Abstract:
An explanation for high-frequency cycles of sea level in non-glacial times has remained elusive, despite more than two centuries of research since Lavoisier's seminal observations were published in 1789. In the development of seismic stratigraphy in the 1970s, putatively global high-frequency changes in relative sea level (Vail third-order cycles) were attributed to an unknown eustatic mechanism, prompting a search for Mesozoic ice ages. Over the last decade, a regional mechanism of sea-level control has been developed from studies of the sedimentary record in high-quality oil-industry data. These geological studies have supported the geophysical prediction that significant regional control of sea level is exercised by mantle-induced vertical motions of the Earth's surface. These vertical motions can occur over time intervals from several tens of million years to less than a million years, with amplitudes of tens of metres or more even at the shorter intervals. The vertical motions are not confined to regions with major hotspots. There are two related controls of surface vertical motion: evolution of mantle-convection cells, and pulsing flow within each cell. The effects are evident in the sedimentary record of North Atlantic basins. Mantle convection provides an alternative, regional, mechanism to eustatic control for explaining medium-frequency to high-frequency sea-level cycles.
Carozzi (1965) has described how the succession of Cenozoic marine transgressions and regressions in the Paris Basin was first identified and discussed by Lavoisier (1789). The cause of this first-order control of the sedimentary record by changes in relative sea level has remained mysterious ever since. The waning of continental ice-sheets has long provided an obvious explanation for rapid rises in global sea level during ice ages, but a plausible mechanism for high-frequency changes in relative sea level in non-glacial times has remained elusive (Immenhauser 2005).
Much of the debate on this topic in recent decades has concerned research by Peter Vail's team at Exxon, work first published over 30 years ago. AAPG Memoir 26 Seismic Stratigraphy—Applications to Hydrocarbon Exploration (Payton 1977) introduced sequence stratigraphy against the background hypothesis of eustatic control of relative sea level; these concepts should now be clearly separated. Wilson (1998) emphasized an important distinction between, on the one hand, the generally accepted Exxon precepts on sequence stratigraphy and, on the other hand, the much-disputed claims of global correlation of those sequences. One of the key disputed papers in AAPG Memoir 26 is that by Vail et al. (1977), on global cycles of relative changes of sea level. These cycles were held by Vail and his colleagues to indicate widespread eustatic control. The cycles of change in sea level were classified by frequency: first-order cycles have durations of 200–300 Ma; second-order cycles 10–80 Ma; third-order cycles 1–10 Ma.
Agreement on the causes of the sea-level control of the third-order sea-level cycles has been particularly elusive. In concluding their discussion of the causes of these cycles, Vail et al. (1977, p. 94) wrote: ‘In summary, the cause for the first-order and some second-order cycles may be related to geotectonic mechanisms. Some of the second- and third-order cycles can be explained by glaciation. The empirically observed rapid falls of sea-level at the ends of the third-order cycles remain unexplained where evidence for glaciation is not known.' That uncertainty concerning controls of eustatic sea-level change, on time scales between hundreds of thousands and tens of millions of years in non-glacial times, has persisted into this century (Miller et al. 2005; see Fig. 1).
Timing and amplitude of relative sea-level change caused by mantle convection (bold outline), compared with timing and amplitude of eustatic sea-level change caused by various geological mechanisms (fine outlines). Figure by S. Jones of Birmingham University, after Miller et al. (2005). SF, sea floor; Cont, continental.
Can regional cycles of uplift and subsidence driven by mantle convection provide an explanation for some of the second- and third-order cycles of sea-level change? It is claimed here that convectively supported vertical motions are not only widespread (Moucha et al. 2008), but also more rapid and of greater amplitude (Rudge et al. 2008) than previously realized (Fig. 1). Two related causes of vertical motion at the Earth's surface are involved in this change of thinking: evolution of upper mantle convection cells (leading to second-order sea-level cycles), and pulsing flow within each cell (third-order cycles) (Fig. 2). Patterns of convective support of the Earth's surface inferred from both present-day observations and theoretical studies can be matched with observations from the geological record. As a result of this matching, a uniformitarian picture is beginning to emerge, of convective support resulting from the planform arrangement of convection cells and from pulsing flow within these cells. The resulting sea-level changes are regional and in places diachronous. In principle therefore, it ought to be possible to distinguish episodic regional control of sea-level change by mantle convection from eustatic sea-level change caused by other, perhaps periodic, mechanisms. This principle is put to the test later in this paper.
