abyssal plain wikipedia , lookup
Large igneous province wikipedia , lookup
Transcript
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy Earth and Planetary Science Letters 272 (2008) 553–566 Contents lists available at ScienceDirect Earth and Planetary Science Letters j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e p s l Do fracture zones define continental margin segmentation? — Evidence from the French Guiana margin C.J. Greenroyd a, C. Peirce a,⁎, M. Rodger b, A.B. Watts b, R.W. Hobbs a a b Department of Earth Sciences, Durham University, Science Laboratories, South Road, Durham, DH1 3LE, UK Department of Earth Sciences, University of Oxford, Parks Road, Oxford, OX1 3PR, UK A R T I C L E I N F O Article history: Received 24 July 2007 Received in revised form 16 May 2008 Accepted 19 May 2008 Available online 30 June 2008 Editor: C.P. Jaupart Keywords: crustal structure continental margins oceanic crust fracture zones A B S T R A C T Plate reconstructions suggest that the French Guiana margin in the west equatorial Atlantic is a highly segmented margin with both rift- and transform-style features. We describe here the results of modelling coincident multi-channel and wide-angle seismic, gravity and magnetic data acquired along two transects of this margin. The resulting models not only highlight the degree of structural segmentation but also demonstrate the effect of trans-tension on margin evolution. As a whole, the margin is characterised by 35–37 km thick pre-rift continental crust which is separated from unusually thin oceanic crust (3–4 km) by thinned continental and/or transitional regions. To the north, the margin exhibits a 320 km wide zone of thinned continental crust adjacent to a narrow ocean–continent transition, and is interpreted as a transform margin where the wide zone of thinned crust is a result of profile orientation being highly oblique to the direction of rifting. Approximately 240 km to the south, the margin is characterised by a 70 km wide zone of thinned continental crust which is wider than typical for transform, and narrower than typical for rifted margins. This crustal structure is interpreted to reflect a “leaky” transform formed by trans-tensional extension. These observations suggest that fracture zone influenced geometry of equatorial Atlantic rifting, did not produce a well-defined margin crustal structure, but instead resulted in margin segments which display characteristics of both rift and transform tectonic processes. The associated abundance of fracture zones has likely also affected the post-rift evolution of the margin, and provided topographic basement highs which acted as sediment dams to the northwards flux of sediment from the Amazon. © 2008 Elsevier B.V. All rights reserved. 1. Introduction A diverse range of rift-related structures are observed at the passive continental margins of the Atlantic; a consequence of variation in mantle temperature, rate and extent of rifting, rift geometry and lithospheric thickness. Observed structures have resulted in margins being classified according to: the orientation of the rifting direction relative to that of subsequent oceanic spreading – rift to transform (e.g. Whitmarsh et al., 1996; Edwards et al., 1997); the degree of rift-related magmatism – volcanic to non-volcanic (Mutter, 1993); the distance over which continental crustal thinning occurs – wide to narrow (Davis and Kusznir, 2002); and whether or not transition zones of exhumed mantle or intruded crust are observed (Dean et al., 2000; Hopper et al., 2006). While the two-dimensional aspects of the mode of formation of these structural end-members is relatively well understood, to better understand their three-dimensional inter-relationships, or the margin segmentation, study of alongstrike continuity of features is required, together with correlation of features between conjugate margin pairs. ⁎ Corresponding author. E-mail address: [email protected] (C. Peirce). 0012-821X/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2008.05.022 Large-scale segmentation is observed along the margins of the eastern Atlantic, e.g. in their magmatic characteristics which vary, north to south, from volcanic (Rockall – Morgan et al., 1989) to nonvolcanic (Iberia – Dean et al., 2000; Congo-Zaire-Angola – Contrucci et al., 2004) to volcanic (Namibia – Bauer et al., 2000). However the nature of the along-strike transition between these end-member types is poorly known. Similarly, these margins are structurally segmented by rift- and transform-style structures; the latter related to long-lived first- and second-order offsets in the Mid-Atlantic Ridge (MAR) which can be traced to each margin as fracture zones or fracture-zone-like features. The relatively narrow width of these zones suggests that the transition from rift- to transform-style structures is similarly abrupt. However, to date, existing studies have focused on better understanding of the origin and evolution of end-member margin characteristics, rather than addressing the nature of the transition between them. In order to understand the origin and development of along-strike segmentation, the western equatorial Atlantic margin was chosen for study because the satellite-derived gravity anomaly indicates that the crustal basement fabric is dominated by fracture zones (Fig. 1), whose intersection with the margin is reflected in the free-air gravity anomaly as small offsets in the ‘edge effect’ high. In addition, this Author's personal copy 554 C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 Fig. 1. French Guiana-Northeast Brazil margin. Top: Lineations observed in the first derivative of the satellite-derived gravity FAA (Sandwell and Smith, 1997) showing the locations of fracture zones which can be traced from the MAR to the continental margin. Middle: Summary of the observed fracture zones traced (black dashed) and interpolated (black dotted) towards the margin. Bottom: Müller et al.'s (1997) seafloor age isochrons, with heavy contours marking 30 Ma intervals. The location of Profiles A and D at the French Guiana margin are annotated. Author's personal copy C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 margin has extensive multi-channel seismic data coverage (both hydrocarbon industry and academic – Fig. 2), and several whole crustal wide-angle seismic transects acquired by the Amazon Cone Experiment (ACE – Watts and Peirce 2004). Here we extend the separate interpretations of ACE seismic and gravity profiles acquired across the French Guiana margin (Fig. 3), to develop a geodynamic model of margin evolution and investigate along-margin structural segmentation. The acquisition and modelling of the data, upon which this study is based, are described in detail in Greenroyd et al. (2007a,b). 