We present data that link Scandinavia’s passive-margin domains under a unified system invoking isostatically driven, postextension phase vertical adjustments to severe crustal thinning. Topographic and geological data indicate that the relative location of the first landward occurrence of total crustal embrittlement or deformation coupling—the Taper Break—controlled and continues to control Scandinavia’s post-thinning geomorphic evolution. Formed during Late Jurassic or Early Cretaceous thinning, yet marked today by seismicity, the Taper Break closely approximates the boundary between (1) less-stretched lithosphere that increases in rigidity both toward land and through postrift time, and (2) the highly attenuated, pervasively faulted, permanently weakened lithosphere of the distal margin. Following the stretching, thinning, and exhumation phases proposed by other workers, an accommodation phase is warranted. Commencing during “sag” basin time and continuing today, it is probably driven by thermal cooling and mass transfer from the escarpment to the basins offshore. The accommodation phase does not entirely coincide with the traditional postrift phase as the former may contain the latter. During accommodation, the original synrift escarpments can be eroded to very low base levels. Sharply tapered margin segments can undergo subsequent rejuvenation by out-of-sequence normal faulting and footwall uplift, probably in response to tensile bending stresses engendered by lithospheric-scale flexure. Accommodation-phase uplift at passive margins is the inexorable and penultimate phase of hyperextension, and may perhaps be followed by the onset of subduction localized by the weakened lithosphere of the distal margin and the ocean-continent transition.
At the scale of half-graben basins and footwall ranges, extension and the consequent evolution of landscape are well understood (e.g., Leeder and Jackson, 1993; Gawthorpe and Leeder, 2000; Densmore et al., 2004). Robust and oft-cited models describe integrated relationships between the tectonic unloading of extensional footwalls and the differential surface uplift that controls subsequent drainage evolution (see above). However, although all passive margins are the final products of continental extension, and the structures that control their crustal thinning are an order of magnitude larger than more frequently studied half-graben bounding faults (Manatschal, 2004), a tectonics-based understanding of their landscape evolution has not yet emerged.
Models predict that seaward-facing escarpments should form at actively extending margins (e.g., Weissel and Karner, 1989). However, the persistence of sharply asymmetric, continental-scale back tilts along many post-breakup margins (see Holtedahl, 1953!) poses a well-known conundrum. Observations of range-front, “out-of-sequence” normal faults are either well documented (Zalán and Oliveira, 2005; Redfield et al., 2005; Osmundsen et al., 2009, 2010; Viola et al., 2005, 2012) or indicated by thermochronology (Oliveira et al., 2000; Kounov et al., 2009) at several margins that are well into their post-breakup phase. Citing a global association between pronounced topographic escarpments and sharp offshore crustal thinning gradients, Osmundsen and Redfield (2011) suggested that the templates of continental extension can exert control over fault reactivation and onshore landscapes for tens to more than one hundred million years after seafloor spreading begins (see GSA Data Repository Fig. ES-11).
Worldwide, the highest and most sharply asymmetric passive-margin escarpments seem to occur at sectors where the crystalline continental crust was tectonically eviscerated from a thickness of ∼40 km to virtually nothing over a relatively short distance (Osmundsen and Redfield, 2011). In at least Scandinavia and SE Brasil, the distribution of drainage patterns, landscapes types, and sediment transport routes is profoundly asymmetric, and appears to be guided by normal faults that are younger than the main phase of rifting (Redfield et al., 2005; Zalán and Oliveira, 2005; Osmundsen et al., 2009, 2010). These observations carry global implications.
In this contribution, we document the remarkable influence exerted by the Mesozoic crustal thinning gradient over Scandinavia’s present-day topography. We define the width of the passive margin to include the uplifted hinterland. We discuss relationships between onshore “out-of-sequence” normal faulting and the tectonic imprint imposed by such faulting on the landforms of the emergent part of the Scandinavian passive margin. We interpret the seismicity of Scandinavia’s proximal margin and the hinterland break in slope from within the framework of hyperextension. We conclude by extending our discussion to a more global scope, presenting a model where reactivation of normal faulting and topographic rejuvenation at sharply tapered passive-margin sectors constitute the inevitable beginning-of-the-end of a Wilson cycle.
HYPEREXTENSION AND THE PASSIVE MARGIN
The term “hyperextended” is increasingly being used to describe margins in which the basement continental crust underwent such a high degree of thinning that, during large-magnitude extension, the brittle-ductile transition migrated to the underlying mantle (Peréz-Gussinyé and Reston, 2001; Sutra and Manatschal, 2012). One possible result can be the exhumation of mantle rocks to Earth’s surface along low-angle detachments (e.g. Péron-Pinvidic and Manatschal, 2009). Alternatively, low-angle faults may simply penetrate the crystalline crust to sole in the mantle (e.g. Ranero and Peréz-Gussinyé, 2010). In either case, crustal reduction to total embrittlement—commonly 10 km thickness—was achieved by a multistage process (Lavier and Manatschal, 2006; Péron-Pinvidic and Manatschal, 2009; Manatschal et al., 2010).
Hyperextended margins have been classified into distinct distal and proximal domains (Fig. 1; see Manatschal et al., 2010). The distal margin is composed of extremely attenuated continental crust, commonly less than 10 km thick (Manatschal et al., 2010). Distal thinning can be enabled by low-angle normal faults with displacements in the order of tens of kilometers (Boillot et al., 1987; Whitmarsh et al., 2001; Manatschal et al., 2001; Peréz-Gussinyé and Reston, 2001; Thinon et al., 2003; Reston, 2005; Tucholke et al., 2007; Osmundsen and Ebbing, 2008; Péron-Pinvidic and Manatschal, 2009). Basement rocks of the distal margin may be completely crosscut by multiple generations of normal faults (Reston, 2005). By contrast, the proximal margin is characterized by faulted basins with structures that penetrate only the upper crust and tend to sole out at ductile, midcrustal levels (Fig. 1; Manatschal, 2004; Péron-Pinvidic and Manatschal, 2009; Manatschal et al., 2010; Sutra and Manatschal, 2012). In the proximal margin, stretching factors are relatively low, and pure shear extension (the stretching phase of Lavier and Manatschal, 2006) dominates. Corresponding to the continental shelf (Manatschal et al., 2010), the crystalline crust of the proximal margin is always greater than 10 km and tends to consistently thicken toward its landward side. The proximal and distal margins are separated by the “necking zone,” a localized area where the crust underwent a dramatic reduction from ∼30 km of thickness (Manatschal et al., 2010).