Sketch to illustrate how mantle convection controls relative sea level. The upper mantle beneath the lithosphere is everywhere stirred by convection cells. The Earth's surface is deflected upwards above hotter, upwelling limbs and downwards above cooler, downwelling limbs to form a tessellating pattern of swells and depressions. The cells vary in their thermal anomaly, and the amplitude of the response at the Earth's surface varies in consequence (A and B). These cells control second-order cycles of sea level. The Rayleigh number of the Earth's mantle is 10 to the power of 6–8, depending on whether convection occurs in one or two layers; we thus expect the convection flow to be time dependent. This is illustrated here by blobs of hotter mantle (grey patches) and cooler mantle, carried round the convection cells. These blobs cause a surface response that is superimposed on the swell pattern (C), here referred to as pulsing. This pulsing controls the higher-frequency third-order cycles of sea level. Figure by S. Jones.
Why does sea level change when there are no ice sheets?
The main ‘geotectonic mechanism' for the longer-duration cycles discussed by Vail et al. (1977) is variation in the rate of sea-floor spreading, as later assessed quantitatively by Parsons (1982). Rowley (2002) suggested that it has not been significant over the last 180 Ma, and Miller et al. (2005, p. 1294) suggested: ‘a modest decrease in the rate of ocean-crust production because the long-term eustatic fall of 70 to 100 meters since the early Eocene cannot be totally ascribed to the permanent growth of ice sheets'. Debate concerning this potential control of global sea-level continues (Seton et al. 2009); however significant this mechanism may prove to be for the long-duration cycles, it does not bear on the problem of the higher-frequency cycles.
This unsolved problem of the control of second- and third-order cycles has required caution in accepting the principle of eustatic control of sea level that was set out in AAPG Memoir 26 in 1977. Caution had become obvious by the time of publication of AAPG Memoir 39, Seismic Stratigraphy II (Berg & Woolverton 1985). In that volume much of the emphasis in discussing sea levels was on application of the theory of thermal evolution of rifted basins developed by McKenzie (1978), rather than on eustatic controls. Hubbard et al. (1985, p. 93) wrote: ‘We find little support for Vail et al.'s proposal (Schlee 1984) that depositional sequence boundaries are largely controlled by global eustatic changes in sea level.' Hubbard and his colleagues appealed instead to major plate rearrangement for the main unconformities they mapped. For the higher-frequency events, they invoked local interplay between sediment supply and various tectonic processes, such as rifting and thermal subsidence.
The later AAPG Memoir 46 (Tankard & Balkwill 1989) reflected the Hubbard et al.-type approach to the higher-frequency events. This was combined with appeal to control of regional sea level by tectonic activity linked to the broad framework of plate activity (e.g. Cloetingh et al. 1989; Petrie et al. 1989), an approach still being pursued vigorously by Holford et al. (2009). Consideration of thermal uplift in Memoir 46 (White 1989; Ziegler 1989) was a harbinger of developments at the turn of the century that form the core of the argument now presented here.
Work by Galloway (1989) may be taken as an example of a great volume of published research on the topic of regional versus global control of sea level. He spelled out the crucial interplay between sediment supply, basin subsidence and uplift, and eustatic change in sea level. Another key publication with similar themes was AAPG Memoir 58 (Weimer & Posamentier 1993). Conditions inherent to a basin clearly control patterns of sedimentation, so a measure of regional control has never been in dispute. The difficulty of quantifying that measure against global controls lies at the heart of the longstanding debate in the literature.
How much control on sea level was exercised by ice sheets in the Mesozoic and Cenozoic?
The problem with applying eustasy to analysis of present-day marginal basins is not simply the strength of the evidence for alternative, regional, controls. Much of the observational science on which the study by Vail et al. (1977) was based appears to come from the Mesozoic and Cenozoic (see Hallam (1981) and Miall & Miall (2002) for critical comment on the cryptic nature of the Exxon database that gives rise to that ‘appears'). It has generally been held that, out of the last 200 Ma, glaciation can be invoked as a control of global sea level only from the Oligocene to the present day: ‘The earliest Oligocene (∼34 Myr ago) is widely accepted as the interval associated with the onset of ‘‘icehouse'' conditions' (Tripati et al. 2005, p. 341). Tripati et al. reported evidence for glaciations in both hemispheres during a transitional period from greenhouse to icehouse, starting in the Eocene about 42 Ma ago, so eustatic control is the obvious first court of appeal in assessing causes of widespread changes in sea level since the Palaeocene, but not for the Mesozoic and earliest Cenozoic.