2. Regional tectonic setting The passive continental margins of French Guiana and its conjugate offshore Liberia, are unique amongst the Atlantic margins 555 as they represent the point of final continental break-up and opening of the Atlantic Ocean. Plate reconstructions (Unternehr et al., 1988; Nürnberg and Müller 1991 – Fig. 4) and commercial seismic data (Pereira da Siva, 1989; Mello et al., 2001; Cobbold et al., 2004) show that French Guiana and Northeast Brazil rifted from Guinea, Sierra Leone, Liberia, Ivory Coast and Ghana during the early Cretaceous at ~110 Ma; this final rifting event being preceded by the formation of both the North and South Atlantic. The tectonic evolution of the equatorial Atlantic has been highly influenced by a series of transform faults and fracture zones (henceforth referred to as transform features) which cross-cut the entire ocean basin, evident in the gravity anomaly (Fig. 1) and seabed bathymetry (Fig. 2). The intersection of these transform features with the MAR is associated with large lateral offsets in the ridge trend, which accommodate a total offset of over 3000 km within the Fig. 2. Bathymetry of the western equatorial Atlantic (Sandwell and Smith, 1997), showing the location of large-offset transform faults along the MAR. Also highlighted are prominent bathymetric features: the Demerara Plateau to the north; and the Amazon Cone deep-sea fan system and Ceara Rise aseismic ridge to the south. The coincident MCS and WA profiles that form the basis of this study (A and D) are highlighted in red, while the remaining seismic profiles (B to G) which comprise the Amazon Cone Experiment, are also shown. The red dashed lines surrounding Profiles A and D mark the extent of the lower panels which show the locations of OBSs (red triangles), OBHs (blue triangles) and land stations (green triangles). The green box adjacent to Profile D shows the location of Erbacher et al.'s (2004) seismic stratigraphic reference profiles which are shown by the solid green lines in the corresponding inset. DSDP and ODP wells are marked by red and blue stars respectively. The Guyaplac profiles, traversing the margin between Profiles A and D, are marked by thin blue lines. See text for details. Bathymetric contours are plotted at 20 m, 50 m, 100 m (dashed line), 500 m (dotted line) and then at every 1000 m (solid line) intervals. Author's personal copy 556 C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 Fig. 3. Models of the French Guiana-Northeast Brazil margin (after Greenroyd et al., 2007a,b). Simplified illustrations of the interpreted crustal units are shown above and below corresponding models. Triangles mark OBS/H locations (see Fig. 2). Velocities are colour-coded in km s− 1, where blue colours represent sedimentary layers, dark green the upper crust, and light green the lower crust. P-wave velocities were converted into density and the free-air gravity anomaly calculated (solid red line) for comparison with that acquired whilst seismic surveying (dashed) for both profiles. The satellite-derived (Sandwell and Smith, 1997), longer-wavelength anomaly is included (blue). Model parameters are summarised Greenroyd et al. (2007a,b). Upward arrows locate fracture zone intersections. Interpreted crustal units are annotated: CC — continental crust; TCC — thinned continental crust; TZ — transition zone crust; OC — oceanic crust. See text for details. equatorial Atlantic region. Generally, where these transform features intersect with the continental margin it would be expected that the margin is characteristically transform in type, while in between riftstyle characteristics predominate. Further evidence for this segmented pattern is provided by studies of the West African margin where adjacent rift (Liberia – Mascle 1976; Ivory Coast – Peirce et al., 1996) and transform (Ghana – Edwards et al., 1997) margins are observed. Post-rift, the equatorial MAR appears to have been slowspreading, although estimates of half-spreading rate vary between 9 mm yr− 1 (Le Pichon and Hayes, 1971) and 28 mm yr− 1 (Nürnberg and Müller, 1991). At ~ 80 Ma, spreading was accompanied by the formation of the Ceara Rise offshore Northeast Brazil and the Sierra Leone Rise offshore West Africa; two oceanic plateaus which, given their conjugacy, were most likely formed at or close to the MAR (Kumar and Embley, 1977). Between ~ 115 and ~ 100 Ma syn- to postrift, coarse clastic sediment sequences were deposited in the region adjacent to the Amazon Cone (Fig. 2) which were later overlain by post-rift fan-delta and platform carbonate sequences (Brandão and Feijó, 1994). A major increase in sediment flux from the Amazon River occurred at ~ 10 Ma (Damuth and Kumar, 1975; Supko et al., 1977) as a result of the uplift and erosion of the Bolivian Andes. These vertical movements altered drainage patterns across South America and resulted in the catchment area for the Amazon River extending across most of the continent. The overall result of this Author's personal copy C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 557 Fig. 4. Plate reconstruction of the equatorial Atlantic (after Nürnberg and Müller,1991). In this reconstruction the North Atlantic rifted prior to 118.7 Ma with rifting in the South Atlantic progressing northward at this time. The equatorial Atlantic was the last region of the Atlantic to open at ~ 110 Ma. The continents of South America and Africa are denoted by + symbols. Plates are labelled: AFR — southern Africa; NWA — northwestern Africa; SAM — South America; PAR — Parana; SAL — Salado; and COL — Colorado. deposition is a sediment column N8 km thick offshore Northeast Brazil (Houtz et al., 1977; Rodger et al., 2006). To the north, offshore French Guiana, the bathymetric trend observed along much of the margin alters from a relatively narrow continental shelf and steep continental slope dipping towards the deep abyssal plain, to a wide, gradual sloping shelf terminated by a near-vertical slope. The latter is the Demerara Plateau, which Benkhelil et al. (1995) suggest, with its conjugate the Guinea Plateau, Author's personal copy 558 C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 evolved by trans-tensional rifting and was the last part of the margin to break-up. Gouyet et al. (1994) conclude that pre-rift sedimentary deposition occurred in an inner shelf environment and that post-rift deposition was open marine. Sediment lithology and seismic stratigraphy of the Demerara Plateau have been obtained from boreholes and site survey data (ODP Sites 1257–1261 – Erbacher et al., 2004) and this is used as the primary stratigraphic reference for the ACE. Gouyet et al. (1994) summarise the structures observed, and divide its geological history into two main stages: 2.) Post-Aptian (113 Ma–present) — Initiated during the Albian (112– 97 Ma), the break-up between the Demerara and Guinea Plateaux represents the final opening of the South Atlantic. This break-up incorporated a combination of perpendicular rifting and dextral shearing, segmenting the French Guiana and northeast Brazil margin into a series of rift- and transform-type structures. Postrift, the depositional environment shifted from shallow to open marine sediments. 