The Taper Break (Fig. 1) was defined by Osmundsen and Redfield (2011) as the first point seaward of the coast where the crust has been thinned to 10 km or less. On sharply tapered margins, this point will be located close to shore, at the outer end of a very pronounced necking zone (Osmundsen and Ebbing, 2008; Osmundsen and Redfield, 2011). On more gently tapered margin segments, the Taper Break may lie very far out to sea. As the crustal wedge thickens landward of the Taper Break, the submerged portion of the proximal margin may blend seamlessly with the emergent portion, presenting the question of where it should end.
Because most industry seismic lines stop well offshore, the transition to “normal” continental crust is very difficult to identify. A mean continental Moho depth of 39 km ± 3 km derived from emergent, nonorogenic areas of the CRUST2 model (Bassin et al., 2000) might be considered to represent a rough guide for the “normal” thickness of unstretched continental crust (Fig. ES-2 [see footnote 1]; see Christensen and Mooney, 1995). It follows that if the proximal margin is composed of extended crust, it commonly reaches well onshore (see Mosar, 2003). Recent studies (Stratford et al., 2009; Svenningsen et al., 2007; Ebbing et al., 2012) show that under these definitions, Scandinavia’s proximal margin extends to, or slightly beyond, the topographic crest.
Because the onshore portions of many extended margin sectors display a well-defined hinterland back slope that dips more gently than the seaward-facing slope (see Weissel and Karner, 1989), the escarpment crest constitutes a natural and easily defined proxy for the landward boundary (Osmundsen and Redfield, 2011). Adorned by preglacial relict surfaces (e.g., Lidmar-Bergsrtöm et al., 2007), Scandinavia’s hinterland slope persists nearly to the Gulf of Bothnia. Thus, for Scandinavia, and likely for a number of other passive margins, we define a third, topographically distinct hinterland (Figs. 1 and 2). Commencing just landward of the proximal margin’s inboard escarpment and ending at a pronounced break in slope (Fig. 1), it presents a sharp contrast to the flatter, near-sea-level continental interior (Figs. ES-3 and ES-4 [see footnote 1]). Incorporating previously published architectural language for the offshore parts of extended margins (e.g., Manatschal, 2004; Manatschal et al., 2010), our terminology describes the passive margin as a unified system, where each element has been uplifted or has subsided and where onshore and offshore components are linked by the structures and concepts of hyperextension. We next identify, analyze, and discuss elements of Scandinavian geology and geomorphology from within this new context.
CRUSTAL TAPER AND ESCARPMENT ELEVATIONS IN SCANDINAVIA
Several magma-poor margins and some magma-rich ones are now considered hyperextended (Manatschal et al., 2010). For example, recent work suggests that Scandinavia’s Eocene magmatic breakup was preceded by a Jurassic–Cretaceous phase of magma-poor hyperextension (Osmundsen and Ebbing, 2008; Lundin and Doré, 2011). Following this phase of extension and prior to breakup, enormous volumes of sediment were deposited in offshore “sag” basins as the synrift escarpments were destroyed by erosion (Osmundsen and Ebbing, 2008). Today, the location of the Taper Break appears to directly control the height of the onshore escarpment (Figs. 1 and 3; Figs. ES-1 and ES-3 to ES-8 [see footnote 1]; Osmundsen and Redfield, 2011). Some tens of millions of years after high-beta thinning, the hinterland and the inboard portion of the proximal margin still dance to the music of hyperextension.
To rigorously investigate spatial relationships among the key elements of Scandinavia’s passive margin, we measured distances along 233 continent-ocean boundary perpendicular profiles. For each profile, we calculated the apparent taper length (Fig. 1), the total beam length (Taper Break to hinterland break in slope; Fig. 1), and the maximum elevation. We ran a variety of experiments testing correlations between elevation, crustal taper, and hinterland morphology (see Figs. ES-3 to ES-8 [see footnote 1]). A full description of our methods, our models, and their estimated geological errors is provided in the supplemental material that accompanies this article (see footnote 1).
Because Scandinavia’s offshore margin structure is well known from seismic-refraction and seismic-reflection studies (Raum et al., 2002; Mjelde et al., 2003; Ebbing et al., 2006; Faleide et al., 2008; Ebbing and Olesen, 2010; Reynisson, 2010, and citations therein), our measurements can be made with confidence. However, like several other passive margins, the Scandinavian margin is characterized by lower-crustal bodies (LCBs) exhibiting seismic velocities higher than expected for the lower crust and lower than expected for the mantle (see Eldholm and Mutter, 1986; Ebbing et al., 2006). Because inclusion or exclusion of the lower-crustal bodies in the calculation of crustal thickness can significantly change the location of the Taper Break, we developed two end-member families of models, including and excluding the lower-crustal bodies.
Separating the apparent taper length data into subsets by geographically contiguous regions exposes a high degree of internal order (Fig. 3A; Fig. ES-6 [see footnote 1]). For example, apparent taper length measurements from the Møre region correlate particularly well with elevation. The best correlations of all are obtained by plotting the maximum elevation along each traverse against the distance measured between the Taper Break and the hinterland break in slope (Fig. 3B; see Figs. ES4 to ES-8 [see footnote 1]). For over 1000 km, the length of a hypothetical two-dimensional thin plate (akin to a crustal beam) beginning at the base of the hinterland slope and ending at the Taper Break is inversely proportional to the elevation of the seaward-facing escarpment.
