Crustal thinning, mantle exhumation and serpentinization in the Porcupine Basin, offshore Ireland: evidence from wide-angle seismic data

Lithospheric extensional processes that lead to the development of both intra-plate sedimentary basins and passive continental margins near plate boundaries have been the focus of considerable scientific research since the global theory of plate tectonics became widely accepted in the late 1960s (Menard 1987). Over the last 25 years or so two end-member tectonic models for the formation of large-scale sedimentary basin systems and passive continental margins have been proposed. The pure shear model of McKenzie (1978) and the simple shear model of Wernicke (1985), with various modifications, have been applied to a multitude of active and dormant continental extensional systems around the world (Kusznir & Ziegler 1992).

Perhaps the most well-studied intra-plate sedimentary basin in the North Atlantic region is the complex graben system of the North Sea, where the concepts of pure and simple shear have been applied to formulate different models of basin formation, which are often based on the same geological and geophysical datasets. These studies have been generally inconclusive about the relative importance of these different shear regimes in controlling the crustal-scale geometry and basin architecture of the moderately extended North Sea basin system, where the crust has a maximum stretching factor (ß) of about two (Klemperer & White 1990; Reston 1990).

In this paper we present results from a wide-angle seismic experiment in the Porcupine Basin, off the SW coast of Ireland, which have resolved deep structure below very highly extended crust. The results shed some light on the problem of the origin of large extensional structures in the Earth's lithosphere within the North Atlantic region and have applications to similar passive margin basins elsewhere around the world.

The Porcupine Basin

The Porcupine Basin is a deep-water sedimentary basin, located offshore to the SW of Ireland in the North Atlantic (Fig. 1 ). Its present shape and major structure are generally thought to have formed during a major episode of lithospheric stretching in Jurassic times, with other, less pronounced rift episodes in the Permo-Triassic and Early Cretaceous (Shannon 1991; Tate 1993; Sinclair et al. 1994; Shannon et al 1995). Inherited tectonic fabrics from Variscan, Caledonian and older erogenic cycles had a strong influence on the pattern and scale of Mesozoic sedimentary basin development in this part of the North Atlantic. The reactivation of some of these older basement structures that are well expressed on free-air gravity gradient maps (as lineaments) influenced the regional sedimentary architecture west of Ireland and in the Irish and Celtic Seas (Readman et al. 1995; McGrane et al. 2001).

However, the control of this pre-existing basement structure is not so apparent in the Porcupine Basin, which has a north-south trend orthogonal to the main Caledonian and Variscan inherited fabrics (Naylor & Shannon 2005). There is evidence of Late Palaeozoic basin development in the region, which crosscuts the axis of the younger Mesozoic and Cenozoic basins (Naylor et al. 2002; Naylor & Shannon 2005). These basins are Carboniferous in age and trend westwards from onshore Ireland across the Irish shelf region towards the eastern margin of the Porcupine and Rockall basins (Croker 1995; Readman et al. 2003) where they generally lie beneath Mesozoic and Cenozoic to Recent sediments.

Geological and geophysical constraints on the geometry of the pre-Cenozoic succession in the Porcupine Basin are particularly poor away from the margins of the basin (Croker & Shannon 1987). The Cretaceous and Cenozoic successions are better understood, and consist of mostly unfaulted, post-rift strata (Shannon et al. 1993; McDonnell & Shannon 2001; Stoker et al. 2005). Previous studies of the large-scale structure of the Porcupine Basin, which attempted to understand the regional structure and evolution of the basin, were based largely on petroleum industry borehole and seismic reflection data (e.g. Tate 1993; Tate et al. 1993; Naylor et al. 2002; Reston et al. 2004; Naylor & Shannon 2005). Some of these (see Naylor et al. 2002, and references therein) identified a number of major, deep-seated structures within the centre of the basin and lying beneath the sedimentary succession or within the crust.