3. Datasets 1.) Liassic to Aptian (213–113 Ma) – Prior to the equatorial rifting of South America and Africa at 118.7 Ma, the Demerara and Guinea Plateaux were adjacent parts of the southern Central Atlantic margin (Benkhelil et al., 1995 – Fig. 5). Sedimentation at that time occurred primarily in an inner shelf environment, with significant continental influxes. The Neocomian period (145.6–131.8 Ma) also represents the start of the progressive northward opening of the South Atlantic Ocean (rifting of Argentina and South Africa); and As part of the ACE, seismic transects across the continental margin of French Guiana were acquired, together with bathymetry, gravity and magnetic data (Watts and Peirce, 2004). This seismic subset comprises two coincident multi-channel and wide-angle seismic (MCS and WA) profiles, A and D, and example data sections can be found in Greenroyd et al. (2007a,b). Here we interpret the results from Profiles A and D in terms of the mode of evolution of the French Fig. 5. Initial rift geometry prior to opening of the equatorial Atlantic, for the Demerara Plateau and adjacent areas offshore French Guiana (after Benkhelil et al., 1995). This model suggests that the initiation of continental break-up is a result of trans-tensional motion between the African and South American plates during the Early Cretaceous which resulted in the inception of seafloor spreading by the Late Cretaceous. Locations of the two profiles modelled in this study are shown by solid lines. Author's personal copy C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 Guiana margin, and if and how the along-margin pattern of segmentation reflects its evolution. Further details of the dataset and modelling approaches can be found in Greenroyd et al. (2007a,b), and profile location is shown in Fig. 2. Several other datasets have been acquired in this region complementing the ACE, which collectively enable a comprehensive evolutionary history of the margin to be determined. Fig. 2 shows the location of the datasets which have been used to define sedimentary and structural characteristics of the region (industry data – Gouyet et al., 1994), and to develop an understanding of the 3D distribution of sediment and the orientation and extent of basement structures (Guyaplac data – F. Klingelhöfer and W. Roest pers. comm., with 2D profile interpretation by the French Petroleum Institute). The Guyaplac survey was conducted over the French Guiana continental margin as part of the French Extraplac (e.g. Loncke et al., 2006) programme. Of the MCS profiles acquired, ten are margin transects, oriented northeast-southwest, and one is oriented east-west (Fig. 2). These profiles, therefore, provide a pseudo-3D structural reference framework extending from ACE Profile D to, and ~100 km beyond, Profile A. Collectively, the MCS data image the sediment column and basement across most of the margin and the corresponding interpretation suggests that oceanic crust is observed oceanward of the toe of the Demerara Plateau, which is interpreted as comprising sediments overlying volcanic sills and intrusives. Significant faulting of the oceanic crust is also observed, resulting in a rough basement surface. Some faulting is identified within the sections, primarily on the Demerara Plateau itself, although also in the abyssal sediments. More extensive faulting is observed at the basement surface. In addition, several regions of basement are interpreted as volcanic sills and intrusions. In particular, two sharp rises in the basement surface are observed at the toe of the Demerara Plateau on Profile D and are interpreted as volcanic in nature. Whilst a detailed 3D description of the basement is not possible with the MCS data available, a pseudo-3D description can be approximated from the 50 km separated Guyaplac 2D profiles. Thus, only large scale features crossing the study area can be mapped. However, given the general trend of lineations in this region (Fig. 1), significant fracture zones are likely to be among these. 4. Summary of previous modelling The modelling of the seismic and gravity data from Profiles A and D is described in Greenroyd et al. (2007a,b). However, a brief summary of results and a comparison of similarities and differences between the two are included here (Fig. 3). Both models comprise nine subsurface interfaces which vertically partition the models as follows: an intra-water column thermocline; the seafloor; four intra-sediment boundaries; the crustal basement; an intra-crustal boundary; and the Moho. The depth to and velocity between these interfaces also varies across the models such that they can be subdivided laterally into regions characteristic of the continental and oceanic crusts, which are further divided into Upper Crust and Lower Crust and Layer 2 and Layer 3 respectively, to be compatible with the standard models of oceanic and continental crustal structure (e.g. Spudich and Orcutt, 1980; Bratt and Purdy, 1984; Christensen and Mooney, 1995). 4.1. Profile A Profile A exhibits 35–37 km thick pre-rift continental crust which thins by a factor of ~6.4 over a distance of ~ 70 km. The P-wave velocity within the Upper Crust ranges from 5.6 to 6.0 km s− 1 while that of the Lower Crust ranges from 6.4 to 6.7 km s− 1 although ray coverage within this region of the model is sparse compared to that further offshore for the OBS/Hs. Oceanward of the thinned continental crust is a 45 km wide region interpreted as a transition zone between continental and oceanic crust. 559 Further oceanward the crust is characteristically oceanic in style, with a hummocky basement reflection imaged in the MCS data and a two layer velocity structure. Layer 2 velocities range from 4.6 to 5.7 km s− 1 and Layer 3 velocities range from 6.4 to 7.5 km s− 1, both noticeably different from the adjacent continental crust. The combined Layer 2 and Layer 3 thickness is just 3-4 km which, compared to the global average of 7.1 ± 0.8 km s− 1 (White et al., 1992), is remarkably thin. 4.2. Profile D Profile D also exhibits 35–37 km thick pre-rift continental crust. The continental crust here thins over a total distance of 317 km, although the thinning occurs mainly in two regions; over a wide zone between 70 and 235 km offset where the crust thins from 34.4 to 21.7 km thickness and over a narrower zone between 320 and 387 km offset with thinning from 21.0 to 10.6 km. P-wave velocities are similar to those observed along Profile A although modelling suggests a lower velocity in the shallow crust. The Upper Crust velocities vary from 4.2 to 6.2 km s− 1 and the Lower Crust from 6.2 to 6.9 km s−1. The wider zone of crustal thinning is interpreted to result from profile orientation in a direction sub-orthogonal to that of rifting, while the more rapid, narrower, oceanward zone of thinning is interpreted to reflect intersection with a transform feature. No transitional crust is observed which suggests a very narrow ocean–continent transition (OCT) at this location. The interpretation of oceanic crust oceanward of 387 km offset is based on the hummocky nature of the MCS basement reflection and also the two layer crustal velocity structure suggested by WA modelling. Layer 2 velocities range from 4.3 to 6.2 km s− 1 and Layer 3 from 6.4 to 7.4 km s− 1, similar to those modelled along Profile A. Again, the combined Layer 2 and Layer 3 thickness is just 3–4 km, suggesting that this may be a margin-wide feature. 