Seismic investigations along East Greenland's Fjord Region completed during the last decade provide fundamental insights into the region's crustal structure and tectonic history. A summary of models along a transect through the Kejser Franz Joseph Fjord provides a view from the Precambrian Shield to the Eocene oceanic crust. We conclude that a change of rifting geometry from an upper- to a lower-plate-style margin occurred in early Mesozoic times and formed the >350-km-wide rift zone. Despite the demonstrated asymmetry of the northeast Greenland and conjugate Vøring margins, the change of rift geometries and the direction of rift jumps remain debatable. A combined model for productivity and duration of magmatism is proposed for the northeast Greenland fjord region. We suggest that magmatism started slowly at 58.8 ± 3.6 Ma with a production rate of 1.5 × 10−4 km3 km−1 a−1, which is similar to the productivity of onshore upper and lower lava sequences on the Geikie Plateau. A peak of 9.4 × 10−4 km3 km−1 a−1 for 0.5 Myr, and a subsequent productivity of 4.4 ± 0.3 × 10−4 km3 km−1 a−1 for 2.5 Myr between 53.3 and 50.8 Ma, produced the majority of melt, but break-up did not occur immediately afterwards. Continuous production of melt, similar to the rate of ocean spreading until C22 (∼50 Ma), contributed to massive magmatic underplating until eventual break-up at 50 Ma. The volumes and production rates show similarities to those obtained from a profile off the southeast Greenland margin but with a major difference in a smaller regional spatial extent.
The northeast Greenland margin opposes the well-explored Norwegian margin across the northern North Atlantic. Margin evolution concepts and rift geometries on the Greenland margin are often assumed to be the same as on its Scandinavian conjugate (e.g. Mosar et al. 2002a). A lack of detailed crustal structure models, especially for the post-Devonian to Tertiary parts of the margin, makes this assumption hard to evaluate. Deep seismic experiments in the last decade (Mandler & Jokat 1998; Schlindwein & Jokat 1999; Voss & Jokat 2007; Voss et al. 2009) and potential field modelling (Schmidt-Aursch & Jokat 2005b) have provided fundamental data for rift system analysis at the northeast Greenland margin. Schlindwein & Jokat (1999, 2000) proposed a model for the late Caledonian extension, but did not discuss post-Devonian events. In this paper, we summarize and review the extensional structures on a traverse from the Precambrian shield west of the Greenland Caledonides, across the continental sedimentary basins and shelf region, and into the oceanic basin off the Kejser Franz Joseph Fjord (KFJF; Fig. 1). The objective of this study is to provide an overview of the crustal structures formed during tectonic extension that started with Devonian extensional collapse and culminated in early Eocene opening of the North Atlantic. The early Tertiary magmatic episodes, that is, the duration of magmatism based on predicted production rates, will be discussed on this northeast Greenland margin cross-section.
The Iapetus Ocean closed during continent-continent collision of Baltica and Laurentia in mid-Silurian times (∼425 Ma). The westward subduction of Baltic crust caused extreme crustal thickening in the Caledonian Belt (Torsvik et al. 1996). The early stage of post-collision extension and formation of the fault belt (Fig. 1) occurred in middle to late Silurian times, perhaps due to gravitational collapse and pure shear stretching of the lower crust (Andersen & Jamveit 1990; Milnes et al. 1997). An alternative model is that the subduction reverted to exhumation (Fossen & Rykkelid 1992; Rey et al. 1997) along a pre-existing, west-dipping shear zone (Schlindwein & Jokat 2000). Discrete extensional phases followed the lithospheric collapse (Dewey 1988; McClay et al. 1986) for almost 350 Myr. Extensional detachments (Hartz & Andresen 1995), fault-controlled Devonian basins (Larsen & Bengaard 1991) and syn-extension granitic intrusions along the detachment faults (Hartz & Andresen 1995) testify to the post-Caledonian extension, which ceased in late Devonian to early Carboniferous time. Price et al. (1997) suggested minor crustal stretching occurred associated with a Carboniferous rifting event on Traill Ø (Fig. 1), but larger amounts of stretching are suggested for Jameson Land (Larsen 1990; Larsen & Marcussen 1992). Existing seismic refraction data in this area (Fechner 1994; Mandler & Jokat 1998) also show evidence for thinning of the Devonian crust. The most prominent rifting event took place in late Jurassic to early Cretaceous times (Surlyk 1990), when marine sediments were deposited over the Devonian sediments on Jameson Land. North of Kong Oscar Fjord (KOF) these two sequences are juxtaposed (Fig. 1) and separated by the Gauss Halvø Fault (GHF; Schlindwein & Jokat 1999; Peacock et al. 2000). Cenozoic magmatism accompanied the final stage of rifting and the opening of the North Atlantic in the early Eocene. Price et al. (1997) provide a detailed analysis of the onshore exposures of the Cenozoic rift-related and magmatic rocks. The contrasting rifting histories north and south of the KOF led to the assumption that pre-existing crustal structures had an influence on magmatism (Schlindwein & Jokat 1999). The total amount of volcanic extrusives in the fjord region between KOF and Shannon Island (Fig. 1) is still under debate. The relatively minor tholeiitic and alkaline basalts exposed onshore may either be