Basin inversion is a fundamental process in the tectonic evolution of sedimentary basins. It may play an important role in the formation of hydrocarbon traps, although it can also have negative effects on the prospectivity of a basin (Turner and Williams, 2004). Many authors have reviewed the concept of basin inversion and the criteria for recognizing the resulting structures (see, e.g. Cooper et al., 1989; Coward et al., 1991; Eisenstadt and Withjack, 1995; Coward, 1996; Kockel, 2003). Mild inversion refers to a small magnitude of shortening relative to the magnitude of earlier extension (Cooper et al., 1989). In this study, we define basin inversion as the shortening of previously extensional basins due to compression or transpression (see Cooper et al., 1989; Turner and Williams, 2004). From here, our use of the term “inversion” refers to basin inversion, unless specifically stated otherwise. Inversion is recognizable from the structures that accommodate shortening and uplift of the basin packages: reverse reactivation of pre-existing normal faults or fault trends, i.e. fault inversion, and the development of shortcut thrusts and folds (e.g. Williams et al., 1989; Turner and Williams, 2004). In recent years, the increased prevalence of borehole-constrained 3D seismic data of high quality has improved our understanding of inversion structures and their development, especially in basins that are only mildly inverted (e.g. Jackson and Larsen, 2008; Grimaldi and Dorobek, 2011; Jackson et al., 2013). Mild inversion is advantageous to such studies because pre-existing structures are still well resolved, making their role during inversion easier to delineate (Jackson et al., 2013).

Several physical model studies have illustrated the structural styles and development of inversion structures (e.g. Koopman et al., 1987; McClay, 1989, 1995; Eisenstadt and Withjack, 1995; Brun and Nalpas, 1996; Bonini et al., 2012; Jagger and McClay, 2018; Dooley and Hudec, 2020), as well as how initial three-dimensional geometries of normal faults affect the development of inversion structures (e.g. Yamada and McClay, 2004). Depending on the reactivation of basement faults below the basin fill, the effects of inversion may vary a great deal between different parts of the extensional basin. Basement faults are more prone to reactivate in a reverse sense with gentle dip angles, oblique angles between their strike and the direction of shortening, low frictional resistances along fault planes, and high connectivity to fluids expelled from juxtaposed rocks (Bonini et al., 2012). Forced folding and short-cut structures tend to occur where the strikes of inherited extensional structures are approximately orthogonal to the direction of maximum compressive stresses, and where steeply dipping normal faults bound the basin (Letouzey, 1990). If instead the incidence is oblique, a strike-slip component is induced that allows for reactivation of steeply dipping faults (Letouzey, 1990; Letouzey et al., 1990).

Additionally, workers have utilized physical models to investigate the structures formed by extension (e.g. Withjack and Callaway, 2000; Ferrer et al., 2017) and inversion of rift basins containing weak evaporite sequences (e.g. Nalpas et al., 1995; Brun and Nalpas, 1996; Bonini et al., 2012; Roma et al., 2018a, b; Ferrer et al., 2017). These have shown that mobile evaporites can play a major role in the structural development of basins during both extension and inversion, due to their ability to decouple deformation (partially or fully) in the substrata and overburden. The model studies of Letouzey et al. (1995) showed that major controlling factors on the localization of reverse faulting and folding in inverted graben include pre-existing extensional structures, salt thickness, and the distribution of older evaporite structures. Following salt deposition, salt ridges tend to form when extension causes salt to flow up-dip on the hanging wall floor away from the main fault because of the deformation of the sub-salt hanging wall and the increased sediment load just adjacent to the main fault (Vendeville, 1987; Vendeville and Jackson, 1992; Nalpas and Brun, 1993). During subsequent shortening and inversion, folding and reverse faulting or thrusting will often initiate above salt ridges with favourable orientations, where the overburden is thinned and possibly faulted. This occurs even when there is no direct link to basement faults (Letouzey et al., 1995; see also Brun and Nalpas, 1996).