4.3. Crustal thinning and the ocean–continent transition The two profiles, despite being separated by only ~240 km alongstrike and both being oriented similarly relative to the coastline, exhibit dramatic differences in their style of continental crustal thinning. While the crust along Profile A thins relatively sharply over a zone ~ 70 km wide, the crust along Profile D thins over ~ 317 km, although much of this thinning occurs adjacent to the OCT. This variation in width of the zone of thinning (the margin width) alone demonstrates significant structural variation exists along this margin. In addition, the characteristics of the adjacent OCT also vary dramatically between profiles from being a narrow, sharp boundarylike feature along Profile D, to being a transition zone up to 45 km wide along Profile A. Both of these features are interpreted to demonstrate the role played by transform features not only in margin evolution, but also as a primary control on along-margin segmentation. For Profile A, a mode of margin evolution is proposed in which a component of transtensional extension results in a margin with typical extensional characteristics, but where the margin width and OCT are considerably narrower than would be expected for a rifted margin. For Profile D, the model is interpreted to reflect a multi-component evolution in which large-offset transform faults, bound purely extensional, rifted regions. The structures imaged on Profile D thus reflect the oblique trend of this transect across the extensional part of the margin relative to the direction of rifting, and its intersection with the much narrower transform-type margin which is interpreted to bound the Demerara Plateau to the north. 5. Model resolution The χ2 fit to the data and the resolution on each interface and layer velocity are tabulated in Greenroyd et al. (2007a,b). Subsequent Author's personal copy 560 C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 testing of model resolution and uniqueness has been undertaken via independent inversion of (Fig. 6), and a statistical analysis (Fig. 7) of model fit to observed traveltimes. An inversion-based approached was chosen, as it also provides an opportunity to assess modeller bias inherent in the forward approach to seismic and gravity modelling used in Greenroyd et al. (2007a,b). An additional source of uncertainty within the models is their primary dependence on seismic data. Thus, further tests of the uniqueness of the resulting models have been undertaken using the magnetic anomaly (Fig. 8). In Greenroyd et al. (2007a,b), gravity modelling provides additional constraint on the variation in crustal thickness and Moho geometry beneath the continental slope and shelf where the ray coverage is limited and, consequently, the P-wave velocity-depth models are poorly constrained. Modelling of magnetic anomaly data is used here primarily to identify true oceanic crust, and consequently better define the location of the OCT, and identify any lineations which might reflect the location of fracture zones. 5.1. Inverse modelling For tomographic inverse modelling of the WA data, tomo2d was used (Korenaga et al., 2000). Using Profile A to demonstrate the outcome, the inverted P-wave velocity-depth model (Fig. 6) matches the observed data to a χ2 of 1.26 and clearly shows a velocity-depth Fig. 6. Results of inverse modelling of the observed traveltimes for Profile A. See text for details. A simple starting P-wave velocity-depth model (top) was inverted using OBS first arrival traveltimes to produce a best-fit velocity model (centre). 1D velocity-depth profiles (bottom) through the inverted model (blue) and the forward model (black) are compared. The velocitydepth profiles are calculated at 250, 300 and 350 km offset, a region identified from the MCS data as most likely oceanic in nature. The profiles show distinct similarities between the velocity models produced by the forward and inverse modelling techniques, although the inverse model is more inherently smoothed. OBS locations are highlighted by red triangles. Author's personal copy C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 561 Fig. 7. Results of Metropolis uncertainty analysis on Profile D. Top: The distribution of model boundaries produced by the Metropolis algorithm, where 66% of the boundaries lie within one standard deviation (dark grey) and 86% within two (light grey). The input model (red lines) is the best-fit forward model, sub-sampled every 10 km. Bottom: One standard deviation velocity errors — where 66% of models created by the Metropolis algorithm lie within the mean plus or minus the velocities shown. profile in which velocity increases with depth. Within the region of dense ray coverage beneath the OBSs, the correlation between the inverse and forward modelled velocity profiles is good (cf. Figs. 3 and 6), with the inverse model being a highly smoothed version of the forward model with no obvious discrepancies. Thus, for Profile A, the inverse modelling is complimentary to the forward modelling, suggesting that between 190–390 km offset the forward model is relatively independent of modeller bias. 5.2. Metropolis To further assess the resolution of the P-wave velocity-depth models, the Metropolis approach was adopted (Pearse, 2002; Tarantola, 2005; Hobbs, 2006). The algorithm is not an inverse technique, although it is computationally intensive and modeller independent, and is particularly suited to this problem as it generates random samples from a probability distribution that is difficult to estimate directly. The approach is, in effect, a more detailed version of the forward modelling approach to resolution testing described in Greenroyd et al. (2007a,b), which is able to test lateral and vertical velocity and depth variations simultaneously. Metropolis is designed to quantify uncertainty within a model by extensively testing models close to the modeller-defined, best-fit, final forward model, by applying small, random perturbations to that model and assessing each resulting model in terms of the statistical fit to observed traveltimes, effectively ‘searching’ the model space around the final forward model. As a consequence, this approach extracts quite different information from the data when compared with the inverse modelling approach. While the inverse modelling has shown that the large-scale structural features of the models are not manifestations of modeller bias, this approach will test the resolution of these features. If, for example, a particular layer is poorly resolved due to a lack of traveltime picks, then a variety of models, each with a slightly different velocity or depth structure, may also fit the data. As part of the Metropolis approach, approximately 20,000 models were created for Profiles A and D, and from these the statistical mean and standard deviation (σ) were calculated. Fig. 7 demonstrates an example of the results for Profile D, with error estimates for the model boundaries and the P-wave velocities, and the best-fit forward model lying approximately in the centre of the suite of models created. The uppermost layers are well constrained across the model. Errors increase with depth, but are not excessively large and at the Moho the Author's personal copy 562 C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 Author's personal copy C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 1σ error is ±0.2–0.5 km. These errors increase landwards, partly as a result