The closer the Taper Break is to the emergent part of the Norwegian margin, the steeper is the hinterland back-slope dip of both filtered and unfiltered digital elevation models (DEMs), whether or not the lower-crustal bodies are considered (Fig. 4). The hinterland back-slope dip is also directly linked to the combined width of the hinterland and the proximal margin. To test this independently, we located a point exactly halfway between the escarpment crest and the hinterland break in slope and collected raw (unsmoothed) DEM data inside a window having a radius equal to one third of the distance between these two markers. From the resultant cloud of points, we computed the mean dip and dip azimuth to the best-fit plane. For the lower-crustal body–excluded model, the recovered mean dips from Møre to Nordland exhibit a consistent relationship with the total beam length (Taper Break to hinterland break in slope; Fig. 4B). The same subset also performs well, albeit less so, when the lower-crustal bodies are included (Fig. 4C). Correlations are poorest in the North Sea and Senja fracture zone subsets.
CRUSTAL TAPER AND THE MORPHOLOGY OF SCANDINAVIA
We have previously shown that normal faults controlled the topography inboard of the sharply tapered Møre and Lofoten-Vesterålen segments of the Scandinavian margin until long after the Late Jurassic–Early Cretaceous rift phase that created the taper (Redfield et al., 2005; Redfield and Osmundsen, 2009; Osmundsen et al., 2009, 2010; Hendriks et al., 2010). We have also shown that the relationship between taper and topography holds for a global set of extended margins (Osmundsen and Redfield, 2011; Fig. ES-1 [see footnote 1]). Our new results indicate that both the height and back-slope dip of the Scandinavian Mountains are directly related to the sharpness of the Jurassic–Cretaceous thinning gradient. They also document the importance of the hinterland break in slope. We next turn our attention to further evidence for tectonic landforms in Scandinavia and their relationships to the Taper Break.
Under the shape factor definition we have adopted, the range 0.50–0.55 appears to represent a transitional geometry between oblate and elongate classes (Fig. ES-9 [see footnote 1]). The hinterland catchments east of Møre are characterized by elongated basins with shape factors less than 0.50 (Fig. 5A). By contrast, shape factors of the lower-elevation hinterland drainage basins east of Trøndelag are greater than 0.50, signifying a more oblate form. The specific hinterland drainage basin where the Møre-Trøndelag Fault Complex tips out has a shape factor greater than 0.55. To the north of the Møre-Trøndelag Fault Complex tip, the escarpments rise again, and hinterland drainage basin shape factors drop below 0.50 (Fig. 5A). From Møre to Nordland, the total length of the crustal beam (Taper Break to hinterland break in slope; Fig. 1), the mean (smoothed) hinterland back-slope dip, and the hinterland catchment shape factor are systematically linked (Figs. 5B and 5C). The relationships hold true for both of our end-member models, although, once again, excluding the lower-crustal bodies from the crustal density package provides the better fit. The along-strike changes in hinterland drainage basin shape factor are consistent with the displacement gradient of a large normal fault system, or even a megarelay, linking the downdropped hanging wall of the Møre-Trøndelag Fault Complex with the back-tilted hinterland of the Nordland (Fig. 5; Redfield et al., 2005).
We have also recovered information from the hinterland drainage incisions. Figure 6 shows histograms for stream segment trunk azimuths for all DEM-derived drainage systems of stream order 5 or greater from Scandinavia’s three topographic zones (Fig. 2). While those in the cratonic zone show considerable variation, and the escarpment zone exhibits a more bimodal distribution, the trunk azimuths from the Møre-to-Nordland portion of the hinterland display a striking degree of internal order. The histogram for the Møre-to-Nordland portion is bell shaped, nearly symmetric, and evinces a single peak at ∼150°–160°. This azimuth is roughly perpendicular to both the generalized continent-ocean boundary and Taper Break between the Møre and Nordland regions. It is also virtually identical to the mean orientation of the maximum cluster of slip lines recorded from faults from the proximal margin along the base of the Great Escarpment (Fig. 7; structural data from Redfield and Osmundsen, 2009; Osmundsen et al., 2010; plus new observations), and is perpendicular to the trend of many of the major continent-ocean boundary-parallel “thinning” faults (Lavier and Manatschal, 2006) that created the crustal taper in the Møre area (Osmundsen and Ebbing, 2008).
Although stream azimuth incisions and slickenlines form by completely different geological processes, their positive correlation can be readily understood through well-established principles of tectonic topography. Fault slip induces hinterland surface tilt, thus governing the drainage slope and its consequent pattern of incision (Leeder and Jackson, 1993). Repeated episodically, over and over, the process will generate conformity between two otherwise unrelated sets of vector data. A striking level of “teleconnection” (see Hoth and Kukowski, 2011) is evident between Scandinavia’s hinterland and proximal margin domains.
From DEM data, Etzelmüller et al. (2007) defined a set of quantitative landscape categories (Fig. 8). In mid-Norway, they identified little to no deep Alpine excision. Rather, their Alpine categories occurred principally in southern or northern Norway, where the Taper Break lies close to shore. This is especially true within the domain of the Møre-Trøndelag Fault Complex (Fig. 8, inset). To analyze this quantitatively, we constructed eight Møre-Trøndelag Fault Complex–perpendicular topographic profiles and superimposed over each the landscape classes of Etzelmüller et al. (2007). At the southeastern border of the Møre-Trøndelag Fault Complex (profile 1), the Tjellefonna fault zone defines the base of the topographic escarpment (Redfield and Osmundsen, 2009). The topographic rise from sea level is nearly 1800 meters, apatite fission track (AFT) apparent age offsets are significant at the 2σ confidence interval, the Taper Break lies close to shore, and the high Alpine (magenta) landscape class is in fault contact with the glacial scour (light blue) class. At profile 4 (Axel’s fault, the northernmost exposed fault outcrop of the Tjellefonna fault zone), the high Alpine and glacial scour classes are no longer directly juxtaposed. By profile 8 the escarpment is less sharp and much lower, 2σ AFT apparent age offsets are not resolvable, and the Taper Break is much more distal. Here, still far from the Møre-Trøndelag Fault Complex tip, but nevertheless well along its displacement gradient, both sides of the escarpment classify as moderate relief (Etzelmüller et al., 2007; Osmundsen et al., 2010). Present-day landscape classification, sea-level AFT apparent age displacement, and the crustal thinning gradient are spatially and quantitatively linked.