evidence for weak activity or alternatively represent relics of larger amounts of intrusions which were since eroded (Upton 1988; Larsen et al. 1989). Schlindwein & Jokat (1999) and Voss & Jokat (2007) proposed from deep seismic crustal velocity models major melt production in the region between KOF, KFJF and the Godthåb Gulf. Observations of high seismic velocities (>7.0 km s−1) in the lower crust are consistently interpreted in this region as magmatic underplating of the northeast Greenland continental crust. The seismic evidence for underplating diminishes rapidly in northern direction (Voss et al. 2009). Voss & Jokat (2007) proposed the presence of concealed basaltic extrusives mixed with syn-rift sediments in ∼2–6 km depth within the northeast Greenland shelf region, based on seismic velocities and gravity modelling (Fig. 1). Associated magmatism is related to the opening of the North Atlantic between Greenland and Scandinavia, starting at around 56 Ma from south to north (Larsen 1988). The earliest seafloor spreading is marked by the oldest ocean-spreading anomaly, C24B (∼54 Ma), along the North Atlantic margins (Fig. 1). Voss & Jokat (2007) proposed a delay in break-up in a zone north of the Jan Mayen Fracture Zone (JMFZ), which was locked as a result of long-term extension of the continental crust. Based on the location of the continent ocean boundary (COB) and its obliquity with respect to seafloor spreading anomalies, Voss et al. (2009) estimated break-up not earlier than 51.5 ± 0.2 Ma (C23) and 50.1 ± 0.3 Ma (C22) off Godthåb Gulf and KFJF, respectively, which is about 3.5 Myr later than that further north. Initially, enhanced oceanic crustal accretion started with half-spreading rates of about 30 cm a−1 (Voss et al. 2009), and decreased rapidly to 14–17 cm a−1 and gave rise to the production of thinner than normal (5–7 km) oceanic crust (White et al. 1992). In contrast, the oceanic crust forming off southeast Greenland at the same time is 8–12 km thick (Korenaga et al. 2000; Holbrook et al. 2001; Hopper et al. 2003).
Crustal Scale Characteristics of Rift Episodes
A 3-D density model (Schmidt-Aursch & Jokat 2005b) covered parts of the Caledonian hinterland, the fjord region and the Greenland Sea basin up to the Mohns Ridge. Inferences on the structure of the continental crust and sedimentary basins based on six deep seismic profiles (Schlindwein & Jokat 1999; Schmidt-Aursch & Jokat 2005a) over KOF, KFJF, Dickson Fjord (DF), Nordvestfjord (NVF), Fønfjord (FF) and Gåsefjord (GF; Fig. 1). A continent-ocean transition (COT) zone could not be constrained due to the poorly resolved crustal structure, and the COB after Escher & Pulvertaft (1995) was used instead. Constraints for the ocean basin came from Klingelhöfer et al. (2000a). The error in extrapolated Moho depths was estimated to ±5 km for the continental domain, and ±3 km in the ocean basin (Schmidt-Aursch & Jokat 2005b). A major uncertainty remains because of poorly known surface geology and westward extent of the Caledonides, hidden beneath the Greenland ice sheet. The newly observed and modelled COT along profile AWI-20030500 (Voss & Jokat 2007) prompt a revised conceptual cross-section through the KFJF (Fig. 1). A composite 1020-km-long traverse is shown in Fig. 2, based on three profiles of seismic refraction data and 3-D density modelling. This traverse provides an excellent basis for re-examining the rifting history and the development of the crustal structural style during post-Caledonian extensional collapse through to the final stage of continental break-up. Multichannel seismic data are not included due to the major focus on deeper crustal structural styles, which is best addressed using wide-angle seismic data. We refer readers to the original studies for details of the velocities, densities, resolutions and uncertainties of the seismic models of the following transects.
The crustal model from the Precambrian shield to the Caledonian mountains (kilometres 0–460) is deduced from a 3-D density cross-section after Schmidt-Aursch & Jokat (2005b). The inland-direction is a prolongation of the offshore seismic transects of the KFJF, and connects to the location of Summit Station (SUM; Dahl-Jensen et al. 2003; Fig. 1). The schematic surface topography (Fig. 2) was not included in the density model (Schmidt-Aursch & Jokat 2005b). The profile merges with its eastern neighbour between kilometre 400 at the top and kilometre 460 at the bottom (Fig. 2).
The post-Caledonian structure of the continental crust and sedimentary basins is based on seismic refraction profile 94320 (Fig. 1) in the seaward prolongation of KFJF (Schlindwein & Jokat 1999), between kilometres 400 and 620 (Fig. 2). Intracrustal reflections were located with an accuracy of ±2 km, and the Moho with ±3 km. The transition to the eastern-most profile cuts through kilometres 560 and 620 from top to bottom.
Profile AWI-20030500 covers the COT and the onset of oceanic crust in the Greenland basin. It forms the youngest part of the cross-section from kilometre 560 to the oceanic end at kilometre 1020. The greater number of recording stations and higher ray coverage results in an estimated accuracy of ±0.5 km for upper layers and ±2 km for lower layer boundaries and the Moho (Voss & Jokat 2007). 2-D Bouguer gravity modelling revealed similar densities for the crustal layers, sedimentary basin and upper mantle in the two profiles just described, and it confirms the COT location and high-velocity lower crust.