Were the Mesozoic and early Cenozoic really so bereft of glaciation that there is no appeal to waxing and waning ice-sheets as a significant control of sea level? There is currently dispute on this point. Miller et al. (2003, 2004, 2005) are firmly in the ice camp. They proposed control by ephemeral Antarctic ice sheets as an explanation for million-year-scale changes of c. 15–30 m in Late Cretaceous and Palaeogene global sea levels. There is also evidence of Early Cretaceous cooling from deep-sea sediments (Stoll & Schrag 1996) and some support from models of Mesozoic climates (Valdes & Sellwood 1992; Price et al. 1998).
In the warmer camp we find Jenkyns et al. (2004), who argued for high temperatures for the Late Cretaceous Arctic Ocean, implying (p. 891) ‘a total absence of polar ice at these high latitudes': any ice sheets of that age would have been in the southern hemisphere. Moriya et al. (2007) and Ando et al. (2009) also found against Cretaceous glaciation on the strength of stable isotope evidence from the Demerara Rise and Blake Nose, in the western North Atlantic.
Gale et al. (2002, 2008) and Kuhnt et al. (2009) identified periodic variations in global Cretaceous sea levels and suggested that these variations were astronomically controlled. In the Jurassic, orbital variations may have triggered the formation of temporary ice sheets in the uplands of southernmost Gondwanaland, leading to ‘rapid and small-scale (a few metres) sea-level change' (Sellwood & Valdes 2008, p. 10). Glacioeustatic or other climatic control of some low-magnitude third-order changes in sea level during the long warmth of the Mesozoic and early Cenozoic remains highly plausible. The meticulous high-resolution global stratigraphy being developed by Gale and others will be the key to distinguishing such climatically controlled, periodic, sea-level changes from the episodic changes here attributed to heterogeneities in mantle convection. These episodic changes are considered in the next section.
More certain help from below?
By the time of the publication of the Exxon hypothesis in 1977, it was well established that mantle convection was a first-order control of the elevation of the Earth's surface at plate boundaries. From the late 1970s it became apparent that mantle convection is also a first-order control of surface elevation in present-day plate interiors, as well as on plate margins (McKenzie 1977; Parsons & Daly 1983). That first-order control varies on geological time scales: convection in the Earth's mantle is strongly time dependent (Rudge et al. 2008).
One generally accepted control of relative sea level that operates locally not globally is that associated with mantle plumes or hotspots (Ernst & Buchan 2003). The regional influence of this type of mantle convection over tens of millions of years is well recognized: ‘While the deep origin of mantle plumes is controversial, their surface expression is dramatic and widely observed. Perhaps the two most ubiquitous manifestations of mantle plumes are voluminous volcanism and a broad regional topographic swell' (Parsons et al. 1994, p. 83). The sedimentary record may be used to quantify in time and space this transient ‘broad topographic' vertical movement caused by mantle convection (Nadin et al. 1995; Jones et al. 2001; Xu et al. 2004; Mackay et al. 2005; He et al. 2006; Saunders et al. 2007).
Such control of surface elevation by mantle convection appears to be pervasive on the present-day Earth: it is not confined to obvious hotspots ‘with voluminous volcanism'. We can happily echo Moucha et al. (2008): ‘There is no such thing as a stable continental platform.' Vertical movements caused by mantle convection may have a magnitude of a kilometre or more (Saunders et al. 2007). Here is a mechanism for the second-order Vail cycles, operating regionally, not globally, over tens of millions of years. The application of this uniformitarian argument to the interpretation of the Mesozoic and Cenozoic stratigraphical record of the NE Atlantic Ocean (Fig. 3) is discussed in the following two sections of this paper.
Four North Atlantic swells, sketched on the strength of evidence from the sources cited in Table 1. The inset diagram in (c) shows the range of estimates of maximum Eocene uplift, based on data from locations marked by encircled numbers in the main diagram. Location 7 is Porcupine Basin. Figure by author and S. Jones.
But what of the third-order Vail cycles? The sedimentary record can also be used to measure the episodic pulse of mantle convection over a few millions of years (White & Lovell 1997). Shaw Champion et al. (2008) have quantified the effects of one such pulse on the Palaeogene record offshore Scotland. This has allowed Rudge et al. (2008) to develop a model of that pulsed mantle convection. The application of this model as a controlling mechanism for some third-order Vail cycles is central to the hypothesis of sea-level control advanced here.