In suggesting that Scandinavia’s present-day landscapes are dominated by tectonic processes, we are hinting that many fault systems may have been reactivated well into the Cenozoic, or even today. Satellite interferometry data covering the Lyngen Alps suggest strain is accumulating (Osmundsen et al., 2009), but a confirmation of this interpretation remains pending. However, seismicity in Norway is well documented (Fig. 9; Bungum et al., 1979; Ambraseys, 1985; Hicks et al., 2000; Tvedt et al., 2002; see supplemental material [see footnote 1]). Events exceeding Mw = 5.0 have been recorded between southern Norway and points north of Lofoten-Vesterålen, and paleoseismic data document large-magnitude prehistoric earthquakes, the effects of which may even have helped shape Nordic mythology (Mörner, 2007).
As discussed in the supplemental material (see footnote 1), the available seismic catalogues differ greatly from each other in content. At present, we have no simple explanation why this is so, and we have therefore opted to present, analyze, and discuss data from two archives: the International Seismological Center (ISC) catalogue and the University of Helsinki (HEL) catalogue.
A pronounced offshore belt of seismic activity runs along the outermost edge of the proximal margin, just inboard of and roughly parallel to the no-lower-crustal-body Taper Break (see Byrkjaland et al., 2000; belt labeled a in Figs. 9B and 9C). Its focal plane solutions tend toward reverse faults (Hicks et al., 2000). Many hypocenters are estimated to be deeper than 15 km (Hicks et al., 2000), placing them in the crystalline crust or the underlying lower-crustal bodies. A similar nearshore to onshore belt of seismicity runs along the inner proximal margin edge, roughly parallel to and seaward of the topographic escarpments and just outboard of the 39 km crustal thickness line defining the thickness of “normal” unstretched crust (labeled b in Figs. 9B and 9C, see ES-2 [see footnote 1]). These hypocenters tend to be shallower than 12 km, commonly exhibit oblique-normal slip focal plane solutions, and are spatially more diffuse where the escarpments are lower in the Trøndelag region (Fig. 9A; see Hicks et al., 2000). The belts converge at both the northern and southern ends of Scandinavia, where the crustal taper is sharp and the proximal margin is relatively narrow. Few to no seismic events have been recorded on the intervening, gently tapering Trøndelag Platform. Thus, present-day seismicity very clearly defines the seaward and landward limits of the proximal margin. Norway’s earthquakes are systematically located with respect to Scandinavia’s Late Jurassic or Early Cretaceous thinning gradient.
The hinterland is also seismically quiescent between the escarpment and the hinterland break in slope. However, a third seismic belt (labeled c in Figs. 9B and 9C) marks the transition from unextended but upwarped lithosphere to unextended, unwarped lithosphere. Inboard of the hinterland break in slope, the lithosphere tends to be very old, very thick, and very rigid (Poudjom Djomani et al., 1999; Kozlovskaya et al., 2004; Pérez-Gussinyé and Watts, 2005; Artemieva, 2007). Although many of the earthquakes that constitute this third belt are quite small (Mw ≤ 2.0), they do not record man-made events: for example, Böðvarsson et al. (2006) explicitly removed all events suspected to be from anthropogenic sources. A significant amount of seismic energy may also be released per annum in the center of southern Finland (Fig. 9B). Most of these events are also small to very small in magnitude, and many probably do have the potential to be anthropogenic. Present in the ISC catalogue, these earthquakes are largely absent in the HEL data set, probably the result of filtering for explosions and other non-earthquake sources.
Møre-Trøndelag Fault Complex Topographic Displacement Gradient
Although made throughout much of Scandinavia, our observations can be well summarized within the geographic confines of mid-Norway (Fig. 10; Fig. ES-10 [see footnote 1]). Stacked topographic profiles drawn parallel to the Møre-Trøndelag Fault Complex from within its hanging wall and footwall sectors show that the topography of the southern Scandinavian Mountains is consistent with the scissors effect seen in large Basin and Range Province normal fault systems (e.g., Densmore et al., 2004; see Redfield et al., 2005). Direct relationships among (1) topography parallel to the Møre-Trøndelag Fault Complex, (2) DEM-determined landscape type, (3) hinterland drainage azimuths and catchment shape factors, (4) hanging-wall–footwall AFT apparent age pairs, and (5) fault slickenlines can be resolved both with each other and with respect to the Jurassic–Cretaceous crustal thinning gradient, or apparent taper length (Fig. 10).
In the Langfjorden region of the Møre-Trøndelag Fault Complex, the Taper Break lies close to shore, the escarpment is elevated and sharply defined, and the Tjellefonna fault zone is known to exist as a reactivated normal fault system (Redfield and Osmundsen, 2009; Nasuti et al., 2009; Redfield et al., 2011). In its hinterland, average back slopes are steep, and their drainage catchment shapes are correspondingly elongate. Sea-level 2σ AFT apparent age offsets are documented (Redfield et al., 2005; Redfield and Osmundsen, 2009, and citations therein), with younger ages in the upthrown footwall. The Tjellefonna fault zone footwall landscapes are dramatically Alpine, contrasting with the glacial scour and moderate relief of the less-incised hanging-wall landscapes (Fig. 8; see Osmundsen et al., 2010). To the northeast, the topography decays along strike as the Tjellefonna fault zone tips out, and the landscape contrasts across the fault system diminish. Footwall AFT samples age progressively away from the maximum displacement, becoming statistically conformable with the hanging-wall apparent ages as the tip is approached. The Taper Break becomes progressively more distal, and the apparent taper length becomes much longer. The Møre-Trøndelag Fault Complex is thus best interpreted as an ancient fault system (L.M. Watts, 2001) that was regionally reactivated in a normal mode during or at some point after the Jurassic–Cretaceous thinning phase of deformation, and that remained sufficiently active thereafter such that Cretaceous to latest Cenozoic–Holocene geological and geomorphic processes were controlled by its tectonic template.