A review of the typical crustal units (Fig. 2) follows the interpretations of these studies, in combination with the geological maps of Escher & Pulvertaft (1995) and Henriksen et al. (2000). The surface geology, faults and large-scale crustal structures are correlated, and the main tectonic events are summarized, in a simplified timescale on Fig. 3.
Precambrian shield and caledonian orogen
Gravity modelling by Schmidt-Aursch & Jokat (2005b) yielded a thickness of the Proterozoic crust of 35 km (Fig. 2; 0–300 km) with moderate (2.93–3.00 × 103 kg m−3) lower crustal densities. This is consistent with the average of other Precambrian shields worldwide (Meissner 1986; Durrheim & Mooney 1994; Christensen & Mooney 1995; Zandt & Ammon 1995). However, Dahl-Jensen et al. (2003) inferred a Moho depth of 47 km from receiver function analysis at the SUM, which is about 5 km off the gravity transect near kilometre 125 (Figs 1 and 2). Schmidt-Aursch & Jokat (2005b) showed that using this deeper Moho in the landward prolongation of the KFJF required a density increase from 2.93–3.0 × 103 kg m−3 between 25 and 35 km to 3.1 × 103 and 3.2 × 103 kg m−3 in the 12 km lower crustal layer, to fit the Bouguer anomaly for the Moho (Fig. 2). There is no independent evidence to favour either set of densities, although lower crustal layers with high densities and seismic velocities have been predicted (Durrheim & Mooney 1994) beneath Proterozoic shields. We prefer the simple model associated with the shallower Moho, the lower deep crustal densities and a smoother Moho topography, but without rejecting the other model.
The distinct crustal root (Fig. 2; kilometres 350–500) beneath the Caledonian orogenic belt has a maximum Moho depth of 49 km below sea level, deeper than either of the alternative Moho depths modelled to the west. Clear evidence for this root came from a deep seismic profile in the NVF, and was confirmed by 3-D gravity modelling (Mandler & Jokat 1998; Schmidt-Aursch & Jokat 2005a,b; Fig. 1). The significant Bouguer anomaly low correlates with the highest surface elevations and the crustal root, revealing an overall crustal thickness of 51 km including the caledonian mountains. A crustal root is absent beneath the conjugate Scandinavian Caledonides (Meissner 1986; Kinck et al. 1991), where a significant Bouguer anomaly low is instead attributed to lower densities in the mantle (Theilen & Meissner 1979; Bannister et al. 1991). Schmidt-Aursch & Jokat (2005b) discuss different orogenic roots and their associated gravity anomalies in detail. An example for an old and preserved crustal root was also found beneath the Proterozoic Torngat Orogen, northeast Canada, between 35 and 38 km thick Archean crust in the west and the Nain Province in the east, which has been preserved for ∼1.8 Gyr (Funck & Louden 1999). A wide Bouguer anomaly low coincides with the ∼50 km deep root, whose formation is suggested either to be the result of a flip in subduction direction (eastward to westward) or, alternatively, from westward underthrusting in a late stage of collision. However, Funck & Louden (1999) attributed the preservation of the crustal root to the absence of post-orogenic heating and ductile reworking, consistent with the lack of post-collisional magmatism. The East Greenland Caledonian crustal root formed in Silurian time (∼425 Ma) during the palaeo-westward subduction of Baltic crust (Torsvik et al. 1996). Schlindwein & Jokat (2000) considered gravitational collapse or subduction reverted to exhumation along a west-dipping shear zone. Whether this and/or later heating, for example, due to the Iceland plume thermal anomaly, affected the Caledonian crustal root is presently not understood.
The major east-dipping Fjord Region Detachment (FRD; Fig. 2; Hartz & Andresen 1995; Andresen et al. 1998) separates an area unaffected by significant upper crustal extension to the west (Andresen et al. 1998), from the Eleonore Bay Supergroup to the east (Fig. 2; 450–550 km). The FRD overlies a steeply westward-dipping Moho at 40–30 km depth. Schlindwein & Jokat (2000) proposed that the FRD terminates at a lower crustal reflector in ∼13 km depth, and that it therefore does not represent a crustal-scale detachment. Those authors proposed that the overthickened post-orogenic Caledonian crust collapsed along a west-dipping shear zone between kilometres 500 and 600 (Fig. 2), which is marked by a prominent lower crustal reflector. They suggest lower crustal displacement along this shear zone, following either a simple shear or a delamination model. The Western Fault Zone (WFZ; Fig. 2) developed during the Devonian subsidence of the thinned crust, and the basin filled with Devonian continental sediments (Larsen & Bengaard 1991; Escher & Pulvertaft 1995). The location of the WFZ at the surface correlates with a Moho high in 30 ± 3 km depth. Schlindwein & Jokat (2000) concluded that this step in the Moho was preserved when the first major rifting phase gradually shifted to the east between late-Devonian and early-Carboniferous times.