Cryptic features of mantle behaviour are revealed to geophysicists by analysis of the sedimentary record, and geologists thereby discover at least a partial answer to longstanding problems with the mechanism of changes in regional sea level. Certain implacable facts provide a bulwark against an obvious line of circular reasoning. From a geophysical angle, the Rayleigh number of the Earth's mantle indicates clearly that the pattern of convection in the mantle will be time dependent (see Fig. 2 and Rudge et al. 2008). From a geological viewpoint, North Sea explorers can be as sure of the alternation of marine and non-marine Palaeogene sediments proved by their wells as was Lavoisier in the Paris Basin when, over 200 years ago, he was confronted with rocks of comparable age and type. The geological evidence for the advances in understanding claimed in this paper has been collected during a half-century of exploration for oil and gas in the North Atlantic; that evidence is considered in the following two sections of this paper.
Key linkages in the North Atlantic
The first set of publicly available palaeogeographical maps featuring the North Sea was published in the proceedings of the historic 1974 Bloomsbury conference, edited by Woodland (1975). One of these maps (Ziegler 1975, fig. 16) depicted Palaeogene palaeogeography. For the first time a confident outline of the Palaeogene shores could be drawn, following emergence of land from the Chalk sea. There is something familiar about this outline: for the first time in the geological history of Britain and Ireland there is recognizable anticipation of present-day coastlines. Early Scotland is prominent, supplying sediment into the North Sea Basin, including the sand that formed the recently discovered reservoirs at Forties and Frigg. The map implied that sand was also being supplied to the Palaeogene sea floor west of Scotland, a notion later to be tested successfully. Geologists long familiar with the obvious onshore evidence of removal of the Chalk and older cover from Britain during the Palaeogene could now balance the sedimentary books by looking offshore. Attempts to quantify that regional uplift have followed, using techniques such as fission-track analysis and vitrinite reflectance (Green et al. 2002).
The concept of broad regional thermal uplift lasting some tens of millions of years was beginning to be accepted by the 1970s, so an explanation of the uplift of Palaeogene Scotland was to hand. A connection could be made between the widespread volcanic activity shown on Ziegler's map and regional uplift associated with the inception of the North Atlantic igneous province, a plausible mechanism for a Vail second-order cycle. Two Cenozoic swells, Palaeogene and Neogene (Table 1; Fig. 3), are controlled by the mantle convection that is so dramatically in evidence at the present-day Iceland mantle plume (White 1989). The effects of this have been considered in detail (see, e.g. Doré et al. 2002).
Summary data for four Cenozoic–Mesozoic swells in the North Atlantic region
Despite the increasing wealth of data from the oil industry and related research since the 1970s, a mechanism for the third-order cycles subsequently mapped on the flanks of Palaeogene Scotland (Table 1) remained elusive until recently. Discussing the sea-level cycles identified by Stewart (1987) from the Palaeogene of the North Sea, Milton et al. (1990, p. 345) wrote: ‘The mechanism causing the ten short-scale relative sea-level cycles is unknown; possibly eustatic sea-level variations, intra-plate stress or short-scale tectonic variations.' Knox (1996, p. 209) recognized three main uplift phases in NW Europe, linked to the proto-Icelandic mantle plume, ‘expressed in the North Sea as long-term regressive–transgressive facies cycles'. White & Lovell (1997) proposed that the depositional episodes of Stewart (1987) provide a measure of the pulse of a mantle plume. The associated magmatic underplating provides a mechanism for both regressions of the sea (on intrusion of melt) and transgressions (on solidification of melt) (Maclennan & Lovell 2002).
This link with magmatic underplating may be considered plausible in explaining some, but not all, of the third-order cycles in the Palaeogene igneous province. Underplating is far less convincing as a mechanism for any of the Jurassic and Cretaceous vertical movements. These Mesozoic events have little associated igneous activity compared with the Palaeogene (Table 1). Therefore the appeal to magmatic underplating may be regarded as valid in particular but not general circumstances. Is there a more general control exercised by pulsing mantle convection, which does not necessarily involve significant melting and production of magma? We have seen that the search for such a mechanism has long been unsuccessful. That situation has now changed, with a fusion of geology and geophysics that provides the stratigraphical insights claimed in this paper.