A deep seismic reflection profile in the Porcupine Basin (Croker & Klemperer 1989) failed to clearly image the seismic reflection Moho because of high seismic energy attenuation (low seismic Q values) within the Mesozoic to Cenozoic sediments. Some earlier seismic refraction (wide-angle reflection) experiments in the region investigated the deep crustal structure of the Porcupine High (Whitmarsh et al. 1974) and the basin margin structure along the Irish Mainland Platform (Makris et al. 1988), as well as the continent-ocean boundary at the mouth of the Porcupine Seabight Basin. Here we present the first well-resolved results of P-wave velocity variations in the deeper parts of the Porcupine Basin along a transverse wide-angle seismic profile that crosses the northern part of the basin (Fig. 1).

The wide-angle seismic data were gathered across a structural feature known as the Porcupine Arch, identified in recent interpretations of vertical incidence seismic reflection data (Naylor et al. 2002). This structure coincides with a large free-air gravity anomaly visible on satellite gravity data (Fig. 2). The origin of this gravity high is unknown. One interpretation is that it is due to dense and high-velocity magmatic material underplated at the base of stretched continental crust (Tate et al. 1993). Recent 2D gravity modelling and seismic stratigraphie analysis suggest that it may be due to very thin continental crust or to exhumed anomalous serpentinized mantle lithosphere (Reston et al. 2004; Readman et al. 2005). Leg 4 of the Rockall and Porcupine Irish Deep Seismic (RAPIDS4) wide-angle seismic experiment was undertaken to test these different ideas about the nature of the Porcupine Arch and the origin of the gravity anomaly (Fig. 2).

The RAPIDS4 seismic experiment

In the summer of 2002 a large quantity of high-quality wideangle seismic data was acquired in the northern part of the Porcupine Basin (Fig. 1) during the RAPIDS4 experiment, using the Russian research vessel Akademik Boris Petrov. The expertment was designed so that the resolution of crustal structure was optimized across the region straddling the Porcupine Arch and the associated free-air gravity high (Figs 1 and 2; see also Fig. 3). It took into consideration the likely propagation distances of the seismic (airgun) sources. The 230 km profile runs east to west (at 51.8

A total of 65 ocean bottom seismometers (OBSs) with four channels (one hydrophone and three orthogonal 4.5 Hz geophone components) were deployed at 3-4 km intervals along the profile. Of the deployed instruments 61 were recovered and 58 of these contained usable data. The seismic source was an array of nine tuned airguns with a total capacity of 3820 in3, which were towed 75 m behind the ship's global positioning system (GPS) antenna at 10m water depth. These were fired at about 120 m intervals (i.e. 1 minute intervals) across the profile (a total of about 2000 shots) and produced a source signal that is rich in higher acoustic frequencies of up to 40 Hz, particularly in the near-seismic wavefield where attenuation of the seismic signal is less. Seismic energy propagation distances were very good and often exceeded 100 km, although at large offsets the signal-to-noise ratio is lower, which is typical for this type of seismic data (Figs 4-9).

The precise geographical locations of the recovered OBSs were determined using direct water (P-wave) arrivals, as significant wind-driven and thermohaline currents are present within the water column of the Porcupine Basin (O'Reilly et al. 2003), which is up to 1060 m deep along the central part of the experimental profile (Fig. 1). These currents, if not corrected for, can lead to substantial errors in positional accuracy and hence the exact arrival time of seismic energy from the airgun sources at the OBS receiver positions. Immediately after recovery of the instruments a time drift correction was applied to the OBS clocks, as this can also be a source of significant error during data interpretation and modelling of the observed travel-times.

Modelling methods and strategy

The interpretation and picking of first arrivals on each of the recovered 58 OBS seismograms, which contained usable data, was carried out on the vertical geophone component using a variety of reduction velocities of between 4.0 and 8.0 km s^sup -1^. The construction of the initial model used a priori geophysical and geological knowledge about the regional P-wave velocity structure built up from both past marine seismic experiments in the Porcupine Basin region (Whitmarsh et al. 1974; Makris et al. 1988; O'Reilly et al. 1996; Morewood et al. 2005) and recent onshore wideangle seismic experiments in SW Ireland along the Irish Atlantic seaboard (Landes et al. 2003). These earlier results were used to define the deeper crustal structure at the eastern and western extremities of the model, towards the Celtic Sea Platform and the Porcupine High, respectively.