The initiation of a second major rifting phase in middle Jurassic times (Surlyk 1990) led to the evolution of Mesozoic sedimentary basins, which lie to the east of the Devonian basin. Middle to upper Jurassic sediments were deposited in fluvial and shallow marine settings (Price et al. 1997). A second Moho slope developed during that rifting episode, underlying the GHF (Fig. 2) between the two major sedimentary sequences. The increased seismic velocities of the lower crust in 16–30 km depth (Fig. 2; kilometres 600–680) were interpreted to be a result of (1) the displacement of lower crustal material along the west-dipping shear zone (Schlindwein & Jokat 2000) as described above and (2) Tertiary magmatism (Schlindwein & Jokat 1999; Voss & Jokat 2007). The strong lateral velocity gradient at kilometre 600 can be either a structural boundary between rifted lower continental crust and a pure magmatic underplated body or as the result of an increase in lower crustal intrusions. The correlation between a strong negative magnetic anomaly off the coast (Fig. 1) and the expected magnetization of a lower crustal body led Schlindwein & Jokat (1999) to conclude that the high velocities between kilometres 600 and 680 represent a magmatic underplate.
Continent-ocean transition zone
Rifting persisted until late Cretaceous/early Tertiary times and formed deep-marine clastic wedges of up to 2600 m thickness exposed in Wolaston Foreland (Surlyk 1978, 1990). Voss & Jokat (2007) suggested that the latest stage of the Cretaceous to Tertiary rifting phase might have been accompanied by magmatism that significantly influenced the style of the COT. Extrusive basalts intercalated with syn-rift sediments form up to 5.4-km-thick layer between kilometres 690 and 810 (Fig. 2). A further conclusion was that a large degree of magmatic intrusions had resulted in an increase in seismic crustal velocities between kilometres 670 and 800 (Fig. 2) in 6–18 km depth (Voss & Jokat 2007) to values (6.6–6.8 km s−1) that are significantly above the global average for extended crust at such depths (Christensen & Mooney 1995). Voss & Jokat (2007) related magnetic anomalies to such intrusions.
A major structure of the COT is a 210 km wide and up to 15-km-thick lower crustal body, interpreted as a solidified magmatic underplate beneath the rifted continental crust (Voss & Jokat 2007). The current transition to mantle rocks occurs beneath the underplate at a depth of 26–28 km, shallowing rapidly eastwards towards the oceanic crust. Schlindwein & Jokat (1999) concluded from the presence of minor exposures of onshore Tertiary plateau basalts on Bontekoe Ø (Fig. 2; kilometre 670), Hold with Hope and Traill Ø (Fig. 1), that the underplate formed contemporaneously with the Tertiary plateau basalts on the Geikie Plateau south of Scoresby Sund (Fig. 1). We cannot preclude the possibility that the thick high-velocity lower crustal body (HVLC) contains fragments of inherited and exhumed Caledonian crust, as in the Norwegian Vøring basin (Gernigon et al. 2004). A major difference is, however, the complex magnetic pattern in the northeast Greenland margin associated with major intrusions in the crustal layers. To what extent the HVLC contributes to these magnetic anomalies is open to question, given its depth range of 15–30 km and the unknown level of the Curie temperature (540–570 °C). Schlindwein & Jokat (1999) attributed the large negative magnetic anomaly off the northeast Greenland fjord region (Fig. 1) to the magmatic underplate and expected the demagnetization level to lie below 20 km. We favour, for further consideration, a massive magmatic body and assume the majority of the melt accumulated at the crust-mantle boundary.
Thermal subsidence of the Norwegian-Greenland rift system initiated the deposition of Cenozoic sediments (Fig. 2; kilometres 690–820), which form the top layer of the present East Greenland shelf. The onset of oceanic crust is marked by a deep Cenozoic sedimentary basin (kilometres 810–840) and a steep rise of the Moho to ∼14 km (Voss & Jokat 2007). The accumulation of the magmatic underplate terminated at the time of break-up, and normal accretion of oceanic crust began.
Voss & Jokat (2007) proposed rift propagation from north to south along the Fjord region margin. Break-up was estimated to have occurred last off the KFJF, at close to C22 (∼49.4 Ma) time. Anomaly C21 (∼47.1 Ma) is the first clearly identified magnetic ocean spreading anomaly, at kilometre 850 (Figs 1 and 2). The total thickness of the early Eocene oceanic crust decreases from 7 to 4.8 km (Fig. 2; kilometres 820–980), but with a local maximum of 11.5 km beneath a fragment of the JMFZ (Voss & Jokat 2007; Fig. 2; kilometre 925).
The composite cross-section (Fig. 2) demonstrates the relative dimensions and extents of tectonic and magmatic structures of the COT, the extensional basins and rifted continental crust. Compared to the Precambrian crust and the Caledonian crustal root, a high degree of crustal thinning occurred over a 350-km-wide region from Devonian to Cretaceous times. Post-collision extension involved a significant initial vertical movement of crustal material as seen on the steep eastern flank of the crustal root (Fig. 2; kilometres 460–530). An effective mechanism for crustal thinning and vertical displacement of crustal material is extension along an asymmetric crustal-scale detachment models. The simple shear model (Wernicke 1985) requires a low-angle fault cutting through the entire lithosphere. Extension is accommodated along the fault, and produces related normal faults from upper crustal brittle deformation in the hanging wall. Here, fault-bounded basins develop and fill with clastic sediments. A Moho slope forms where the shear zone offsets. An alternative scenario is the delamination model (Lister et al. 1986), which considers a crustal-scale detachment fault, which shears horizontally beneath the brittle/ductile layer boundary of the crust and beneath the Moho. Extension along the shear zone produce equivalent normal faulting at the surface. In either case, asymmetric margins evolve depending on their location relative to the shear zone, as conceptually illustrated in Fig. 4 after Lister et al. (1986). The hanging wall is referred to as the upper-plate margin with rocks originally above the shear zone, and with a simple structured basement (Lister et al. 1986). Uplift of the continental crust is a response to lateral translation of dense lithospheric material and upwelling of warmer and less denser asthenosphere. Underplating of igneous material at the crust-mantle boundary and normal fault sequences dipping at the surface towards the newly developing ocean are characteristics of an upper-plate margin configuration. The lower-plate margin refers to the footwall, which exposes deeper crustal rocks and hosts wider sedimentary basins. Movement along the major detachment fault and the removal of upper crustal material accompany crustal thinning, subsidence and upward buckling of the lower crust. Characteristics of both upper- and lower-plate margins can be identified along the margin transect in Fig. 2. We propose a rifting model involving a change from an upper- to a lower-plate margin, and which involves subsequent magmatic overprinting.