of the lack of MCS control on the basement surface and partly due to the lack of ray coverage at the most landward OBS. Similarly, the velocities are well constrained and 1σ errors are estimated at 0– 0.2 km s− 1 in the sediment layers and 0.2–0.4 km s− 1 in the underlying crust. Errors in the mantle appear slightly banded, primarily as a result of the reduction in ray coverage at depth. The banding also highlights an issue regarding how model parameterisation can affect the results. A typical upper mantle ray path is 20–60 km in length, whereas the model is laterally sampled every 10 km. The banding therefore shows that the Moho boundary and upper mantle velocity may be over sampled, resulting in instabilities on a wavelength of ~20 km. Hence, this demonstrates that the ray-trace approach to modelling cannot resolve anomalies in the mantle of less than ~ 20 km in size. 6. Magnetic modelling and crustal lineations Magnetic data offer a non-seismic approach by which to constrain crustal properties at a continental margin. In previous studies, crossstrike variations in crustal magnetization have been observed at several margins (e.g. Ghana – Edwards et al., 1997; Nova Scotia – Wu et al., 2006). They are generally used to locate magnetized oceanic crust and also to assess spreading rate with respect to magnetic field reversals. However, changes in magnetization may also be associated with serpentinization and the presence of fracture zones (Lin et al., 2005). As such, magnetic modelling of the French Guiana profiles was conducted with three aims: to analyse structural variations which may resolve the cause of minor misfits in the seismic and gravity models; to distinguish between oceanic and non-oceanic crust, i.e. the location of the OCT; and to correlate magnetic anomalies along-strike the margin to assist spreading rate calculations and the mapping of fracture zone trends. The modelling results (Fig. 8) proved to be highly non-unique, although a region of increased magnetization was observed. This anomaly is suggested to result from a fracture zone in the oceanic crust and, consequently, it is assumed that using its magnetic characteristics the areal extent of the major offset fracture zones may be mapped in the regional magnetic data. Fig. 8 shows the Guyaplac regional data and an interpretation of possible lineations within it. A distinct low runs approximately eastwest at ~ 8°N. Using this trend, three areas of positive anomaly can be identified between 8° and 10°N. These areas are displaced from one another by ~90–130 km in an east–west direction and, in conjunction with the 2D magnetic modelling, support the inference of fracture zones since they also correlate with offsets in the basement surface interpreted in the MCS data and lineations in the FAA. Two of the interpreted features correlate with the fracture zones which intercept Profile D at ~ 385 and ~ 440 km offset. The first of these corresponds to the location of the OCT in the WA model, which suggests that the thinned continental crust along Profile D terminates at a fracture zone and, despite the wide zone of thinned crust, the margin here should be termed a transform margin. The second of these zones intersects the profile at the site of the large topographical change in the basement surface identified in both the WA and MCS data. Two further fracture zones are interpreted to lie at ~ 530 and ~470 km offset. No fracture zones are interpreted to cross Profile A, although one is extrapolated, from the gravity anomaly further oceanward, to 563 intersect the margin at 5.5°N, close to the continental slope on Profile A. The location of fracture zones in relation to the structural variation observed, along Profile D in particular, highlights their importance in the evolution of the French Guiana margin. 7. Discussion The aim of the ACE was to study the along-margin-strike variation in crustal structure at the French Guiana-Northeast Brazil continental margin. Comparison of the results of modelling of data from ACE Profile A (Greenroyd et al., 2007a) with that of Profile D (Greenroyd et al., 2007b), coupled with interpretation of the Guyaplac MCS profiles, suggests significant along-strike variation exists within the ~ 240 km of the margin imaged by these transects. Both of the whole crustal WA profiles are interpreted to extend from pre-rift continental to post-rift oceanic crust and thus may also be used to inform understanding of the geometry and mode of opening of the equatorial Atlantic. 7.1. Margin evolution The crustal structure of the equatorial Atlantic margins reflect its tectonic evolution from the onset of continental thinning to the establishment of seafloor spreading which continues to the presentday. Oblique rifting of South America from West Africa resulted in trans-tensional extension which is manifest along-strike by structural segmentation, whose boundaries define adjacent margin regions which do not display distinctly rifted or transform characteristics but instead show elements of both. The margin as a whole evolved amagmatically, most likely as a result of the long rift duration and, perhaps, an anomalously cool underlying mantle. Following final break-up, oceanic crustal accretion resulted in unusually thin oceanic crust margin-wide, most likely a result of a combination of factors including a slow spreading rate and a high density of fracture zones. Post-rift, sediment deposition and accumulation occurred slowly until the mid-Miocene, when the uplift of the Bolivian Andes altered drainage patterns across South America and resulted in the catchment area for the Amazon River extending across most of the continent. The uplift was accompanied by erosion which resulted in increased rates of terrigenous sediment flux into the Atlantic where it was, and still is, transported northwestwards by deeper ocean water circulation. The bulk of the sediment influx is deposited as the Amazon Cone deep-sea fan system, with the remaining deposition being controlled by basement topography associated with fracture zones offshore French Guiana and by the Ceara Rise aseismic ridge to the east. Distinct structural segmentation is observed along the French Guiana margin (Fig. 9). The key structural features (continental, thinned continental, and oceanic crust and transition zone) from Profiles A and D, and B and F (Rodger et al., 2006) have been used to map out this segmentation. In addition, the OCT is traced along the margin in two ways. Firstly, to the south, the location of the 8 km depth-below-surface contour, i.e. the base of the sharp increase in depth to basement, is taken from the 3D basement grid of Rodger et al. (2006). Secondly, to the north of this basement grid, the zero crossing of the gravity FAA was traced along the margin. Both of these have been observed to be good indicators of the OCT location at the points at which they intersect the ACE profiles. In addition, the fracture zones Fig. 8. Interpretation of magnetic lineations within the regional magnetic anomaly. Top: The Guyaplac magnetic anomaly data (F. Klingelhöfer and W. Roest – pers. comm.