Travel-time inversion of the data based on the subjective picking of first arrivals using the approach of Zelt & Forsyth (1994) was unsatisfactory in the central region of the experiment profile, where apparent P-wave velocities exceeded 5km s^sup -1^, because of the complexity in the parts of the seismograms defining these higher velocities. Large lateral heterogeneity in structure over distances of less than 10 km occurs across the centre of the seismic profile because of local effects, caused by large variations in structure (compare Fig. 5a and b). A consistent classification of the different primary arrivals at the basin centre, particularly at small source-to-receiver offsets where the apparent velocities of the first arrivals were 5.5-6.5 km s^sup -1^ could not be objectively established because of this heterogeneity in structure. This led to the adoption of a purely forward modelling approach that ensured that the large uncertainties in picking and classifying various phases from one seismic record section to the next across the basin were minimized so that the lateral variation in structure could be assessed more objectively (see Figs 4-9).

The forward modelling approach used the raytracing software of Zelt & Smith (1992) that employs an 'across-and-down' modelling strategy (Zelt 1999) incorporating very well-established a priori knowledge regarding both the known late Phanerozoic geology of the region and the crustal structure of the Porcupine Basin and its environs, obtained from the previous seismic experiments. This forward modelling strategy involved superimposing calculated theoretical travel-times on the actual seismograms, until a good fit was obtained for the arrivals close to every OBS position across the RAPIDS4 profile. These arrivals define traveltime branches with apparent velocities of between 2 and 5 km s^sup -1^ and are derived from the predominantly Mesozoic and Cenozoic basin-fill sedimentary sequence that is present across the basin.

Once the structure for the basin-fill sediments had been defined well, the forward modelling approach was extended to greater source-toreceiver offsets where the interpretation was more subjective, especially in the region where short travel-time branches with apparent velocities typical of P^sub g^ (V^sub p^ c. 6km s^sup -1^) become apparent at the central part of the basin (see above). Subsequently, the very prolonged P^sub n^ travel-time branches from the deeper part of the structure, which are very clear on OBS positions across the basin centre and extend out to distances often exceeding 90 km (Figs 5-8), were included in the forward model.

Prominent secondary reflected phases such as the basement reflection (P^sub g^P) and the Moho reflection (P^sub m^P) were also included in the forward modelling procedure, where these could be identified. These secondary phases are generally harder to pick as accurately as the first (refracted) arrivals. They are not clearly present in all of the seismograms, especially those that were deployed across the Porcupine Arch, indicating that 'continental crust' is highly thinned or absent towards the axis of the Porcupine Basin at this latitude.

Calculated arrival times for first arrivals fit the observed arrival times within the estimated picking uncertainties across the basin centre. These uncertainties (assigned on the basis of signal-to-noise ratio on neighbouring seismic traces) are at worst ±50 ms for the near-field picks and are usually better than ± 100ms for most of the far-field picks out to distances in excess of 100 km (see Figs 4-9). Uncertainties in picking the secondary reflected arrivals are somewhat higher (i.e. ±1000 -150 ms) because of the generally complex pattern of P-wave coda (including multiples, reverberations and diffractions), which trail behind the primary P-wave arrivals on all of the seismic sections. An indication of how well the model is resolved by the ray coverage is indicated in Figure 3, where a small subsample of rays calculated using the model are shown. At the model's edges the crustal structure is based on previous wide-angle seismic experiments and the far-field P^sub n^ arrivals recorded on OBS record sections at distances of between 30 and 200 km across the model (Fig. 10). The best fitting travel-time curves for the various phase branches (i.e. P^sub g^, P^sub n^ and P^sub m^P), calculated from forward modelling, are indicated by the thin white lines in Figures 4-9. Calculated arrival times for each phase, defined by these curves, generally lie within the assigned uncertainties described above.