Rudge et al. (2008) and Shaw Champion et al. (2008) quantified a general mantle-related mechanism for third-order cycles, on the strength of analysis of high-quality oil-industry data. The first step in this analysis was the recognition of large and rapid vertical movements on the flanks of Palaeogene Scotland (Underhill 2001; Smallwood & Gill 2002; Biskopsto 2004); transient uplift of such magnitude that glacioeustatic control can be firmly ruled out. Shaw Champion et al. (2008) integrated their detailed interpretation of 3D seismic with biostratigraphy to quantify further these vertical movements and to correlate them over some 400 km from west to east over Palaeogene Scotland (Figs 4 and 5). Uplift in the west peaked earlier (between 56.1 and 55.0 Ma) and was of greater magnitude (at least 490 m). Peak uplift to the east was later (between 54.7 and 54.5 Ma) and of lesser magnitude (minimum c. 300 m). A demonstrably diachronous erosion surface is the result.
Simplified palaeogeographical reconstruction of the North Atlantic Ocean during Early Eocene times (55 Ma). Crossed circles, possible locations of Icelandic plume centre at this time (WM89, White & McKenzie 1989; LM94, Lawver & Müller 1994; JW03, Jones & White 2003); hatched areas marked J and B, Judd and Bressay areas where observations of vertical motions have been documented; light grey areas, Late Palaeocene–Early Eocene volcanic rocks; plain white areas, putative landmass; white stippled regions, deltaic tops where ticked line indicates topset–foreset break; dark grey areas, zones of marine deposition; grey stippled areas, sand-dominated fans sourced from Scotland–Shetland landmass. FSB is Faroe–Shetland basin. Inset is idealized reconstruction of North Atlantic Ocean showing extent of influence of Early Eocene Icelandic plume after White & McKenzie (1989) and location of main figure. From Rudge et al. (2008, fig. 1). Reprinted with permission of Elsevier.
Stratigraphical columns of Judd and Bressay areas located in Faroe–Shetland and northern North Sea basins, respectively. (See Shaw Champion et al. (2008) and Underhill (2001), respectively, for further details.) Lithostratigraphical and biostratigraphical correlations follow Mudge & Jones (2004). Dating of biostratigraphy follows Luterbacher et al. (2004). BS, Bressay Sandstone. First and last appearances of key biostratigraphic indicators are shown: W.a., Wetzeliella astra; A.a., Apectodinium augustum; A.h., Apectodinium homomorphum; A.g., Areoligera gippingensis; A.m., Alisocysta margarita. From Rudge et al. (2008, fig. 2). Reprinted with permission of Elsevier.
From these numbers, a mantle-convection model of transient diachronous uplift can be developed. In this model, vertical movement is caused by a temperature anomaly, a hot blob, in asthenospheric material flowing away from a central conduit: the early Iceland mantle plume (Rudge et al. 2008; Fig. 6). The position of the centre of the plume implied by the Rudge et al. model is compatible with earlier work (Fig. 4). We may visualize the stately progress of the hot blob moving from west to east, travelling deep beneath Palaeogene Scotland at a velocity of a few hundred kilometres per million years, while the Earth's surface above it ripples upwards and downwards over vertical distances of hundreds of metres.
Schematic illustration of the geometry of the simplified plume model of Rudge et al. (2008, fig. 5). (See also Fig. 2.) Reprinted with permission of Elsevier.
The evidence presented by Shaw Champion et al. (2008) and the model of Rudge et al. (2008) provide the connection between third-order pulses and the mantle convection responsible for the main regional (Vail second-order) vertical movements over tens of millions of years. Episodic pulses in the convecting mantle will not necessarily cause magmatic underplating, but they can cause high-frequency uplift and subsidence at the Earth's surface, on the scale of the third-order cycles. The original underplating model of White & Lovell (1997) and Maclennan & Lovell (2002) now appears to be a special case of the more general Rudge et al. (2008) model. Evidence of igneous activity is no longer a prerequisite for invoking control by mantle convection of episodic high-frequency changes in regional sea level.
Therefore it may now be suggested with some confidence that the behaviour of mantle convection provides a plausible explanation for at least some second- and third-order Vail cycles. Examination of the stratigraphical record in that light should prove fruitful. One key to this will be recognition of episodes of transient diachronous uplift across a region, a process that will require data of the high quality commonly available only in well-explored hydrocarbon provinces, and the use of high-resolution stratigraphy as practised by Gale et al. (2002, 2008) and Huang et al. (2010) among others. Such examination of the stratigraphical record may also contribute to a better understanding of the nature of time-dependent flow in different varieties of mantle convection.