} shows a clear east–west trending low at 8°N and a series of positive anomaly blocks between 8° and 10°N. Lineations (solid black), block divisions (dotted black) and the locations of Profile A and D (red) are shown. Bathymetric contours are shown to illustrate the current seafloor topography. Bottom: Results of magnetic modelling along Profile A (left) and D (right). In each case the model is divided laterally into several blocks, each assigned a remnant magnetization in A m− 1, with a corresponding inclination of 13° and declination of 6° calculated from the palaeomagnetic pole at the time of crustal accretion of 84°N 224°E (Gordon and Van der Voo, 1995) and the current survey location of 6.5°N 309°E. The calculated anomaly (red) is compared with the ACE shipboard (black dotted) and Guyaplac (black solid) data. The seaward extent of the corresponding WA model is shown by the vertical blue line. Bathymetric contours are plotted at 20 m, 50 m, 100 m (dashed line), 500 m (dotted line) and then at every 1000 m (solid line) intervals. Author's personal copy 564 C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 Fig. 9. Interpretation of structural segmentation along the French Guiana-Northeast Brazil margin. Bottom: Zones of rifted and thinned continental crust are shown by green shading, thin oceanic crust by orange shading and transform margins by purple shading. ACE Profile A, D (red), B, E and F (blue) are shown. Fracture zone traces are marked by black solid and dashed lines (cf. Fig. 1). Top: Reconstruction of the equatorial Atlantic conjugate margins. Interpreted margin orientations are rotated about a pole located at 60°N 35°W. The inset compares this pole (red dot) with those of Rabinowitz and LaBrecque (1979) showing total rotation poles as green circles and early poles of opening as green triangles. Pole labels are in Myr. The distance that the margin spread varies along strike. The southern part of the margin has spread ~ 370 km further that the northern (red; spreading path in blue) which, at a full spreading rate of 40 mm yr− 1, suggests that the southern margin rifted 9 Myr prior to the northern margin. Bathymetric contours are plotted at 100 m (dashed line), 500 m (dotted line) and then at every 1000 m (solid line) intervals. picked from the gravity and magnetic data were added and used to guide interpretation of rift and transform segments of the margin. Within Fig. 9, two rifted segments of the margin lie roughly northsouth. To the far south of these rifted segments is a transform margin which, given that this segment of the margin lies approximately parallel to local transform faults, is likely to be a ‘standard’ transform margin. The segment of margin modelled by Profile A lies in between the two rifts, and is interpreted as a “leaky” transform margin, although this term is used here to describe purely structural aspects and no implications are inferred as to associated magmatic activity. The strike of the northern and southern rifts differs by ~ 15°, a change which is most likely accommodated by the trans-tensional motion observed along Profile A, i.e. the plate rotation required to shift from the southern rift direction to the northern rift direction is the cause of a degree of oblique motion along the transform fault separating the two. This rotation may be part of a very long lasting Author's personal copy C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 shift in the plate orientations or may represent a ‘shimmy’ which occurred over a shorter time scale, possibly related to the release of tension when the South American and African plates finally parted, at or close to the Demerara Plateau (Gouyet et al., 1994). The northernmost transform segment on Fig. 9 is oriented similarly to the segment which is crossed by Profile A. However, the structure appears to be more similar to a ‘standard’ transform margin than the Profile A segment. For example, a marginal ridge is observed (Gadd and Scrutton, 1997), which is possibly a consequence of the greater incidence of first-order fracture zones intersecting this segment of the margin, or it may be a flexural feature associated with differential buoyancy forces following the development of the Ivory Coast transform margin. The obliquely rifted section of the margin described above is not easily explained by current models of margin evolution and, hence, a revised model is proposed, based largely on the model of Peirce et al. (1996); which was developed from Mascle and Blarez (1987) and Mascle et al. (1997). The revised model created to explain the oblique rifting observed in the equatorial Atlantic comprises five main stages: 1) Initial intra-continental rifting begins between South America and Africa; 2) Rifting continues and the lithosphere extends, thinning the continental crust. However, stresses are trans-tensional, i.e. rifting is oblique to the rift axis, and rather than forming strike-slip transform faults, crustal thinning occurs parallel with the rift axis. Consequently, segments of highly thinned continental crust are offset by regions of sharply thinned crust rather than abrupt transform faults; 3) Crustal thinning proceeds to break-up and oceanic spreading centres form; 4) As the two lithospheric plates drift apart, strike-slip motion occurs within the sharply thinned segments of crust. This results in the juxtaposition of old continental lithosphere against young oceanic lithosphere at a “leaky” transform margin; 5) The continental plates continue to drift apart, with the spreading accommodated by strike-slip motion along fracture zones stemming from the “leaky” transforms. The structure associated with rifting has now developed, showing typical rifted segments and atypical “leaky” transform segments. The margin then continues to evolve as increased sedimentation causes a progressively larger degree of subsidence. Post-rift, the continents of South America and Africa spread apart, with associated accretion of oceanic crust. This spreading may be described by movement of the continents relative to poles of rotation which, for the Atlantic, have been described by several authors (e.g. Rabinowitz and LaBrecque 1979) and are generally located within the North Atlantic between 40–70°N 15–45°W. Prior to ~107 Ma the poles of rotation are often placed further south, in West Africa (Rabinowitz and LaBrecque, 1979). Using the margin orientations shown in Fig. 9, the French Guiana and northeast Brazil margin may be projected eastwards toward the conjugate West African margin using such poles of rotation. Fig. 9 shows a reconstruction of the conjugate margin using a single pole of rotation positioned at 60°N 35°W and suggests that the conjugate margins can be located accurately using projection about a single pole. The reconstruction also suggests that the southern margin has spread ~ 370 km further than the northern margin within the ACE study area which, at a full-spreading rate of 40 mm yr− 1, corresponds with a 9 Ma difference in the time of final break-up between the two sections of the margin or asymmetric margin evolution. This gradual break-up may suggest that the trans-tensional features observed at this margin may be a result of the gradual northward motion of both the rift and the stress field which, given the progressive development of the margin, is oriented obliquely to the rift direction. 