As a result of the large number of OBSs used in the experiment and the very close instrument spacing of 3-4 km, the uncertainty in both the derived velocity structure and the interface depths across the central part of the model are small and are determined by lateral and vertical changes in ray coverage. For the upper part of the structure (with velocities between 2.0 and 5.0 km s^sup -1^) the velocity uncertainty is estimated as between 0.05 and 0.1 km s^sup -1^ for the shallowest and deepest layers, respectively, and the depth uncertainty is 0.1-0.2 km. Where the subcrustal structure is well sampled by rays, velocities are determined to within ±0.1 km s^sup -1^. These estimates are based on perturbing the model interface and velocity nodes until the fit between the calculated and the observed travel-times became noticeably degraded.

The geological model

The model developed for the sedimentary and crustal structure is illustrated with reference to ten wide-angle seismic record sections from the central region of the basin and its eastern and western margins. During the forward modelling approach the emphasis was focused on the centre of the basin in the vicinity of the Porcupine Arch. Seismic data from this part of the RAPIDS4 experimental profile resolve the deep velocity structure below the basin centre, and the pattern of long-range P^sub n^ arrivals out to 100 km on both sides of the OBSs constrains deep crustal structure on the basin margins. Deep structure to midcrustal levels from the basin margins to the basin centre is also constrained by observations from the shelf and slope-break regions of the basin. The main features of the derived model are outlined below and are related to the seismic observations presented in the seismic record sections shown in Figures 4-9.

Sedimentary structure

A number of general features of the sedimentary structure are defined across the central part of the model. The sedimentary succession along the seismic profile consists of three fundamental layers, defined by distinct travel-time branches in the observed travel-time curves. The first layer is about 1.5 km in thickness at the deep-water basin centre and thins towards both margins of the basin to about 0.2 km; it is defined by the short distinct travel-time branch labelled S^sub 1^ in Figures 4-9. Its P-wave velocity increases from about 2.0 to 2.4 km s^sup -1^ with depth. The layer is interpreted as a Cenozoic (mostly Neogene) succession that can be correlated with seismic reflection data northwards in the basin, where the age is confirmed by well data (e.g. Naylor et al. 2002).

The second sedimentary layer varies from 3 to 7 km in thickness, with velocities of between 3.2 and 3.8 km s^sup -1^. This layer is defined by a distinctive travel-time branch (the branch labelled 82 in Figs 5-9) across the entire basin and disappears abruptly onto the basin margins (Fig. 4). Two connected sediment accumulations are present on either side of a broad basement high that correlates with the Porcupine Arch defined in interpretations of seismic reflection data (Naylor et al. 2002). A simple topography at the base of this layer suggests that it is not heavily faulted and it has the geometry of a post-rift thermal subsidence succession. The layer is interpreted as a Cretaceous to early Cenozoic (Palaeogene) sedimentary sequence. Again, this age is supported by correlation with drilled strata further north in the basin (Naylor et al. 2002).

Short but strong phase branches (with high apparent velocities of 5.5-6 km s^sup -1^) extend laterally from the main S^sub 2^ travel-time branch at c. 3 s in reduced time (Figs 5-9). These branches represent a series of thin layers of high-velocity material present throughout the sedimentary sequence defined by S^sub 2^. The deepest of these short branches, which occur at 4.5 s in reduced time, can be easily confused with the crustal basement arrivals (P^sub g^) that are very clear on the Porcupine Basin margins (Fig. 4). These thin high-velocity layers may represent basalt flows or sills within the early Cenozoic succession, as volcanism of this age is commonplace across the entire North Atlantic region (White et al. 1987; Ritchie et al. 1999; Lundin & Doré 2005). The more noisy and diffuse branches, which occur at about 4.5 s, may be related to small amounts of late-stage synrift magmatism, or to mafic debris flows, near the Jurassic-Cretaceous boundary (see Discussion).