Upper-plate margin segment
The main evidence for an upper-plate margin geometry is the inferred westward-dipping lower crustal shear zone (Figs 2 and 5a) with marked reflectivity in 22–40 km depth (Schlindwein & Jokat 2000). Fountain et al. (1984) related lower crustal reflectivity to the seismic anisotropy of mylonites along shear zones. The proposed shear zone can be associated with the landward-dipping detachment fault (Lister et al. 1986), marked in Fig. 4. Additional structures supporting this model are the major faults, FRD and WFZ, which formed during the first stage of rifting (Fig. 3). Initial vertical displacement of the lower Caledonian crust (Schlindwein & Jokat 2000) is consistent with the early stage of evolution of an upper-plate margin. An eastward prolongation of the proposed shear zone, which cuts through the upper crustal layer (Fig. 5a), can be deduced from the strong lateral upper crustal velocity increase indicated by the dashed line in Fig. 2, from ∼6.3 to ∼6.6 km s−1 (Voss & Jokat 2007). Lavier & Manatschal (2006) describe differential crustal motion of up to 10 km vertically accommodated at a ductile shear zone in the thinning mode of the extensional rifting model for non-volcanic margins.
The eastward decrease in exposure of upper crustal material between kilometres 550 and ∼800 in Fig. 2, and the accompanying evolution of deep extensional basins underlain by large-scale magmatic underplating, do not fit into a full upper-plate margin model. Thus, we suggest a change from the upper-plate margin configuration during the first Devonian/Carboniferous rifting stage to a lower-plate margin configuration in the following rifting stages.
Lower-plate margin segment
We propose a change to a lower-plate margin with the initiation of the second major phase of rifting further east in the mid-Jurassic (Fig. 3). Supporting evidence comes from Mosar et al. (2002a), who inferred as much from the extensional normal faults and the east-dipping FRD, WFZ and GHF. Extension was accommodated by wide-fault blocks (Price et al. 1997), whereas, in lower Cretaceous times narrow-fault blocks developed along detachment faults as a consequence of the movement of the eastern (upper) plate. Upper crustal material thinned to some degree during the extensional movement. It is not resolved and thus questionable, whether the post-Caledonian sedimentary basins show eroded half-graben structures at the bottom (Fig. 5b). We suggest from the trend of the isopachs in Voss & Jokat (2007) that the lower crustal material, vertically displaced along the crustal shear zone during the Devonian upper-plate margin stage, bows further upwards during the lower-plate margin stage. To some extent, the lateral velocity increase from ∼6.6 to ∼6.8 km s−1 (Voss & Jokat 2007; Fig. 2, marked by dashed line at kilometres 600–650) may mark the transition between the brittle upper crust and ductile lower crustal material in 15–20 km depth, as well as the presence of later magmatic intrusions. A consequence of crustal thinning to less than 10 km was subsidence and the emplacement of syn-rift sediments. Lister et al. (1991) describe a similar sequence of crustal thinning and subsidence for Atlantic-type rifted margins based on observations from the US rifted margins. We envision that intrusions into the rifted crust, during Tertiary magmatism at the end of the second rift phase (Fig. 5c) further increased the seismic velocities within the COT, so that originally rifted upper and lower crust cannot be distinguished. There may have been stages of uplift during magmatic underplating, but we neither quantify this nor the accompanying erosion. However, post-magmatic thermal subsidence of ∼3 km can be deduced from the thickness of the post-break-up sedimentary deposits (Fig. 2).
It remains unresolved to what degree the rifted crust is intruded within the COT, and to what extent the high-velocity lower crustal body consists of heavily intruded and stretched continental crust. Sills in continental crust are likely to result in a heterogeneous structural style with enhanced internal seismic reflectivity, but Schlindwein & Jokat (1999) and Voss & Jokat (2007) reported only clear top and bottom reflections of the high-velocity lower crustal body, as shown in Fig. 2.