Let us start this examination in the same North Atlantic region. We have available abundant data, although not yet the integration of 3D seismic data with biostratigraphy that is essential to the results reported by Rudge et al. (2008) and Shaw Champion et al. (2008). Can we apply the model of pervasive pulsed mantle convection developed for the Palaeogene to two earlier (Mesozoic) swells in the same area?
Pulsing Mesozoic swells?
In another of his slides at the Bloomsbury North Sea conference Ziegler (1975, fig. 14) showed an ‘Igneous Centre' in the Central North Sea, and Jurassic rocks from this centre were described in a separate talk by Howitt et al. (1975). In the discussion following Howitt et al.'s presentation, Sellwood (1975, p. 387) noted that thermal uplift associated with this centre was probably a major control on regional sedimentation. This assertion was borne out by a large volume of subsequent work.
Ziegler & Van Hoorn (1989) discussed in more detail the ‘thermal dome' associated with this Jurassic igneous centre. Underhill & Partington (1993a,b) provided a detailed analysis of the growth and decay of this swell that shows its dominant influence on the regional geological record, extending over a period of some 30 Ma. In their concluding remarks, Underhill & Partington (1993a, p. 478) commented on the implications of their work for the global hypothesis of Vail et al. (1977): ‘care is needed when comparing similar trends in relative sea level from correlative exposures with charts purporting to show global eustacy, without considering regional effects'.
Is this caution by Underhill and Partington justified? How can we establish whether the episodic changes in relative sea level mapped by them on the flanks of the North Sea Jurassic swell are eustatic or a regional effect of pulses in mantle convection? Comparison with a reliable global Jurassic sea-level curve would be an obvious means of identifying the magnitude and timing of those pulses. What is much less obvious is how such a global curve can be established, given the probability of pervasive control of surface elevation by mantle convection through geological time.
One test is to compare the Jurassic relative sea levels mapped by Underhill & Partington on the flanks of the North Sea swell with those identified elsewhere in the same c. 20 Ma of Jurassic time (Table 2). Here I follow Galloway (1989) and Sharland et al. (2001) in concentrating on maximum flooding surfaces. Thirteen of these transgressions have been mapped by Underhill & Partington in the North Sea, following detailed biostratigraphical work by Partington et al. (1993). This North Sea work employed classical techniques of seismic stratigraphy, yet was free of assumptions concerning eustatic control of relative sea level. Indeed, as we have seen, Underhill & Partington questioned the application of the eustatic model in their interpretation of their data. In contrast, the eight main Jurassic transgressions recognized in Russia by Sahagian et al. (1996) and the seven recorded in Arabia by Sharland et al. (2001) were held by those workers to be most probably eustatic in origin.
Comparison of Bajocian to Kimmeridgian maximum flooding surfaces in North Sea, Russian Platform and Arabian plate
How sure is the correlation of maximum flooding surfaces within each of these regions? The uncertainties were clearly set out by Sharland et al. (2001, pp. 51–52) in their study of Arabia. As for the North Sea, Wilson (1998) showed the difficulty of reaching agreement on sequence boundaries even in the intensively studied Kimmeridgian of northwestern Europe. We may take the biostratigraphical resolution achieved on the Russian Platform as typical of the three regions: this varies from stage level (several million years) to substage level (few million years) (Sahagian & Jones 1993, pp. 1110–1111). With all this in mind, it might still be claimed that correlation of maximum flooding surfaces between the three regions, the North Sea, Russia and Arabia, is good at two, possibly three, zones (E. coronatum and A. eudoxus, also possibly P. baylei). Correlation of transgressions is not obvious apart from those times.
Yet more caution is in order. We have seen in the previous section of this paper how the model of control of high-frequency changes in relative sea level by pulses in mantle convection itself involves diachronous regional uplift over c. 1 Ma episodes: simple time-correlation of the effect of these pulses is not to be expected. Even after taking this new complicating factor into account, the data in Table 2 still suggest to me that we do have a hypothesis well worth testing; that there is regional rather than global control of at least several of the North Sea maximum flooding surfaces mapped by Underhill & Partington.
The hypothesis presented here is that both the waxing and waning of the North Sea Mid-Jurassic swell were punctuated by episodic higher-frequency variations in regional relative sea level. These (Vail third-order) episodes were caused by the horizontal advection of hot blobs in the convecting asthenosphere, moving away radially from the centre of the hotspot (Figs 2 and 6). The waxing swell would experience, on its rise out of the Jurassic ocean, episodes of more rapid uplift followed by rapid subsidence; evidence for this is unlikely to be well preserved on the flanks of the emerging landmass. Then the later subsidence of the waning swell, back below the waves, was interrupted by episodes of rapid uplift followed by rapid subsidence. These high-frequency episodes took place over a million years or so, followed by a return to the trend of overall subsidence of the Jurassic island over some 20 Ma. The evidence for this is preserved in places on the flanks of the subsiding swell (Underhill & Partington 1993b).