Our observations indicate that normal faulting following the main phase of crustal thinning was, and remains, an integral process in the shaping of the emergent sectors of the Scandinavian proximal margin and its hinterland. While evidence from the type-section Iberian margin indicates that during the main rift phase faulting propagated sequentially seaward in the distal margin (e.g., Péron-Pinvidic et al., 2008; Ranero and Peréz-Gussinyé, 2010; Sutra and Manatschal, 2012), our data suggest that normal faulting also propagates landward within the proximal margin. Although it has been called such, this is not really “out-of-sequence.” Rather, we suggest it is the natural and isostatic consequence of the evolving accommodation of the proximal margin to continually changing, long-term loads imposed by erosion and deposition. Next, we present some important implications for the development of Scandinavia, and perhaps other margins as well.
1. Age of the Hinterland Landscape
The ages of formation of many individual landscape elements and the timing of reactivation of coast-parallel faults such as the Møre-Trøndelag Fault Complex remain unresolved. However, higher-order streams tend to be less affected by transient tectonism than lower-order tributaries (Merritts and Vincent, 1989; Burbank and Anderson, 2001). It seems likely that channelization of the present-day hinterland drainage trunks into conformity with the direction of brittle faulting and continental breakup reflects a relatively long-term process. The linear, fault-bounded belts of alpine topography that adorn the high-elevation escarpments also bespeak of tectonic drainage patterns that predate the Pliocene–Pleistocene onset of glaciation (Osmundsen et al., 2010). Slickenlines are found in several generations of cataclasite and on mineralizations of epidote, laumontite, and calcite, commonly in the context of fault rocks that indicate multiple phases of reactivation (Osmundsen et al., 2010). Offshore, Mesozoic to mid-Cenozoic strata are upwarped and truncated toward the coast, documenting pre-Pliocene uplift of the margin (e.g., Riis, 1996). In the Danish basin, an Oligocene influx of sediments derived from mainland Norway indicates uplift of the Scandinavian margin at this time (Rasmussen, 2009). Sømme et al. (2009) suggested that coarse clastic deposits of the Maastrichtian–Danian Ormen Lange fan were sourced from an elevated catchment of some 3 km relief. These data imply that normal fault–controlled uplift of the proximal margin is not solely a Neogene phenomenon.
On the other hand, the stark present-day patterns and contrasts in Scandinavia’s landscapes (this work; see detailed discussions in Osmundsen et al., 2009, 2010) suggest that a vertically significant topographic rejuvenation occurred relatively recently. Corroboration can again be found in the offshore sedimentary record. Laterally extensive, coarse clastic, prograding deltaic deposits of the Late Tertiary Molo Formation (Rokoengen et al., 1995, Eidvin et al., 2007) probably represent preglacial erosion. Similarly, the Pliocene–Pleistocene Naust Formation deposits document a basinward transfer of material, probably involving significant erosion of the nearshore mainland as well as sedimentary units from the inner shelf (Dowdeswell et al., 2010). These offshore deposits represent enormous isostatic loads, a mass transfer almost certainly capable of engendering a change in flexure.
The classical “Davisian” concept of landscape evolution from youth through maturity to old age was built on a single impulsive event occurring at the beginning of a geomorphic cycle (Davis, 1899). However, one single uplift event cannot have created Scandinavia’s stepped escarpment topography. Like many other workers (e.g., Riis, 1996; Lidmar-Bergström et al., 2000, 2007), we consider the evidence to be more consistent with a series of similarly oriented uplifts occurring episodically throughout the Cenozoic, each followed by a period of stability during which erosional surfaces were incised to a relatively low base level. Under this modified Davisian perspective, we envision multiple geomorphic cycles, beginning perhaps as early as “sag” basin time (i.e., Early Cretaceous onward) and persisting throughout much—or perhaps even all—of the Cenozoic. The progressive and episodic tilting of the hinterland caused slope-dependent landforms to become rejuvenated, reinforcing or overprinting the preceding landscape until a new equilibrium became established. Scandinavia’s hinterland thus provides an extraordinary archive of information describing the tectonic evolution of the adjacent, proximal margin.
2. Seismicity and the Structure of Scandinavia’s Passive Margin
The source of the stress field behind Fennoscandian seismicity has been the subject of much discussion (e.g., Gudmundsson, 1999; Lindholm et al., 2000; Bungum et al., 2010). Whilst Bungum et al. (2010) concluded that the sources of Scandinavia’s seismicity are multiple and diverse in nature, they considered the dominant forces to be related to plate tectonics and/or to lateral variations in lithospheric structure. However, stresses stemming from postglacial rebound are considered by many to produce passive-margin and intracontinental earthquakes (e.g., Adams, 1989; Wu and Hasagawa, 1996; Wu et al., 1999). Using a simple thin plate model, Gudmundsson (1999) computed surface tensile stress of up to ∼30 MPa near the center of Scandinavia’s postglacial uplift and compressive shear stress of up to ∼10 MPa at the margins (see Figs. 9B and 9C). These values are sufficient to initiate or reactivate fractures in crystalline rocks (Chase and Wallace, 1988; Gudmundsson, 1999; A.B. Watts, 2001), leading Gudmundsson (1999) to suggest that many, if not most, of Scandinavia’s larger-magnitude events are well explained by postglacial doming.
Seismicity can be depicted by maps showing the total energy released by earthquakes per unit area (Figs. 9B and 9C; see supplemental material for details of mapping procedure [see footnote 1]). These maps indicate that Fennoscandia’s annual energy release is not radially symmetric, but rather somewhat concentrated near the three passive-margin benchmark locations (Taper Break, escarpment/innermost limit of extension, and hinterland break in slope). Given the relationship of crustal taper to tectonic geomorphology described herein, these spatial correlations are probably not coincidental.