Along with salt diapirism, basin inversion is responsible for a significant number of structural traps for hydrocarbon reservoirs in Cretaceous Chalks in the Danish Central Graben (Megson, 1992; Vejbæk and Andersen, 2002). The deformational histories and geometries of these reservoir strata are closely related to basin inversion that occurred mainly in the Late Cretaceous, when they were buried only at shallow depths or were still being deposited (Vejbæk and Andersen, 1987; Cartwright, 1989; Vejbæk and Andersen, 2002). Duffy et al. (2013) analysed the controls of mobile evaporites on the structural development during Triassic and Jurassic rifting in a smaller part of the Danish Central Graben (hereafter DCG; Fig. 1). Their work highlights the DCG as a well-suited area for studying the structural controls of deep evaporites, as the thickness of the Zechstein units here range from near zero in some areas to thousands of metres in salt structures. In addition, the combination of a mild yet significant degree of inversion and the well-preserved extensional fabric provides an excellent opportunity to link the kinematics of inversion structures formed near the surface to deeper inherited structures and evaporites. This is the focus of the present study. We provide an updated kinematic analysis of a portion of the DCG (Fig. 1) in order to explain the role of mobile evaporites and salt tectonics during Late Cretaceous basin inversion, and the relationship between deformation in the sub-salt basement and supra-salt overburden. So far, the studies dedicated to inversion in the DCG focused on spatial variations in timing of inversion (Vejbæk and Andersen, 1987, 2002; Cartwright, 1989) – we instead focus on the kinematic connections between the upper inversion structures, the Zechstein units and pre-Zechstein basement below. This revision is due as a wealth of general and conceptual knowledge has since then surfaced on extensional and compressional deformation of basins containing mobile evaporites (e.g. Vendeville and Jackson, 1992; Letouzey et al., 1995; Stewart and Clark, 1999; Withjack and Callaway, 2000; Bonini et al., 2012; Stewart, 2014). Our analysis incorporates these findings with our own results from seismic mapping and structural interpretation of a high-quality 3D seismic data set. The main objective of the present study is thus to relate and discuss our observations to published results about the following topics.

The controls of evaporites on inversion structures in the study area and mildly inverted basins in general.

The North Sea area has accommodated sediments since Cambrian times while experiencing compressional, extensional, and inversion tectonic regimes (Ziegler, 1990; Vejbæk, 1997; Coward et al., 2003; Lassen and Thybo, 2012). The present outline of the North Sea Basin relates to an early Permian thermal event and crustal extension (Vejbæk, 1997; Glennie et al., 2003). Early Permian rifting involved widespread volcanism, fault block rotation, and erosion of footwall crests (Stemmerik et al., 2000; Mogensen and Korstgård, 2003; Coward et al., 2003; Glennie et al., 2003; Clausen et al., 2016). Post-rift thermal subsidence (sensu McKenzie, 1978) followed and formed the extensive Northern- and Southern Permian Basins separated by an approximately E–W-striking basement high, which would later separate into the Mid North Sea and Ringkøbing–Fyn highs (Vejbæk, 1990, 1997; Sørensen 1986). In late Permian times, the evaporites of the prominent Zechstein Group were deposited in these basins (Fig. 2a). These units pinch out towards the basin boundaries where marginal evaporite facies dominate while large volumes of halite formed in the basin centres (Stemmerik et al., 2000; Glennie et al., 2003). Inherited regional topography largely controlled the extent of the evaporites although rifting generated fault-controlled subsidence in, e.g. the Central Graben and Horn Graben areas, which allowed thicker packages to accumulate (Clausen and Korstgård, 1993b; Korstgård et al., 1993). Due to later halokinesis, it is unknown whether faulting occurred during or prior to evaporite deposition (e.g. Duffy et al., 2003).