565 7.2. Effect of sediment thickness The assessment of the basement morphology and seismic velocity structure within the equatorial Atlantic has shown that the oceanic crust along Profile D is transected by at least two, possibly three transform features. The most distinct of these, located at 440 km offset, is associated with a 1.9 km change in basement topography along Profile D. Furthermore, the fault can be traced across several of the Guyaplac MCS data sections in which associated basement topography gradually decreases eastward, i.e. oceanward. In addition to the basement topography the fault is also clearly observed in the gravity FAA. This first-order transform fault is associated with a N100 km offset in the MAR and, hence, appears to be very significant in the post-rift evolution of the region. To the north of the fault, several further fracture zones are clearly identified, some of which can be traced over 1800 km from the MAR toward the Caribbean at ~ 54°W. To the south no fracture zones are observed in the FAA to the west of 46°W, although several hundred kilometres south of the Amazon Cone, they can once more be traced to the margin. Such an observation would be readily explained if the margin in this region was formed by purely rift type processes. However, as previously established, this is not the case and rifted and transform structures are observed. A possible explanation for the lack of fracture zones observed within the regional gravity data (Fig. 1) is the thick sediment cover which, despite their close proximity, differs significantly between Profiles A (6.5 km thickness) and D (3.9 km). The extra ~3 km of sediment observed along Profile A may result in higher densities within the lower sediment column, which would reduce the density contrasts and, hence, reduce the FAA characteristics associated with fracture zones. The reason for this change in sediment thickness along-margin-strike is partly related to the Amazon River, the major local source of suspended sediment (Cobbold et al., 2004). However, the fracture zones observed in this study may also play a role in controlling the distribution of sediment. It is likely that the large rise in basement topography, observed at a fracture zone along Profile D, has dammed the sediment which has been carried northward up the coast from the mouth of the Amazon River by prevailing currents (Johns et al., 1998). The Ceara Rise has had a similar effect, damming the eastern edge of the thick Amazon Cone. This has resulted in an unusually thick sediment column not just within the Amazon Cone fan, but also within the region south of 8°N and west of ~ 46°W, as shown in Fig. 9. 8. Conclusions In this paper two models of the deep crustal structure of the passive continental margin of French Guiana-Northeast Brazil have been interpreted in the context of margin evolution and along-strike margin segmentation. The models were created by modelling coincident MCS and WA seismic data, acquired as part of the Amazon Cone Experiment. As a whole, the margin is characterised by 35–37 km thick pre-rift continental crust which is separated from unusually thin oceanic crust (3–4 km thick) by thinned continental and/or transitional crustal regions. To the north, Profile D exhibits a 320 km wide zone of thinned continental crust adjacent to a narrow OCT, and is interpreted as a transform margin where the wide zone of thinned crust is a result of profile orientation highly oblique to the direction of rifting. Profile A, ~ 240 km to the south, shows a 70 km wide zone of thinned continental crust, wider than typical for transform and narrower than typical for rifted margins which generally thin over b40 km and N100 km respectively. The crustal structure along this profile is interpreted to reflect a “leaky” transform formed by trans-tensional extension. The contrasting crustal features between the models demonstrate that this margin is highly segmented structurally and Author's personal copy 566 C.J. Greenroyd et al. / Earth and Planetary Science Letters 272 (2008) 553–566 that the pattern of segmentation defines regions in which the margin displays characteristics of both rift and transform evolution. The location of fracture zones are demonstrated to strongly influence the variation in along-margin-strike structural characteristics, with the origin of these fracture zones being related to the initial break-up geometry. Acknowledgments We wish to thank the master, officers and crew of the RRS Discovery, together with the sea-going staff of NERC's UKORS, Tom Oliva of Seamap UK, and Anne Krabbenhöft and Cord Papenburg who operated the IFM-Geomar seabed instruments during the cruise. Land-based assistance was provided by Dr Jesus Berrocal, Prof. Cleverson Silva, and Prof. Alberto Figueiredo in Brazil, Andy Louch in NERC's RSU, Lourenildo Leite in Belem and Phillippe Weng and Pierre Laporte in French Guiana. This research was funded by the NERC through the Ocean Margins LINK programme and by an Advanced Research Fellowship awarded to RWH. The GMT and Seismic Unix software packages (Wessel and Smith, 1998; Cohen and Stockwell, 2000 respectively) were used to create the figures, and ProMAX to process the MCS data, for this paper. Finally, we also thank the two reviewers for their helpful and positive comments on this paper. References Bauer, K., Neben, S., Schreckenberger, B., Emmermann, R., Hinz, K., Fechner, N., Gohl, K., Schulze, A., Trumbull, R.B., Weber, K., 2000. Deep structure of the Namibia continental margin as derived from integrated geophysical studies. J. Geophys. Res. 105 (B11), 25829–25853. Benkhelil, J., Mascle, J., Tricart, P., 1995. The Guinea continental margin: an example of a structurally complex transform margin. Tectonophysics, 248, 117–137. Brandão, J.A.S.L., Feijó, F.J., 1994. Bacia do Foz do Amazonas. Boletim de Geociências da Petrobrás (Rio de Janeiro), 8 (1), 91–100. Bratt, S.R., Purdy, G.M., 1984. Structure and variability of oceanic lithosphere on the flanks of the East Pacific Rise between 11° and 13°. J. Geophys. Res. 89, 6111–6125. Christensen, N.I., Mooney, W.D., 1995. Seismic velocity structure and composition of the continental crust: a global view. J. Geophys. Res. 100 (B6), 9761–9788. Cobbold, P.R., Mourges, R., Boyd, K., 2004. Mechanism of thin-skinned detachment in the Amazon Fan: assessing the importance of fluid overpressure and hydrocarbon generation. Mar. Pet. Geol. 21, 1013–1025. Cohen, J., Stockwell, J., 2000. CWP/SU: Seismic Unix release 34: a free package for seismic research and processing. Centre for Wave Phenomenon, Colorado School of Mines. Contrucci, I., Matias, L., Moulin, M., Géli, L., Klingelhöfer, F., Nouzé, H., Aslanian, D., Olivet, J.-L., Réhault, J.-P., Sibuet, J.