The reverse earthquakes associated with the mechanically weak (Péron-Pinvidic et al., 2008; Mohn et al., 2010; Lundin and Doré, 2011) Taper Break are spatially associated with the zone of compressive stress identified by Gudmundsson (1999). The location and tensile nature of the escarpment zone earthquake (see Hicks et al., 2000; Bungum et al., 2010) also fit Gudmundsson’s (1999) doming model. However, all the catalogues we have tested show prominent seismic gaps at the Trøndelag and Horda Platforms (TP, HP; Fig. 9) and the hinterland east of the escarpment. The systematically located seismic gaps and the asymmetric distribution of seismic energy within the dome of post-glacial uplift might appear puzzling, but in fact may provide an important clue.
Although the cause of the stress field that generates seismicity probably has little to do with hyperextension per se, many earthquakes may be related to faults, whose existence can be traced to Scandinavia’s large-magnitude extension. Stress concentration can occur where two mechanically contrasting materials are in contact (Jaeger et al., 2007). Landward from the belt of earthquake epicenters that mark the Taper Break, the crustal wedge begins to thicken, and the large fault arrays that exist tend to sole out at midcrustal levels (Fig. 1). Toward the sea, the crystalline continental crust is hyperextended, pervasively faulted, and structurally much less competent. In addition, faulting and serpentinization may have affected the uppermost parts of the distal margin’s lithospheric mantle. These contrasting structural conditions probably generate contrasting stiffness: For a given stress, more strain can be accommodated in the distal margin than in the less-faulted proximal margin. Thus, whatever the mechanism responsible for the generation of stress, the necking zone (Fig. 1) may constitute a natural location for earthquakes to concentrate. Similarly, inboard of the Taper Break on the gently thinned Trøndelag Platform (TP, Fig. 9), faulting is not penetrative. There, contrasting structural conditions do not exist, and proximal margin seismicity is negligible.
In addition to the three seismic energy belts described previously herein, a fourth belt follows the thinned crust of the North Sea–Viking graben rift systems outboard of the thicker Horda Platform (HP; belt labeled d in Figs. 9B and 9C). Because they occur within very thin, densely faulted crust that is closest both to the spreading axis and to the relatively high-density contrast of the continent-ocean boundary, a temptation exists to ascribe the earthquakes of the Taper Break and the North Sea–Viking graben rifts to a horizontal far-field stress such as ridge push or gravitational potential energy. However, along the southwestern trace of the Taper Break off Scotland and Ireland, seismicity becomes increasingly sparse (Fig. 11C). Although in places this distal margin crust has been hyperextended as much as in the Møre basin (see Lundin and Doré, 2011), insufficient stress is being applied today to either reactivate thinning-phase faults or to generate new ones. Yet, important plate-tectonic conditions such as the age/density of the oceanic crust at the continent-ocean boundary and the seafloor spreading azimuth at the ridge are essentially the same offshore Scotland and Ireland as in Scandinavia.
Nevertheless, the seismicity of the North Sea–Viking graben rift indicates that its faults can be reactivated by today’s stress field. Rather than ridge push or another far-field stress, the process that appears most likely to differentiate the seismic response to hyperextension along the Anglican and Scandinavian sectors of the eastern North Atlantic Taper Break is the land surface vertical motion due to glacial unloading (Gudmundsson, 1999). As a first-order test of this hypothesis, we briefly consider the region offshore Labrador, Canada (Fig. 11A). Because the land surface is undergoing rapid uplift, Labrador has been a focus of studies concerning postglacial seismicity (e.g., Adams, 1989; Wu and Hasagawa, 1996). Offshore, the Taper Break can be identified in one location (see Osmundsen and Redfield, 2011) using modern seismic-reflection data (Reston, 2009; Chian et al., 1995). Hyperextended continental crust may be present offshore, as well as a belt of exhumed mantle serpentinite (Lundin and Doré, 2011). Although we cannot constrain the Taper Break elsewhere on the Labrador margin, the rule of taper and topography (Fig. 3A; Fig. ES-1 [see footnote 1]) appears to hold: escarpment elevations are generally higher in the north of Labrador than in the south. A belt of earthquakes tracks the Taper Break on the Labrador side for some distance, but becomes diffuse or absent to the south as the width of the proximal margin increases (Reston, 2009; Lundin and Doré, 2011). Seismicity may thus also mark the Taper Break offshore of Labrador in the region associated with postglacial vertical deformation (see land surface mapping by Andrews and Tyler, 1977).
The concentration of Scandinavian seismicity within the region affected by recent deglaciation as defined by present-day land uplift is impressive (Fig. 11C). However, it presents a fundamental question. Our observations suggest that significant post–thinning phase faulting and consequent escarpment topography (and therefore seismicity) affected the Norwegian proximal margin long before the onset of the Pliocene–Pleistocene ice world (see also Osmundsen et al., 2010). If ridge push is relatively unimportant for Taper Break or escarpment seismicity and should no postglacial rebound be ongoing, how can a passive continental margin such as (for example) SE Brasil be seismic? Also, in the absence of seismicity, how can footwall uplift proceed? One plausible solution may lie in the simple, yet inevitable, geological processes that accompany the transition from active extension through breakup and beyond.
3. Taper, Tilt, and Flexure
While teleconnections between the proximal margin and the hinterland are perhaps most prominent in the Møre sector (Fig. 10), more than 1000 km of taper relationships (Figs. 3–5) illustrate that vast areas of Norway and Sweden are tectonically linked. From northern Scandinavia, Lidmar-Bergström et al. (2007) described a diverging set of paleosurfaces that fan to the west, a geometry hinting strongly toward episodic uplift guided by lithospheric-scale flexure. We suggest that flexure is controlled by the Taper Break. Mechanically weak, it marks the onset of minimal to negligible coupling between the proximal margin and the post-breakup oceanic lithosphere. Similarly, the termination of bending near the Gulf of Bothnia is coincident with a sharp increase in lithospheric flexural rigidity (see Pérez-Gussinyé and Watts).