Asymmetric rifting of conjugate margins
Torske & Prestvik (1991) proposed a rift-configuration model with an east-dipping main crustal detachment south of the JMFZ and a west-dipping one north of the JMFZ, and suggested an upper-plate style for the East Greenland margin and a lower-plate type Vøring margin. Mosar et al. (2002a) also demonstrated asymmetric rifting and the development of upper- versus lower-plate margins for Norway and East Greenland based on cross-sections north and south of the JMFZ. However, the latter authors proposed a lower-plate margin development for East Greenland north of the JMFZ, and did not discuss the influence of magmatism for the latest stage of rifting at either margin. Mjelde et al. (2003) favoured a delamination model for the Vøring margin which started as a lower plate in late Cretaceous and early Tertiary times and switched to an upper-plate geometry in response to the arrival of the Icelandic hotspot and a westward jump of the rifting axis. Our model shows a similar jump from an upper- to a lower-plate margin, but suggests the timing of this jump was probably in mid-Jurassic times. The detachment surface is proposed near the top of ductile and heavily intruded lower crust in both models. In light of the uncertainties of the conjugate positions of Kejser Franz Joseph Profile and the Vøring Plateau profile (Voss & Jokat 2007), these two models agree on the evolution of asymmetric margins during the late stages of rifting. They disagree, however, in the timing of tectonic events and styles of rifting in the earlier stages.
Duration and Production Rates of Northeast Greenland Magmatism
Estimates of the production rates and the duration of magmatism are very rough because of the large uncertainty in the amount of magma intruded into the lower crust, the proportion of melt that was erupted as basalts and the effect of erosion (Upton 1988; Larsen et al. 1989). It should be noted that these rates represent half production rates in units of km3 per unit length per year (km3 km−1 a−1) based on calculations along only one side of the rift zone.
Table 1 summarizes the total volumes for each part of the rift zone influenced by magmatism and the assumed proportion of melt within the layers. We cannot predict the proportion of basalt and sedimentary rocks at kilometres 690 and 810 (Fig. 2). Thus, we assume it to vary between 0 and 100 per cent with an average of 50 per cent, that is, 210 ± 210 km3 of basalts. Increasing seismic velocities within the continental crust (kilometres 615–690) and the COT between kilometres 690 and 815 (Fig. 2) could be due to uplifted, dense lower crust, as described above, and/or to intrusions. Therefore, we assume a low variability of 0–20 per cent, averaging 10 per cent, of volcanic intrusions in the rifted continental crust, that is, 210 ± 210 km3 (Table 1), which is most likely an underestimate. The magmatic underplate (kilometres 600–810) is assumed to consist of entirely magmatic material and has a volume of 1990 km3 (Table 1). The total volume of basalts, crustal intrusions and magmatic underplating is thus estimated at 2410 ± 420 km3. An equivalent amount can be assumed from the adjacent profile off the Godthåb Gulf (Voss & Jokat 2007).
Voss et al. (2009) proposed a N-to-S propagation of break-up at the northeast Greenland margin, starting at the Greenland Fracture Zone at 54.2 Ma and ending at 50 Ma off the KFJF. We assume, in a first approach, that magmatism was continuous throughout this period and to have had a constant production rate (Table 2). Published half production rates are used for a second calculation and an alternative duration of magmatism is estimated for the region off the KFJF (Table 3).
Voss et al. (2009) proposed break-up off the KFJF at 50 Ma near the maximum of the normal polarization of C22. A constant half production rate as high as 5.7 ± 1.0 × 10−4 km3 km−1 a−1 must have continued for 4.2 Myr to account for the estimated total magmatism off the KFJF. This corresponds to a half rate of 4.15 × 10−4 km3 km−1 a−1 (Table 2), which is both lower and sustained for 1.2 Myr less than our estimate. As an alternative to constant production, the latter authors suggested melt production increased during break-up, and then underwent a gradual decrease.
The present day Mohns Ridge (Fig. 1) is an ultra-slow spreading ridge (∼0.8 cm a−1; Klingelhöfer et al. 2000b) with a mean crustal accretion of 4.0 ± 0.5 km for the period between 20 Ma and present. This corresponds to a magma productivity of just 0.3 × 10−4 km3 km−1 a−1 (Table 2). More recent publications (Mosar et al. 2002b) proposed initial half spreading rates of 1.3–1.8 cm a−1 for the Mohns Ridge and ∼1.5–2.1 cm a−1 for the Aegir Ridge. Voss et al. (2009) have shown that half spreading rates decreased from an initial maximum of 2.2–2.9 cm a−1 between C23 and C24 to about 1.4 ± 0.1 cm a−1 until C22 with a corresponding decreasing crustal thickness from ∼10 to 5 km. These values would yield a magma production rate of 0.7–2.9 × 10−4 km3 km−1 a−1 (Table 2). Comparable values were found for central-eastern Greenland plateau basalts (Table 3), which range between 1.15 and 1.5 × 10−4 km3 km−1 a−1 for the northern Geikie Plateau to ∼2.0 × 10−4 km3 km−1 a−1 for the southern Geikie Plateau and the area between 69°N and Kangerdlugssuaq (Larsen et al. 1989). Basalts on Iceland and onshore East Greenland were emplaced at similar rates (Nielsen & Brooks 1981; Table 3).