Triassic rifting in the central and southern North Sea area generated approximately N–S-striking faults as part of the development of the Central Graben and Horn Graben (Fig. 2b; Coward et al., 2003). The Mid North Sea and Ringkøbing–Fyn highs greatly influenced Triassic sedimentation, as evident from lithological differences in the two basins. Continental clastics with local and thin evaporite beds dominated to the south (Fig. 2b), while occasional marine transgressions from the Tethys introduced marine conditions and flooded the Mid North Sea High. The most northward extent of these marine transgressions occurred in the easternmost part of the North Sea Basin (Bertelsen, 1980; Michelsen and Clausen, 2002). The Triassic sediments in the Danish Central Graben show a close relationship to the area south of the Mid North Sea and Ringkøbing–Fyn highs (Michelsen and Clausen, 2002). At this time, halokinesis involving Zechstein units initiated, and differential subsidence due to halokinesis possibly controlled the distribution of fluvial sediments (Goldsmith et al., 2003; McKie et al., 2010; Jarsve et al., 2014).

The Jurassic was dominated by intense rifting, which highly influenced the Moray Firth, the Viking Graben, and the Central Graben (Fig. 2c). Thermal doming at the triple junction initiated in Early Jurassic times and erosion of older deposits generated the widespread and easily recognizable Mid Cimmerian Unconformity (Coward et al., 2003). Intense Late Jurassic rifting occurred over two main phases. The first reactivated N–S-striking normal faults during E–W-directed extension. In the Danish Central Graben, this generated mainly eastward-dipping half-grabens (Roberts et al., 1990). The second phase formed NNW–SSE-striking faults that connected the older faults. This combination caused a rather complex fault pattern in the Danish Central Graben. Thick syn-rift mudstone deposits of especially the Farsund Fm. provide good source rocks locally in the DCG (Michelsen et al., 2003).

Deposition of marine shales characterize the Early Cretaceous, while faulting gradually ceased in the central North Sea area (Fig. 2d; Coward et al., 2003) and post-rift thermal subsidence formed accommodation (Sclater and Christie, 1980). Extensive deposition of pelagic shelf carbonates followed in Late Cretaceous and Danian times, forming the Chalk Group, while siliciclastic units were restricted to the basin margins (Fig. 2e; Ziegler, 1990; Surlyk et al., 2003). Widespread Late Cretaceous and early Paloegene inversion tectonism in the Alpine foreland caused uplift and erosion of former depocentres, as well as re-deposition of sediments in the southern and central North Sea area (e.g. Ziegler, 1987; Kley, 2018). Inversion generated a significant number of structural traps for producing hydrocarbon reservoirs in, e.g. the Danish and Norwegian parts of the Central Graben (Damtoft et al., 1987). The post-Danian Paleogene was dominated by siliciclastic deposition from the emergent areas surrounding the North Sea basin, meaning that the deep basin above the Mesozoic rift system gradually filled (Gołedowski et al., 2012; Anell et al., 2012). Inversion activity in the North Sea Basin continued, although the activity migrated westwards relative to the areas affected in the Cretaceous and Paleogene (Ziegler, 1990; Kley, 2018).

Locally in the DCG, inversion movements may have initiated as early as in the late Hauterivian (Vejbæk, 1986). Although punctuated by pulses of increased activity, inversion was continuous in the DCG throughout the Late Cretaceous (Vejbæk and Andersen, 1987, 2002). Around the beginning of the Campanian, it became a major influence on the basin morphology, forming a prominent unconformity above earlier rift depocentres (van Buchem et al., 2018). The relative uplift caused by inversion occurred on a background of post-rift thermal subsidence and was small enough that erosion was limited in comparison to other inverted basins in the southern North Sea. Subsequent Paleogene inversion generated only a wide and gentle flexure above the rift (Fig. 1; Vejbæk and Andersen, 2002; Clausen and Korstgård, 1993a; Clausen et al., 2012).