At extended margin escarpments, lithospheric flexure can exhibit either upwarp or downwarp (Campanile et al., 2008). We suggest that the style of synrift warp is controlled by the sharpness of taper. For example, extreme excision over a short horizontal distance that is accompanied by mantle exhumation or near-exhumation may impose a structural template akin to a very nearly broken thin plate. Erosion of what may be a 4–5-km-high synrift escarpment (Osmundsen and Redfield, 2011; Fig. ES-1 [see footnote 1]) will generate an enormous mass deficit at the thin plate edge, a condition amenable to flexural upwarp. Alternatively, more gradual crustal thinning along an array of detachment faults will distribute isostatic unloading over a much wider zone. A likely result might be a series of lower, less dramatic escarpments that step inland toward the hinterland and a lithosphere that is perhaps better modeled as a weakened, but continuous thin plate. Unlike a sharply tapered sector, the locus of erosion and the “sag” basin depocenter may be widely separated on gently tapered sectors. Under this load structure, lithospheric downwarp may be the more common flexural response.
Footwall uplift during extension is well-established (e.g., Weissel and Karner, 1989; Kusznir et al., 1991). Topographic reduction by erosion may commence almost immediately, and the erosion of the rift flanks and deposition in the subsiding hanging wall and subsequent “postrift” basins can induce further flank uplift (Weissel and Karner, 1989; Masek et al., 1994; Small and Anderson, 1995). Following extension, excision-related proximal margin uplift must give way to post-thinning lithospheric cooling and subsidence (McKenzie, 1978; Weissel and Karner, 1989). The episodic rejuvenation of kilometer-scale topography at the innermost proximal margin following such a reduction thus appears isostatically challenging. However, one consequence of cooling will be an increase in the effective flexural rigidity of the proximal margin lithosphere (Stephenson, 1984; Chase and Wallace, 1988). Continuous transfer of mass from escarpment to basin atop a sufficiently rigid thin plate pinned at its cratonic end will erode the escarpment and eventually generate a significant lithospheric-scale bend (A.B. Watts, 2001; Watts et al., 2009). In this way, the reduction of rift-era escarpments throughout much of the “sag” period may be accompanied by an ever-increasing tensile bending stress at the proximal margin (A.B. Watts, 2001; Watts et al., 2009).
Tensile stress in the uppermost crust can eventually break rock (A.B. Watts, 2001; Chase and Wallace, 1988; Gudmundsson, 1999). In the case of a downwarped gentle taper, near-surface deformation may be limited to tension fractures, although normal faults may form at greater depths (Gudmundsson, 1999) At sharply tapered upwarps, high topography may combine with unbuttressed, seaward-directed gravitational potential energy (Pascal and Cloetingh, 2009) to initiate new, shallow normal faults or reactivate old ones. The sharp contrast in earthquake epicenter density between the seismic escarpment zone and the aseismic, well-buttressed hinterland back slope may reflect this process. The time span between the end of extension to the point when the “sag” basin is sufficiently full such that brittle normal faulting becomes an important process on the proximal margin must depend on the sedimentation rate, but in terms of geological time it may not be too dissimilar worldwide—a few tens of millions of years may suffice. By their very architectural nature, continental margins adjacent to hyperextended ocean basins may be doomed to reactivate.
The ratio of fragmentation to reactivation (the generation of new fractures compared to the reactivation of old ones) tends to decrease with time as crystalline bedrock becomes saturated, such that any new stress orientation can be geometrically accommodated by preexisting structures (Munier and Talbot, 1993). Similarly, because broken rock is weaker than unbroken rock, the instigation of slip will require less bending as the margin sector evolves with time. Thus, escarpment sectors that are controlled by normal faults or heavy fracturing may tend to localize both in space and orientation. Whereas the synrift escarpment and the innermost limit of lower-crustal extension may initially be widely separated, post-thinning inboard migration of faulting will encourage the escarpment crest to migrate landward, stalling only where structural saturation occurs. The near-Gaussian symmetry of the hinterland stream azimuth histogram (Fig. 6C) reflects the geometric consistency of Cenozoic escarpment zone fault reactivation, suggesting that effective saturation occurred long ago. Measuring an actual and time-transgressive process, the apparent taper length—our proxy for the width of the proximal margin—may in fact have a physical meaning.
4. Some Global Implications
Our results point strongly toward a mechanically weak distal margin, and they support the hypothesis that post-rift cooling does not increase the strength of the lithosphere where it has undergone significant mechanical damage (see previous work by Péron-Pinvidic et al., 2008; Mohn et al., 2010; Lundin and Doré, 2011). In turn, this suggests that horizontal far-field forces can be readily accommodated by deformation in the distal margin and the outermost proximal margin, and are thus not responsible for onshore proximal margin deformation. The patterns of North Atlantic seismicity described above support this interpretation.
Loads generated by escarpment erosion, offshore sedimentary deposition, and postglacial rebound have been superimposed in Scandinavia throughout the Neogene period’s multiple glacial cycles. Their vertically oriented stresses are mutually reinforced during periods of deglaciation. However, compared to that of the postglacial dome, the zone of maximum uplift generated by escarpment erosion will be longer, linear, and located along the emergent proximal margin. The zone of offshore maximum deposition will be similarly oriented. This observation may help to partly explain the asymmetric expenditure of Fennoscandia’s annual seismic energy budget. It may also explain the conundrum posed earlier in this discussion: If the vertical load generated by erosion and deposition is sufficiently great, fault reactivation and consequent seismicity can occur at any passive margin sector. In Scandinavia, episodic footwall uplift throughout much of the Cretaceous and Cenozoic may have been driven by just such a mechanism.