Assuming a constant production rate of 1.5 × 10−4 km3 km−1 a−1, based on an average of the published values for extrusive basalts, a time interval of 16.1 ± 2.8 Myr would be required to produce the observed 2410 ± 420 km3 of magmatic material off the KFJF. This, in turn, would place the initiation of magmatism at around 66.1 ± 2.8 Ma, which is older than most estimates. For example, lower plateau basalts were related to a 60–62 Ma igneous phase around Kangerdlugssuaq (Saunders et al. 1997) and from drill site 917 off southeast Greenland (Saunders et al. 1998). Plateau lavas on Hold with Hope and Wolaston Foreland were also related to the lower series (Upton et al. 1995). Price et al. (1997) concluded a main period of volcanism in northeast Greenland at 60–54 Ma.
The two estimates, one with a production rate of 5.7 ± 1.0 × 10−4 km3 km−1 a−1 for 4.2 Myr, and one with a period of magmatism of 16.1 ± 2.8 Myr for an average productivity of 1.5 × 10−4 km3 km−1 a−1, represent end-member models. A model with a heterogeneous, rather than constant, productivity rate appears to be most likely. For instance, the eastward increase in thickness of the magmatic underplate may point to an eastward (i.e. later) increase in productivity. Larsen et al. (1989) also calculated an increased productivity for the latter episode of NAVP volcanism (Table 3).
We propose a four stage productivity model including production rates derived from onshore plateau basalts and crustal accretion rates of northeastern Greenland. The derived initial magmatism correlates well with the dating of earliest emplaced onshore basalts. Backwards calculation from the time of break-up is necessary to achieve the initial duration. Production rates, duration of the stages and total magmatic volumes are listed in Table 4 and schematically shown in Fig. 6. Half spreading rates for northeast Greenland's oceanic crust and average crustal thicknesses are taken from Voss et al. (2009), and the ages of spreading anomalies from Cande & Kent (1995).
Stage IV: The latest stage comprises the magmatic material, which was accreted as oceanic crust from C23 (50.8 Ma) on, but remained beneath the COT zone until the local break-up at 50 Ma. The average half spreading rate for this time interval is proposed as 1.4 ± 0.1 cm a−1 and an average crustal accretion of 8 km yields a production rate of 1.1 ± 0.1 × 10−4 km3 km−1 a−1, corresponding to a volume of 88 ± 8 km3 (Table 4).
Stage III: An average productivity of 4.4 ± 0.3 km3 km−1 a−1 is deduced from studies of the NAVP after Eldholm & Grue (1994) (4.15 × 10−4 km3 km−1 a−1) and the lowest limit of the proposed range of the rate for the region off KFJF (4.7 × 10−4 km3 km−1 a−1; Table 2). This stage has a duration of 2.5 Myr, starting from C23 (50.8 Ma) and lasting until 53.3 Ma (C24B), and produces a magma volume of 1100 ± 75 km3.
Stage II: A peak of a magma production of 9.4 × 10−4 km3 km−1 a−1 within 0.5 Myr (Eldholm et al. 1989; Eldholm & Grue 1994) is used between 53.3 and 53.8 Ma. This results in a contribution of 470 ± 32 km3 for an equivalent error to that for stage III of 7 per cent.
Stage I: The remaining magmatic volume produced in the initial stage would be 752 ± 535 km3 (Table 4) deduced from the difference of the total volume along the transect (2410 ± 420 km3) and the calculated volumes of the later three stages (1658 ± 115 km3). At a production rate of 1.5 × 10−4 km3 km−1 a−1, based on that for the emplacement of the plateau basalts (Tables 3 and 4), the duration of stage I would be 5.0 ± 3.6 Myr.
The model yields a total duration of melt production of 8.8 ± 3.6 Myr (Table 4) and suggests an initiation of magmatism at between 55.2 and 62.4 Ma (Fig. 6). This timing correlates well with the dating of lower plateau basalts (60–62 Ma) around Kangerdlugssuaq (Saunders et al. 1997) and the plateau lavas on Hold with Hope and Wolaston Foreland (60–54 Ma; Upton et al. 1995; Price et al. 1997).
From the poly-productivity model, we conclude that the maximum magma production at the northeast Greenland margin, that is, off the KFJF and probably also off the Godthåb Gulf, is consistent with the general estimated production rates of extrusives and intrusions for the NAVP. Initiation of magmatism is suggested to have started several million years before the maximum burst of magma in stages II and III (C24B; 53.8 Ma-C23; 50.8 Ma). However, whereas in other regions along the Greenland margin stage III coincides with break-up and emplacement of extensive SDRS basalts, this peak of magmatism off the fjord region preceded break-up there by a few million years. We infer from the change of upper- to lower-plate margin configuration and the associated seaward shift of the rifting centre that the rifted continental crust is only partially weakened, which had an effect on the delayed break-up.