The notion of a pulsing Mid-Jurassic hotspot in the North Sea remains hypothetical, in contrast to the more firmly established case for control by mantle convection of the sedimentary record on the flanks of Palaeogene Scotland discussed in the previous section. Work to test the Jurassic hypothesis is in hand; it will be significantly aided by developments in astrochronology (e.g. Huang et al. 2010). Meanwhile I endorse the view of Underhill & Partington that caution is required in ascribing a eustatic origin to the changes in relative sea level recorded on the flanks of the Jurassic swell. I go further and suggest that during at least half of the Mesozoic and early Cenozoic, mantle-convection effects of this type exercised significant control of regional sea level in the North Atlantic area.
The Mid-Jurassic swell is the earliest of the postulated four mapped in the region (Table 1, Fig. 3); three of these have been considered so far. The fourth is the Early Cretaceous swell, the second of four in sequence of time. Ziegler (1989, p. 117) identified this swell as covering a wide area: ‘During the earliest Cretaceous ‘‘late Kimerian'' rifting pulse, the North Atlantic area became thermally domed up'. This Early Cretaceous uplift has been further considered by McMahon & Underhill (1995), McMahon & Turner (1998) and Jones et al. (2001). Its regional extent and duration may be quantified, with some uncertainties (e.g. Holford et al. 2005). Following Jones et al. (2001), the area of peak Berriasian uplift mapped by McMahon & Turner (1998, fig. 14) may be extended further west to include Porcupine Basin (Fig. 3). Pending further study, I do not at this stage follow Ziegler and include the ‘earliest Cretaceous' (Early Berriasian) peak uplift of the basins offshore eastern Canada, which precedes the Late Berriasian peak uplift of the Wessex and Porcupine Basins by as much as 2–4 Ma.
In the Wessex and Porcupine Basins surface evidence of Early Cretaceous igneous activity is negligible. Given its pervasive nature, mantle convection may still the key mechanism involved in development of higher-frequency changes in relative sea level on the flanks of the postulated Early Cretaceous amagmatic swell that affected those basins. This notion remains to be tested to the standard applied to the Palaeogene swell discussed in the previous section.
Is the North Atlantic special? I argue that it is not. Existing evidence for the influence of mantle convection on the sedimentary record beyond the North Atlantic may be drawn from studies in Australia (Al-Harthy 1998, 2001), Africa (Cox 1993; Gurnis et al. 2000; Walford & White 2005), China (Xu et al. 2004; He et al. 2006; but see also Peate & Bryan 2008), India (Halkett 2002; Maclennan et al. 2003) and elsewhere (Saunders et al. 2007). Uniformitarian support for the specific notion of pulsing mantle plumes is also demonstrated across the oceans. Examples may be examined in both the Pacific Ocean, as at Hawaii (Stolper et al. 1996) and Foundation Chain (O'Connor et al. 2002), and also in the Atlantic Ocean, as at Iceland (Vogt 1971), the Canary Islands (Paris et al. 2005) and the South Atlantic (Adam et al. 2007).
The four sets of vertical movement in the North Atlantic discussed here all have in common the ‘broad regional topographic swell' of Parsons et al. (1994, p. 83), but only in the case of the Palaeogene uplift is there truly the ‘voluminous volcanism' of Parsons et al., accompanying the opening of the NE Atlantic Ocean. We may imagine that convective support was as widespread on Mesozoic Earth as it is now, not just associated with major hotspots, and not necessarily associated with large igneous provinces.
Application to prediction of the subsurface
This is a final, anecdotal section, in which I seek to illustrate practical applications of the science discussed above. Those averse to such material may wish to skip to the Conclusion.
In the mid-1980s the BP Exploration office in Dublin was focusing on two areas offshore Ireland. One was the established gas province in the Cretaceous of the North Celtic Sea Basin; the other was Porcupine Basin, then as now a deep-water frontier area. As leader of that exploration team, I am now looking back somewhat wistfully at what more might have been achieved if the hypothesis set out in this paper had been available as a framework for considering possible targets. For the conventional exploration in the Celtic Sea, it would have been useful to know that the well-studied Cretaceous to the east, in the onshore Wessex Basin in southern England (McMahon & Underhill 1995, fig. 1) could be used with some confidence to predict sequences south of Ireland (McMahon & Turner 1998, fig. 4). This would be based on the notion that both sequences developed on the flanks of the same uplift and reflect the same pulses in mantle convection.