SE Brasil may offer a glimpse of the way in which Norway’s margin might manifest itself seismically in the absence of postglacial rebound. Earthquake data from the SISBRA catalogue (ftp://ftp.iag.usp.br/deptos/geofisica/seismic/) and Assumpção et al. (1997) suggest two seismic belts may exist (Fig. 12). One, offshore, follows the thinned crust of the ultradeep, hyperextended (Zalán et al., 2011) Campos and Santos Basins. Onshore, earthquakes occur more commonly in the elevated highlands of the escarpments, and they track especially the long, linear ranges such as the Serra de Mantiqueira and Serra do Espinhaço. Seismicity is more rare in the coastal lowlands, and largely absent in the Brasilian hinterland. Although never glaciated since the time of hyperextension and characterized by significantly fewer earthquakes in toto, SE Brasil’s pattern of seismicity closely mimics Scandinavia.
Scandinavia’s present-day escarpment elevations correlate clearly and directly with the length of a two-dimensional thin plate, or crustal beam, defined at one end by the hinterland break in slope and at the other by a crustal thickness of ∼10 km that was tectonically shaped during Jurassic–Cretaceous time. In the hinterland, slope-dependent geomorphic elements such as dip and catchment shape factors also correlate positively. These data, plus today’s seismic record, indicate that uplift and landscape-forming tectonic activity are recent and even ongoing (see also Osmundsen et al., 2009). Yet the similarity in azimuth between higher-order back-slope drainages and pre-Quaternary brittle fault slickenlines suggests that the teleconnection between the hinterland back slope and the proximal margin has roots in a deeper past. Tectonically generated topography is a reality in Scandinavia, and it appears to have been so for many tens of millions of years.
Using these data, we have redefined the Scandinavian passive margin from the perspective of the new models of extended margins. Fundamental to our redefinition are the hinterland break in slope and the Taper Break. The hinterland break in slope marks the landward limit of the passive margin. Although its crust is of “normal” thickness, the inclusion of the hinterland follows logically: It has been vertically deformed as a result of a lithospheric-scale flexure that follows a marginwide series of topographically significant footwall uplifts. The Taper Break marks the seaward limit of this flexure, and it delimits the boundary between the proximal margin and distal margin.
Both the landward and seaward edges of Scandinavia’s proximal margin are seismically active. However, seismicity appears to be controlled more by local vertical deformation or loads stemming from local topography than by horizontal far-field forces. A belt of reverse seismicity located within attenuated crystalline continental crust or the underlying mantle delimits the seaward edge of the Scandinavian proximal margin. Located near the junction between penetratively faulted, hyperextended crystalline crust and crystalline crust of considerably greater structural integrity, it may mark a point of maximum flexural downwarp of thinned and weakened lithosphere (Watts et al., 2009), a zone of coupling-decoupling that formed during high-beta thinning (see Sutra and Manatschal, 2012), or both.
The processes that shape the Scandinavian passive margin are intimately linked to a crustal architecture largely inherited from the thinning phase of Lavier and Manatschal (2006). In keeping with the previously proposed terminology for hyperextended passive margins, we suggest that an accommodation phase follows the end of active thinning and/or exhumation. Thinning-phase and accommodation-phase escarpments adhere to similar, perhaps even identical scaling laws. During accommodation, cooling of the landward proximal margin and hinterland conspires with increasing flexural rigidity to permit widespread topographic reduction onshore. In turn, the eroded debris contributes to the filling the offshore accommodation space. We suggest that escarpment rejuvenation is a function of an ever-increasing flexural arch that accompanies this redistribution of mass. At some point, tensile bending stresses must exceed the strength of the rock. Tension fractures in the uppermost crust may open at gently tapered downwarps, while renewed “out-of-sequence” normal faulting and episodes of footwall uplift might mark sharply tapered upwarps. As an example, crustal rupture is replaced by crustal flexure as the Møre-Trøndelag Fault Complex megarelay is approached, a process closely reflected in the landscape. Ultimately, the reactivation of faults and fractures will localize, stabilizing the escarpment over time. In this way, the difference between the apparent taper length and the actual width of the proximal margin may reflect the degree of maturity of a physical process.
In general terms, sharply tapered margin sectors may impose long-term controls on both proximal margin sediment source areas and on the fault-controlled routing systems that deliver those sediments to the depocenters (Osmundsen and Redfield, 2011). Similarly, the formation and reactivation of onshore normal faults can impregnate the inner proximal margin with penetrative zones of structural weakness. In Norway, several very large-magnitude rockslides and rock-slope instabilities can be directly attributed to tectonic fabrics that in turn are associated with zones of sharp taper (Osmundsen et al., 2009; Redfield and Osmundsen, 2009). Lastly, the now well-documented taper relationship hints strongly toward a distal margin in which lithosphere is mechanically compromised and correspondingly very weak (see Masson et al., 1994; Péron-Pinvidic et al., 2008; Lundin and Doré, 2011). As suggested by previous workers (see Faccenna et al., 1999; Péron-Pinvidic et al., 2008; Mohn et al., 2010), this weakened zone may potentially localize the onset of subduction when plate convergence begins. Thus, the onset of the accommodation phase may represent the beginning of the end of a Wilson cycle.
As products of the sequential re-imposition of tilt-block topography throughout the accommodation phase, Scandinavia’s landscapes are both old and young. Tectonically linked to the development of the proximal margin, the hinterland back slope is an important archive of information. Although perhaps best expressed on Scandinavia’s recently de-glaciated margin, similar teleconnections may exist on other of the world’s passive margins. We suggest that our hypothesis applies to more than just the Scandinavian passive margin, and perhaps to all extended margins in general. A testable, process-based understanding of landscape evolution of not-so-passive margins can provide a powerful tool whose implications for many aspects of geology may be profound.
C. Talbot and A. Gudmundsson provided excellent and thought-provoking reviews that greatly improved our manuscript. We gratefully acknowledge many informative conversations and collaborations with S. Buiter, J. Ebbing, G. Péron-Pinvidic, S. Gradmann, E. Lundin, R. England, R. Schmitt, N. Stanton, the Norwegian Geological Survey natural hazards team members, and plenty other fine folk. J. Ebbing, F. Reynisson, and M. Peréz-Gussinyé kindly provided their original gridded data and much positive discussion.