-C., 2004. Deep structure of the West African continental margin (Congo, Zaïre, Angola), between 5°S and 8°S, from reflection/ refraction seismics and gravity data. Geophys. J. Int., 158, 529–553. Damuth, J.E., Kumar, N., 1975. Amazon Cone: morphology, sediments, age and growth pattern. Geol. Soc. Am. Bull., 86, 863–878. Davis, M., Kusznir, N., 2002. Are buoyancy forces important during the formation of rifted margins? Geophys. J. Int., 149, 524–533. Dean, S.M., Minshull, T.A., Whitmarsh, R.B., Louden, K.E., 2000. Deep structure of the ocean–continent transition in the southern Iberia Abyssal Plain from seismic refraction profiles: the IAM-9 transect at 40° 20'N. J. Geophys. Res. 105 (B3), 5859–5885. Edwards, R.A., Whitmarsh, R.B., Scrutton, R.A., 1997. The crustal structure across the transform continental margin off Ghana, eastern equatorial Atlantic. J. Geophys. Res. 102 (B1), 747–772. Erbacher, J., Mosher, D.C., Malone, M.J., et al., 2004. Proc. ODP, Init. Repts., 207: College Station, TX (Ocean Drilling Program). doi:10.2973/odp.proc.ir.207.2004. Gadd, S.A., Scrutton, R.A., 1997. An integrated thermomechanical model for transform continental margin evolution. Geo Mar. Letts., 17, 21–30. Gordon, R.G., Van der Voo, R., 1995. Mean paleomagnetic poles for the major continents and the Pacific plate. Handbook Phys. Contants. Am. Geophys. Union 225–239. Gouyet, S., Unternehr, P., Mascle, A., 1994. The French Guyana margin and the Demerara Plateau: geological history and petroleum plays. In: Mascle, A. (Ed.), Hydrocarbon and petroleum geology of France, pp. 411–422. Greenroyd, C.J., Peirce, C., Rodger, M., Watts, A.B., Hobbs, R.W., 2007a. Crustal structure of the French Guiana margin, West Equatorial Atlantic. Geophys. J. Int. 169, 964–987. doi:10.1111/j.1365-246X.2007.03372.x. Greenroyd, C.J., Peirce, C., Rodger, M., Watts, A.B., Hobbs, R.W., 2007b. rustal structure and evolution of the Demerara Plateau, French Guiana. Geophys. J. Int. 172, 549–564. doi:10.1111/j.1365-246X.2007.03662.x. Hobbs, R.W., 2006. Quantifying uncertainty in margin structure derived from longoffset seismic data. Ocean Margins LINK Programme Science Meeting: From Margins to Mainstream.. Unpublished abstract. Hopper, J.R., Funck, T., Tucholke, B.E., Louden, K.E., Holbrook, W.S., Larsen, H.C., 2006. A deep seismic investigation of the Flemish Cap margin: implications for the origin of deep reflectivity and evidence for asymmetric break-up between Newfoundland and Iberia. Geophys. J. Int., 164, 501–515. Houtz, R.E., Ludwig, W.J., Milliman, J.D., Grow, J.A., 1977. Structure of the northern Brazil continental margin. Geol. Soc. Am. Bull., 88, 711–719. Johns, W.E., Lee, T.N., Beardsley, R.C., Candela, J., Limeburner, R., Castro, B., 1998. Annual cycle and variability of the North Brazil current. J. Phys. Oceanogr. 28 (1), 103–128. Korenaga, J., Holbrook, W.S., Kent, G.M., Kelemen, P.B., Detrick, R.S., Larsen, H.-C., Hopper, J.R., Dahl-Jensen, T., 2000. Crustal structure of the southeast Greenland margin from joint refraction and reflection seismic tomography. J. Geophys. Res., 105 (B9), 21591–21614. Kumar, N., Embley, R.W., 1977. Evolution and origin of Ceara Rise: an aseismic rise in the western equatorial Atlantic. Geol. Soc. Am. Bull., 88, 683–694. Le Pichon, X., Hayes, D., 1971. Marginal offsets, fracture zones and the early opening of the south Atlantic. J. Geophys. Res. 76, 6283–6293. Lin, J., Sibuet, J.-C., Srivastava, S., Manatschal, G., 2005. Origin of deep-sea magnetic lineations: magnetized oceanic crust or serpentinized exhumed mantle. EOS Trans. AGU, Fall Meet. Suppl., Abstract GP33B-06. Loncke, L., Droz, L., Gaullier, V., Basile, C., Roest, W., Patriat, M., 2006. Effect of structural inheritance on the distribution and importance of slope instabilities along a passive transform margin: the example of the French Guiana margin. Geophys. Res. Abstr. 8 EGU06-A-03803. Mascle, J., 1976. Atlantic-type continental margins: distinction of two basic structural types. An. Acad. Bras. Cienc., 48, 191–197. Mascle, J., Blarez, E., 1987. Evidence for transform margin evolution from the Ivory coast-Ghana continental margin. Nature, 326, 378–381. Mascle, J., Lohmann, P., Clift, P., 1997. Development of a passive transform margin: Côte d'Ivoire-Ghana transform margin — ODP Leg 159 preliminary results. Geo Mar. Letts., 17 (1), 4–11. Mello, M.R., Mosmann, R., Silva, S.R.P., Maciel, R.R., Miranda, F.P., 2001. Foz do Amazonas area: The last frontier for elephant hydrocarbon accumulations in the South Atlantic realm. In: Downey, M.W., Threet, J.C., Morgan, W.A. (Eds.), Petroleum provinces of the twenty-first century. AAPG Memoir, vol. 74, pp. 403–414. Morgan, J.V., Barton, P.J., White, R.S., 1989. The Hatton Bank continental margin, III. Structure from wide-angle OBS and multichannel seismic refraction profiles. Geophys. J. Int. 98, 367–384. Müller, R.D., Roest, W.R., Royer, J.-Y., Gahagan, L.M., Sclater, J.G., 1997. Digital isochrons of the world's ocean floor. J. Geophys. Res. 102, B2, 3211–3214. Mutter, J.C., 1993. Margins declassified. Nature, 364, 393–394. Nürnberg, D., Müller, R.D., 1991. The tectonic evolution of the South Atlantic from late Jurassic to present. Tectonophysics, 191, 27–53. Pearse, S., 2002. Inversion and modelling of seismic data to assess the evolution of the Rockall Trough. Ph.D. thesis, unpublished, University of Cambridge, pp211. Peirce, C., Whitmarsh, R.B., Scrutton, R.A., Pontoise, B., Sage, F., Mascle, J., 1996. Côte d'Ivoire-Ghana margin: seismic imaging of passive rifted crust adjacent to a transform continental margin. Geophys. J. Int., 125, 781–795. Pereira da Siva, S.R., 1989. Bacias da Foz do Amazonas e Para (Aguas Profundas): Une anlise sismoestratigrafica, tectonsedimentar e termica. Proc. Conf. Braz. Geol. Soc., 2, 843–852. Rabinowitz, P.D., LaBrecque, J.L., 1979. The Mesozoic South Atlantic Ocean and evolution of its continental margins. J. Geophys. Res. 84, 5973–6002. Rodger, M., Watts, A.B., Greenroyd, C.J., Peirce, C., Hobbs, R.W., 2006. Evidence for unusually thin oceanic crust and strong mantle beneath the Amazon Fan. Geology 34, 1081–1084. doi:10.1130/G22966A.22961. Sandwell, D.T., Smith, W.H.F., 1997. Marine gravity anomaly from Geosat and ERS 1 satellite altimetry. J. Geophys. Res. 102, B5, 10039–10054. Spudich, P., Orcutt, J., 1980. A new look at the seismic velocity structure of the oceanic crust. Rev. Geophys., 18, 627–645. Supko, P.R., Perch-Nielson, K., et al., 1977. Initial Reports of the Deep Sea Drilling Project, 39. U.S. Govt. Printing Office, Washington, p. 139. Tarantola, A., 2005. Inverse problem theory and methods for model parameter estimation. Society for Industrial and Applied Mathematics. Philadelphia, p. 342. Unternehr, P., Curie, D., Olivet, J.L., Goslin, J., Beuzart, P., 1988. South Atlantic fits and intraplate boundaries in Africa and South America. Tectonophysics, 155, 169–179. Watts, A.B., Peirce, C., 2004. Amazon Cone seismic experiment – ACE. RRS Discovery cruise report. Unpublished, pp56. Wessel, P., Smith, W.H.F., 1998. New improved version of the Generic Mapping Tools released. EOS Trans. AGU, 79, 579. White, R.S., McKenzie, D., O'Nions, R.K., 1992. Oceanic crustal thickness from seismic measurements and rare earth element inversions. J. Geophys. Res. 97, B13, 19683–19715. Whitmarsh, R.B., White, R.S., Horsefield, S.J., Sibuet, J.-C., Recq, M., Louvel, V., 1996. The ocean–continent boundary off the western continental margin of Iberia: crustal structure of Galicia Bank. J. Geophys. Res. 101, B12, 28291–28314. Wu, Y., Louden, K.E., Funck, T., Jackson, H.R., Dehler, S.A., 2006. Crustal structure of the central Nova Scotia margin off Eastern Canada. Geophys. J. Int., 166, 878–906.