However, the most dramatic effect would have been in the approach to choosing drilling locations in Porcupine Basin. The exploration manager could have put himself in the position of a time-traveller sitting on Porcupine Bank. There he would have experienced two (Vail second-order) episodes of vertical motion, one uplift peaking in the Early Cretaceous and one in the Palaeogene (Fig. 3b and c, location 7). His observation of these two main periods of sand deposition would have been a handy guide to locating the wildcat drilled at that time: 43/13-1. In 1988 this was the deepest-water well ever operated by BP, with a cost that fully reflected the newness of the experience as well as the 921 m depth of turbulent ocean between drillship Pacnorse 1 and the Atlantic sea floor. Presence of enough good-quality reservoir had been identified as the greatest risk. The well was spudded on 7 April 1988. By the time it was abandoned on 11 September 1988, with oil shows in thin sandstones, we had the scant consolation of knowing that the likely reason for failure had been correctly assessed in seeking funding from the Board.
Some younger members of that exploration team based in Dublin later moved to Scotland and played a crucial role in establishing the successful deep-water Palaeogene play to the west of Shetland. It would have been good to understand at the time why the well-mapped Palaeogene sequence in the North Sea Basin might be taken as a guide to the (Vail third-order) pattern of transgression and regression to the west.
The cost of gaining access in the world's remaining frontier basins mounts, rivalling even the cost of building production facilities, so the importance of understanding first-order geological controls of sedimentation grows ever more significant in the industry. It would not be too late to apply the notions set out in this paper in Porcupine Basin and the surrounding region. An examination of possible applications elsewhere should also be worth while.
Twenty years ago, at the time of the frontier exploration in Porcupine Basin discussed above, few of us in the oil industry were much concerned with the effects the use of our products by our customers might be having on the Earth's climate. In recent years attitudes have changed, as I have described elsewhere (Lovell 2008a,b,c, 2009), and shall discuss in a second Presidential Address. From now on, at least as much attention should be given to putting carbon safely back underground as has been given in the past to taking it out. If carbon capture and storage is to play a significant role in helping our transition to a low-carbon economy, the use of depleted oil and gas reservoirs for storage will not be enough. Prediction of the suitability of saline formations for the safe storage of carbon dioxide will be crucial. We shall begin a new phase of exploration, seeking reservoirs to be used for the storage of carbon rather than its extraction. This new exploration may be guided by a more complete understanding of the controls of relative sea level than was available to us in looking for the oil and gas in the first place.
Conclusion
Where does the evidence considered in this paper leave the hypothesis of global cycles of sea-level change? The outcome of the search for a mechanism to explain the second- and third-order cycles at times ‘when evidence for glaciation is not known' (Vail et al. 1977, p. 94) brings into question the validity of habitually invoking eustatic controls, especially for high-frequency changes in sea level during the great stretches of geological time when ice sheets were either absent or vestigial. But invocation of regional control of relative sea level by mantle convection does not bring into question the classical techniques of seismic stratigraphy long associated with eustatic control: the amicable public separation of that Exxon couple urged by Wilson (1998) is long overdue.
The key conclusion of this paper is that consideration of mantle convection is essential to a full understanding of the evolution of sedimentary basins: it is a significant part of the persistently elusive answer to the Lavoisier question. This appeal to mantle convection might be read as an echo of the ‘geotectonic mechanisms' of Vail and his colleagues. But there is a crucial difference: the effects of mantle convection on higher-frequency changes in relative sea level are felt regionally not globally.
Acknowledgments
Over half a century a geologist gets a lot of help in research from colleagues in this friendliest of professions. This help comes from many sources; from those in industry, universities, surveys, museums and elsewhere. Since the mid-1990s I have been particularly fortunate, working at Bullard Laboratories, under the leadership of D. McKenzie and J. Jackson, in the research team led by N. White and in collaboration with former colleagues in BP and elsewhere. This paper arises from that joint work: it has benefited greatly from extended discussions with S. Jones and from his skill in creating figures to illustrate several of the key points arising from those exchanges. Comments from T. Elliott, J. Turner and an anonymous reviewer led to significant revision of the original manuscript and are much appreciated.
- © The Geological Society of London