A regional merge of reprocessed data from several 3D reflection seismic surveys with a total area of ca. 6000 km2 was available for this study. This covers most of the DCG area. We used a sub-crop of this dataset with an area of ca. 4000 km2, covering parts of the Heno Plateau, Tail End Graben and Salt Dome Province areas (Fig. 1). The dataset is Kirchhoff pre-stack depth-migrated, has a 12.5 m inline (N–S direction) and crossline (E–W direction) spacing, and extends to a depth of 10 km. The vertical sampling rate is 4 ms, and the seismic wavelet is zero phase with European polarity. Therefore, in the seismic sections shown herein, a negative amplitude (trough) corresponds to a downward increase in acoustic impedance, i.e. a hard response. The velocity model applied for depth imaging is verified with check-shot data and sonic data from wells.

Within the Chalk Group, the vertical seismic resolution is ca. 25 m, i.e. a quarter of the wavelength of the dominant frequency. The signal quality is generally excellent, although it deteriorates close to and below large lateral discontinuities, e.g. salt diapirs, large faults, and sub-vertical strata. We chose to interpret depth-converted seismic data in order to avoid structural misinterpretations of apparent deformation caused by velocity variations, e.g. pull-up below salt structures. Approximately 25 exploration wells within the study area have been included with formation tops constrained by biostratigraphy (Nielsen and Japsen, 1991).

Two major rift faults occur in the study area: the Coffee Soil Fault and the Gorm–Tyra Fault (Figs. 1, 6). The overall structural architecture of the pre-Cretaceous in the study area approximates that of a NW–SE-trending half-graben bounded by the Coffee Soil Fault (Fig. 6) (Gowers and Sæbøe, 1985; Cartwright, 1991). This major structure constitutes the master rift fault along the northeastern boundary of the Central Graben throughout the Danish North Sea sector. In our study area, it consists of three linked normal-fault segments with different strike directions (Fig. 5b): segment 1 in the north and segment 3 in the south strike approximately NNW–SSE, while segment 2 has an approximately WNW–ESE strike. The fault is generally well imaged on the seismic data and is even traceable below the rift basin floor in some places (Fig. 6). This reveals the planar shape of each segment along the Triassic–Jurassic units in the hanging wall, and relatively small dip angles (see also Cartwright, 1991) of segments 1 and 3 with dip angles of ca. 37 and ca. 32∘, respectively (Figs. 6, 7). This contrasts the ca. 50∘ dip angle of segment 2. The other major basement fault in the study area, the Gorm–Tyra fault is a west-dipping normal fault striking approximately N–S (Figs. 5b, 6). It extends southwards from the intersection of segments 1 and 2 of the Coffee Soil Fault and shows a consistent dip angle of ca. 40∘. Its throw at Top pre-Zechstein level is generally ca. 2 km along its length, and unlike the Coffee Soil Fault it does not extend up through the Mesozoic syn-rift succession. In the hanging wall of segment 1 of the Coffee Soil Fault (Fig. 8) and at the hanging wall of the Gorm–Tyra Fault (Fig. 6), the basin floor, i.e. Top pre-Zechstein, dips significantly towards the faults, up to ca. 20∘. The Top pre-Zechstein reaches its deepest points of ca. 9.5 km in the Tail End Graben and nearly 9 km at the confluence of segments 1 and 2 of the Coffee Soil Fault and the Gorm–Tyra Fault (Fig. 5b). This equates to fault throws of more than 6 km, although this is a minimum estimate as no pre- or syn-rift deposits are preserved in the footwall. To the east of the Gorm–Tyra Fault (i.e. in the area between the Coffee Soil Fault and the Gorm–Tyra fault), the basin floor is generally closer to horizontal (Fig. 9) and does not steepen toward the Coffee Soil Fault, except in the immediate hanging wall of segment 2 (Fig. 6). The Poul Plateau at the confluence of segments 2 and 3 and the Mads High in the west constitute the two shallowest basement elements in the study area apart from the Ringkøbing–Fyn High.