The deepest layer thins rapidly from about 4 km to < 1 km onto the Porcupine Arch and its velocity varies from between 4.5 and 5.0 km s^sup -1^ towards the basin centre, where it thins rapidly or pinches out (Fig. 10). It is well defined as a first arrival only towards the eastern margin of the basin, between the Irish Mainland Platform and the Porcupine Arch (the travel-time branch labelled S^sub 3^ in Fig. 9). No first arrivals define the presence of this layer at the basin centre (Figs 5-8). This observation requires that its thickness is less than the resolution limit of the seismic data, which is about 1 km in this region of the model where the layer is deeply buried by the overlying layers, defined by the S^sub 1^ and S^sub 2^ travel-time branches (Fig. 9). The layer is interpreted as a heavily faulted and tectonically stretched synrift sequence of predominantly Jurassic sediments, deposited during the major episode of crustal extension that shaped the present configuration of the basin (Croker & Klemperer 1989; Sinclair et al. 1994).

Crustal and upper mantle structure

Complex changes in deep crustal structure occur from the basin flanks towards the basin centre. These generate abrupt changes in the pattern of first arrivals from east to west across the basin. Very short (<100 -15 km in length), but identifiable, P^sub g^ branches are sometimes present across part of the basin centre, produced by very high P-wave velocity gradients (Figs 5-8). However, on some nearby record sections these are missing, implying a large amount of lateral variation in basement structure (compare Fig. 6a and b or Fig. 8a and b). As was stated previously, these apparent P^sub g^ branches (where interpreted) can be very easily confused with the branches originating from the interpreted thin basaltic horizons embedded in the lowermost part of the sedimentary succession. Thus there is some ambiguity in defining whether continuous pre-rift crust is present everywhere across the basin. These observations and the absence of a prominent P^sub m^P reflected phase (Figs 5-8) suggest that the crust is severely stretched (to <2-3 km in thickness) or even completely attenuated near the basin centre. The central part of the basin (Fig. 10) contains an elevated region of material with P-wave velocities of 6.0-6.5 km s^sub -1^ that correlates with the Porcupine Arch (Naylor et al. 2002). This region of slightly thickened 'crust ' is inferred on the basis of the short Pg branches evident on the western set of seismic record sections presented (Figs 5-8, parts a).

Mantle velocities were determined using the forward modelling strategy described in the methods section. They are anomalously low beneath the highly stretched crust in the vicinity of the Porcupine Arch. They are typically in the range 7.2-7.5 km s^sup -1^ with large vertical and horizontal gradients, rising to more typical P^sub n^ values that can exceed 8.0 km s^sup -1^ (Fig. 10). This large variation in the velocity is based upon detailed travel-time modelling of the very prolonged P^sub n^ arrivals clearly present in the seismic data (Figs 4-8) out to distances that are often in excess of 100km. These higher velocities occur at deeper structural levels close to the basin margins. The highest P^sub n^ velocities ( V^sub p^ c. 8.2km s^sub -1^) occur within the upper mantle within the region of maximum depth energy penetration along the Porcupine High, where the topographic gradient on the Moho is also greatest (Fig. 10). Lowest P^sub n^ velocities occur at the shallowest depths within the interpreted subcrustal mantle lithosphere within the central region of the basin. These velocities are so low that the distinction between lower crust and upper mantle may be impossible when based purely on P-wave velocities (see Discussion).

The Porcupine Basin is clearly asymmetric in terms of both the basin-fill geometry and crustal structure. This asymmetry is very well constrained by far-field refracted and reflected energy from the Moho and upper mantle at source-to-receiver offsets of 45-90 km. In these record sections (Figs 5-8) the P^sub n^ arrival is consistently between 0.5 and 0.6 s earlier to the west of each of the OBS positions at source-to-receiver offsets exceeding 50 km. This clearly demonstrates that the Moho descends at a significantly steeper angle below the Porcupine High than beneath the Irish Mainland-Celtic Platform region (Fig. 10).

Discussion

The highly asymmetrical crustal thickness variation across the Porcupine Basin resembles the asymmetry found across certain conjugate continental margins such as the Labrador and Greenland margins, where late-stage lithospheric simple shear has been proposed (Louden & Chian 1999). This crustal asymmetry is more easily explained by differential extension of the crust and mantle lithosphere and a simple shear mechanism for extension across the region of the Porcupine Arch rather than the coaxial pure shear model proposed by Tate et al. (1993).