A similar sequence can be discerned at the southeast Greenland margin. Lower lava series on continental crust are dated at 60–60.5 Ma (Saunders et al. 1998) from ODP drillhole 917. Hopper et al. (2003) proposed a model of magmatism and accretion of oceanic crust along a seismic transect SIGMA III (Fig. 1). These authors estimated an initial half spreading rate of 3.3 cm a−1 and identified new igneous crust, which has accreted since 56 Ma. Igneous crustal thickness decreases from 18.3 to 13.5 km within 3 Myr, giving an average productivity of 5.3 × 10−4 km3 km−1 a−1. The productivity decreased further to less than 2.3 × 10−4 km3 km−1 a−1 until 50.8 Ma and 1.9 × 10−4 km3 km−1 a−1 until approximately 50 Ma (Table 2). Continental basalts were proposed to have been emplaced within 1 Myr (at some time between 56 and 61 Ma) with a half production rate of ∼1.12 × 10−4 km3 km−1 a−1 (Saunders et al. 1998; Table 3). The crustal thicknesses, half spreading rates and the emplacement of the continental succession reveal a total volume of approximately 2490 km3 (Fig. 6). The volume produced off the northeast Greenland margin thus seems equivalent to that at the southeast Greenland margin. The major difference of these two regions is the regional extent of magmatism at the southeast Greenland margin compared to the more localized occurrence of large magmatic volumes off the fjord region (Voss et al. 2009). The poly-productivity model with its suggested production rates supports Voss et al.'s model of magmatism sourced from one major conduit system in the vicinity of the Icelandic mantle plume and provides a maximum estimate of melt production between 62.4 and 50 Ma for the region off the KFJF, and probably as far as off Godthåb Gulf, which comprises a similar amount of magmatic underplating (Voss & Jokat 2007). Voss et al. (2009) proposed alternative models invoking highly intruded continental crust rather than pure magmatic underplating, or a second phase of magmatism linked to the Oligocene separation of the Jan Mayen microcontinent (Fig. 3; Gudlaugsson et al. 1988; Upton et al. 1995; Price et al. 1997). Late Eocene, Oligocene and Miocene magmatism is not included in our model here, although onshore intrusions do indicate ongoing activity (Upton et al. 1995; Price et al. 1997) associated with the separation of the Jan Mayen and the 30–25 Ma initiation of ocean spreading at the Kolbeinsey Ridge (Larsen 1990; Kodaira et al. 1997). The proposed production rates well explain the observed amount of melt off the KFJF and correlate with dated onshore basalts. If the high-velocity lower crustal body has a higher content of rifted continental crust, or if the volume of Oligocene magmatism was comparable to that of the initial early Tertiary event, then the production rates are overestimated and need to be re-evaluated. But the degree of continental crustal content in the high-velocity body and the amount of sill intrusion in upper crustal layers remain debatable and further investigations are necessary. However, slightly lower production rates and a portion of the volume associated with the earliest stage (752 ± 535 km3) could be related to the Oligocene magmatism.
A crustal transect across the entire East Greenland passive margin is provided, based on deep seismic refraction data and 3-D gravity data modelling. It extends from the Precambrian shield, and through the Caledonian Foreland, to the early Eocene oceanic crust in the prolongation of KFJF. A polyphase rift history and magmatism based on crustal-scale structures can be summarized as follows.
The lithospheric collapse of Caledonian crust initiated the configuration of an upper-plate margin. Devonian rifting corresponds with the displacement of the lower crustal layer along a landward-dipping detachment fault and the initiation of major seaward-dipping faults at the surface. The reflectivity of the deep-seated shear zone can be related to the velocity contrast and/or mylonites along the shear zone.
The geometry changed to a lower-plate margin configuration with the initiation of the second long-term rifting event in Jurrasic times, as suggested by the pattern of velocity increase in the higher crustal levels. The ductile lower crust bowed up, and brittle upper crustal material was eroded during the large-scale extension accompanying the eastward movement of the eastern upper-plate margin. Large low-angle oceanward-dipping faults developed, or were reactivated from the previous rifting event, forming sedimentary basins. An earlier crust-mantle boundary is assumed to be marked by the top reflector of the magmatic underplate at 15–17 km depth.
Margin uplift is assumed with the initiation of early Tertiary magmatism and the emplacement of the magmatic underplate. Basalts erupted either subaerially or in a shallow water setting on uplifted lower crust. Intrusions additionally modified the rearranged and displaced upper and lower crust.
The history of asymmetric rifting for the northeast Greenland lower-plate margin is consistent with the suggestion of an upper-plate margin configuration on the Vøring Plateau in early Tertiary. Our arguments concerning changes of the rift geometry and jumps of the rifting axis during earlier times are, however, at odds with previous studies.
A four stage model is proposed to explain the production of the observed magmatic volume within the rift zone. Magmatism started at 58.8 ± 3.6 Ma, slowly, with the emplacement of onshore plateau basalts. During the peak of magmatism, most of the produced melt accreted along the rift zone but break-up did not occur in the region off the KFJF. Here, ongoing reduced magmatism weakened the highly extended crust. Prior to break-up at C22 (50 Ma), magma remained beneath the transition zone similar to the accretion during seafloor spreading between C23 and C22 elsewhere. The extruded basalts were concealed by the subsequent subsidence of approximately 3–6 km and Cenozoic sedimentation. This sequence of magmatism explains the suggested amount of melt and supports a model of magmatism from one major feeder system into the region off the KFJF and Godthåb Gulf. In the case that our melt volumes are overestimated due to a higher content of continental crust within the high-velocity body or additional Oligocene magmatism associated with the separation of the Jan Mayen Ridge, other models would remain tenable.
We thank M. C. Schmidt-Aursch and V. Schlindwein, who provided the additional digital gravity and velocity models used for the full transect. The manuscript was also greatly improved by comments from V. Schlindwein and Graeme Eagles. The manuscript profited from the thorough reviews of Robert Trumbull and two other anonymous reviewers. All figures were created with GMT (Wessel & Smith 1998).