In the coaxial pure shear model (McKenzie 1978) for continental rift development the post-rift thermal subsidence basin directly overlies and is perfectly symmetrical with the underlying sequence of synrift sediments. The simple shear model, where a large lithospheric-scale low-angle detachment fault system is involved, predicts that the younger thermal subsidence basin is considerably offset from the older synrift basin (Wernicke 1985). Across the Porcupine Basin the crustal model derived from the RAPIDS4 seismic data (Fig. 10) shows that the interpreted (Cretaceous to Recent) post-rift sequence is generally concordant with the highly stretched crust and the underlying Jurassic synrift sedimentary sequence, which have been similarly interpreted in the nearby Rockall Basin (Hauser et al. 1995; O'Reilly et al. 1996; Morewood et al. 2005). This suggests that bulk pure shear prevailed in the deeper parts of the lower crust and mantle lithosphere during the extension process within both of these neighbouring sedimentary basins, whereas very large amounts of strain were focused into the crust during deformation governed by simple shear.

The crust is severely stretched or completely attenuated at the central regions of the Porcupine Basin, suggesting that, in certain areas, local mantle exhumation may have occurred during the latter stages of tectonic extension. The very rapid thinning of the lowermost sedimentary layer (interpreted as mostly a synrift Jurassic sequence) towards the basin centre appears to occur sympathetically with increased amounts of crustal thinning. This can also be explained if extensional deformation was gradually focused towards the basin centre in response to intra-crustal simple shear, as the synrift sediments were deposited.

The degree of crustal thinning is much more extreme than that defined by similar experimental seismic techniques in the adjacent deep-water Rockall Basin (Hauser et al. 1995; O'Reilly et al. 1996). This thinning, together with the low P-wave velocities and the high positive velocity gradients encountered, is interpreted as evidence for substantial serpentinization (PerezGussinye et al. 2001; Reston et al. 2004, and references therein) within the subcrustal mantle in both basins. Similar thermal and rheological processes to those proposed for the Rockall Trough (O'Reilly et al. 1996) and the Galician continental margin (Perez-Gussinye et al. 2001) may have occurred as deformation progressed. However, the amount of stretching is much more severe in the Porcupine Basin than in the Rockall Basin (Hauser et al. 1995). This is surprising in view of the narrowness of the Porcupine Basin at this latitude.

Late-stage simple shear on a system of Theologically controlled detachment surfaces, which descend westwards with shallow dips towards the Porcupine High from crustal to subcrustal levels, can explain the asymmetry in both crustal geometry and P-wave velocity structure (Fig. 10). The presence of weak zones of serpentine in the lithospheric mantle would facilitate the initiation of such low-angle detachment surfaces in the upper mantle during the later stages of extension, prior to mantle exhumation. Once the entire crust becomes brittle and Theologically coupled to the mantle lithosphere seawater circulation and peridotite serpentinization begins (O'Reilly et al. 1996). One-dimensional calculations suggest that these processes can start when the bulk-stretching factor in the crust reaches a value of three (Perez-Gussinye et al. 2001). These calculations show that 'embrittlement' of the entire crust, which is required for the onset of seawater circulation between the hydrosphere (the Jurassic Ocean at the time) and the mantle lithosphere, can occur well before mantle exhumation develops.

The P-wave velocity structure within the subcrust is difficult to reconcile with magmatic underplating. No double P^sub m^P reflection from an underplated body (from the top and base of the body, respectively) is evident and on every seismic record section across the basin centre evidence for a P^sub n^, P branch is either weak or absent (Figs 5-8). In addition, the very high vertical and horizontal velocity resolved in the deepest parts of the model (Fig. 3) are atypical of the P-wave velocity structure found along volcanic continental margins (Vogt et al. 1998) or magmatic rift systems such as the Northern Ethiopian Rift (Mackenzie et al. 2005). Finally, there is no evidence for the significant quantities of intruded or extruded magma in the vicinity of the Porcupine Arch that one would expect if the Jurassic mantle was anomalously hot and the coaxial pure shear model applied during the main Jurassic extensional episode.

Even if strain rate values were very high (i.e. a coaxial ß value of between 10 and 15) over the 40-50 Ma duration of the main rifting event (Tate et al. 1993) significant amounts of magmatic melt production would not occur. Mantle potential temperatures would need to have been very high to generate large quantities of extruded and intruded melt, even in the absence of lateral heat conduction effects (see, e.g. Pedersen & Ro 1992; Bown & White 1995). The inferred mafic layers (sills or flows), near the base of the sedimentary sequence (Figs 5-8) are very minor in volume and their composition is unknown. They could represent low-viscosity debris flows of highly serpentinized peridotites, which, as Reston et al. (2004) have suggested, may be present in this region of the Porcupine Basin. The interpretation of the low P-wave velocities in the upper mantle beneath the Porcupine Basin as a thick region of magmatic underplate is therefore untenable.

None the less, there is good evidence for substantial late-stage synrift magmatism far to the south of the Porcupine Structural Arch. The most prominent feature in this regard is the 'Porcupine Median Volcanic Ridge' (for details, see Naylor & Shannon 2005). This ridge has been interpreted, on the basis of clear seismic-stratigraphic relationships (Shannon et al. 1993; Tate 1993) as a volcanic edifice of Early Cretaceous age. Late-stage synrift magmatism of this age is not incompatible with cold serpentinized mantle, as this has been observed along the distal parts of the heavily serpentinized Galician continental margin, one of the most well-studied non-volcanic continental margins in the world (Perez-Gussinye et al. 2001).

Basin-wide variations in the stretching factor (ß) based on analysis of synrift and post-rift subsidence from well data predict ß values between two and three across the region of this study (Tate et al. 1993; Reston et al. 2004). Our results, based on crustal thickness from wide-angle seismic data, demonstrate that ß values are at least 6-10 across the RAPIDS4 profile. These values are not compatible with the standard pure shear model, employed in the earlier basin subsidence studies. This very large discrepancy can be explained if extra lithospheric buoyancy is provided by a subcrustal wedge of serpentinized (low-density) mantle peridotite that developed in response to a simple shear mode of extension, during the later stages of deformation. The amount of synrift subsidence would be reduced as soon as the serpentinization front began to propagate into the mantle so that ß values based on subsidence observations will significantly underestimate the true amount of crustal extensional strain.

An interpreted seismic reflection profile that is coincident with the RAPIDS4 profile (seismic reflection profile 103 of Reston et al. (2004) and Readman et al. (2005)) shows evidence of domal uplift of basement above an undulating strong seismic reflection, which Reston et al. (2004) interpreted as a westward-dipping detachment surface (the 'P detachment' in their fig. 3) above which synrift faults dip towards the basin centre. The seismic stratigraphie geometries and the pattern of tilting of the interpreted pre-rift basement as a result of fault block rotation are consistent with domal uplift of the basin centre associated with the progressive development of a serpentinization front.

If serpentinization of the mantle occurred, a reduction in the amount and rate of subsidence late in the Late Jurassic-Early Cretaceous synrift basin development is expected, as a result of volumetric expansion of the parent peridotites and the ensuing provision of chemical buoyancy. There are no independent data to test this prediction across the basin centre (within the region of the Porcupine Arch). However, the presence of Early Cretaceous deltas, alluvial clastic fans and related turbidite fans (Croker & Shannon 1987; Moore & Shannon 1995) along the palaeo-shelf and slope regions of both margins of the basin suggests that basin margin uplift occurred at about the expected time.

These shallower water sedimentary facies (proven from well data) overlie deep-water Late Jurassic submarine fan facies to the north of the Porcupine Arch at about 52.7°N (Robinson & Canhan 2001 ). Early Cretaceous strata are cut by a major angular unconformity, of approximately Valanginian age, particularly pronounced across fault block crests (Sinclair et al. 1994) in much of the northern margin region of the basin. Later significant shallowing, with the basinward progradation of shoreface and deltaic conditions, took place in the region in Aptian and Albian times (Croker & Shannon 1987). The overall shallowing of faciès, especially in the mid-Cretaceous, cannot be correlated with any noticeable lowering of eustatic sea levels. In contrast, a long-term rise in eustatic levels is recorded from the Aptian to the Albian (Haq et al. 1987), at the time when pronounced regression and sediment progradation was occurring in the Porcupine Basin.

The base Cretaceous sequence overlies the late synrift Jurassic sequence unconformably towards the basin margins, where shelf to inner slope sedimentological facies are present, but becomes conformable towards the basin centre, where basinal to outer slope sediment facies prevail that locally onlap pre-existing upstanding structural blocks (Moore & Shannon 1995). These sedimentological facies characteristics persisted until mid-Cretaceous times, when a deeper basinal facies assemblage developed in response to long-term post-rift subsidence, which persisted until the early Palaeogene.

Recent studies of the subsidence history close to the Porcupine Basin margins also record the anticipated change in subsidence behaviour from the Late Jurassic to Early Cretaceous (Jones et al. 2001). A detailed analysis using a suite of petroleum exploration wells between 52.5° and 53°N in the Porcupine Basin indicates that 200-700 m of transient uplift, with respect to theoretical subsidence curves, occurred in Late Jurassic to Early Cretaceous time. This could, at least in part, be related to the onset of serpentinization proposed in this study. However, Jones et al. (2001) ascribed it to hot mantle, rather than to cold serpentinized buoyant lithosphere.

In summary, the model of coaxial pure shear extension that involves the entire lithosphere as a mechanism for forming the Porcupine Basin is unlikely to be completely valid. A mixed shear tectonic model in which bulk distributed pure shear occurs predominantly in the mantle lithosphere and the lower crust throughout the deformation, and simple shear occurs in the upper to mid-crust at the later stages of deformation generating the observed very high bulk crustal extension factors (ß c. 15 to 8), is our preferred model for the formation of the Porcupine Basin at this latitude. The coincident industrial reflection profile (profile 103 of Reston et al. (2004), their fig. 3), where an interpreted undulating detachment surface (the 'P detachment') descends westwards from synrift basement to lower crustal levels, lends further support to this type of model.

Conclusions

The RAPIDS4 wide-angle seismic experiment has resolved the deep crustal and upper mantle structure of the Porcupine Basin between the Porcupine High and the Irish Mainland-Celtic Platform. The pronounced asymmetry of the crust across the basin is inconsistent with the uniform pure shear model that has been assumed in previous studies. This asymmetry suggests that a simple shear mode of extension was prevalent, especially during the latter stages of deformation in the Late Jurassic. Low seismic P-wave velocities in the subcrustal lithosphere, combined with high velocity gradients, are evidence for partial serpentinization of the upper mantle below the Porcupine Arch. They are not compatible with a magmatic origin of the arch, although there is evidence for late-stage synrift magmatism.

Our preferred model is one that involves simple shear within the entire crust during the late synrift phase of deformation, the propagation of low-angle detachment surfaces into serpentinized subcrustal mantle, and ultimately localized mantle exhumation. The presence of serpentinites would provide the conditions for the development of low-angle detachments, therefore increasing the asymmetry of crustal deformation.

These results show that extension of the crust in the Porcupine Basin was more severe than in the adjacent Rockall Basin, where the proposed mode of tectonic extension is similar but where smaller amounts and more local degrees of serpentinization are thought to have occurred.

This publication uses data and survey results acquired during a project undertaken on behalf of the Geological Survey of Ireland (GSI) and the Porcupine Studies Group (PSG) of the Irish Petroleum Infrastructure Programme Group 3. The PSG comprised: Agip Ireland BV, Chevron UK Ltd, EIf Petroleum Ireland BV, Enterprise Energy Ireland Ltd, Marathon International Hibernia Ltd, Phillips Petroleum Company United Kingdom Ltd, Statoil Exploration (Ireland) Ltd and the Petroleum Affairs Division of the Department of Communications, Marine and Natural Resources. We thank M. Gaye from GeoPro GmbH, Hamburg, for help with the initial stages of the data processing. This paper is Contribution GP178 of the Geophysics section of the Dublin Institute for Advanced Studies. We also wish to thank two anonymous reviewers and R. England for their very constructive comments and helpful suggestions.

© 2006 Geological Society Publishing House Provided by ProQuest LLC. All Rights Reserved.