A large number of subsidiary normal faults offset the Top pre-Zechstein with a significantly higher density in the western part of the study area (Heno Plateau) compared to east of the Gorm–Tyra Fault (Figs. 1, 6, 7). Although they are often difficult to map with certainty due to limited data quality at depth, their strike trends are generally between E–W and N–S, i.e. equivalent to those of the Coffee Soil Fault segments and the Gorm–Tyra Fault. We have observed no reverse faults at Top pre-Zechstein level.

Lower Permian Rotliegend units underlie the Top pre-Zechstein in the western part of the study area, where Zechstein units are absent (Figs. 4a, 5c). No wells penetrate beyond the Triassic in the deeper parts of the DCG in the study area, meaning that the nature of the pre-Zechstein basement is unknown here. Parallel reflections on both sides of the Gorm–Tyra Fault suggest that Paleozoic deposits may extend across much of the study area below the Zechstein salt (marked “A” in Fig. 6a)

Zechstein units range in thickness from thin (< 50 m) or below seismic resolution in the central part of the study area, to extreme local maxima (> 3000 m) in diapiric structures in the southern (Salt Dome Province) and very northwestern parts (Figs. 5e–f, 9). Zechstein deposits are absent where the Top pre-Zechstein is the shallowest i.e. on the Heno Plateau to the west (cf. Vigeland Ridge of Gowers and Sæbøe, 1985) and on the Ringkøbing–Fyn High. Only a few significant faults offset the Top Zechstein surface within the larger basin (Figs. 5d, 6). Along the western pinch-out of Zechstein deposits on the half-graben dip-slopes, we have interpreted three reverse faults or thrusts that sole out into Zechstein units (Figs. 6, 8, 10). The two largest occur along the Arne Ridge and Bo–Jens Ridge, where they reach Cretaceous levels (Fig. 12b and marked “B” in Figs. 6, 8, 10). The third occurs in relation to folded overburden strata at the Edna structure (marked “C” in Fig. 6). Also detaching into the Zechstein, a prominent normal-fault trend links the Gorm, Skjold, and Dan structures (marked “A” in Fig. 5d). The northern part of the Gorm–Tyra Fault offsets the thin package of Zechstein found here (Fig. 6), while to the south (Fig. 9), an initial thicker Zechstein deposit “drapes” the Gorm–Tyra Fault, forming pillows at the footwall crest and up-dip in the hanging wall. Therefore, the Gorm–Tyra Fault is only expressed as a large monocline at shallower levels (Fig. 9).

The constructed mobile-salt pinch-out effectively marks the boundary of the Salt Dome Province in the south, while only a small area in the north contains potentially mobile Zechstein (Fig. 5f). In the Poul Basin area, only pockets of Zechstein reaching thicknesses > 200 m occur (Fig. 5e). These pockets fill topographical lows in the Top pre-Zechstein surface that seem to have formed from erosion rather than faulting (Duffy et al., 2013). The Salt Dome Province and the northern Tail End Graben area are characterized by large salt pillows, ridges, and rollers with extreme thicknesses of Zechstein (> 400 m; Fig. 5f). In the Salt Dome Province, the areas of thin or absent Zechstein between these structures show typical signs of salt deflation and evacuation, which led to the formation of minibasins, e.g. in the hanging wall of the Gorm–Tyra Fault (Figs. 6 and 9) and west of Kraka (Fig. 7). The Triassic–Jurassic cover is folded into synclinal minibasins that trend parallel to the salt structures and show thickening in their centres. Late Jurassic units show significant onlap onto the margins of the synclinal mini-basin centres. However, onlaps also occur in older units (e.g. Fig. 9). We therefore infer that Zechstein salt was initially deposited relatively evenly across the Salt Dome Province (in agreement with Duffy et al., 2013), including the area west of the Gorm–Tyra Fault. The Zechstein salt was subsequently mobilized due to rifting and eventually evacuated from below subsiding minibasins and into growing salt structures and diapirs. This process left behind the overburden welded (sensu Wagner and Jackson, 2011) to the pre-Zechstein units (see also Rank-Friend and Elders, 2004; Duffy et al., 2013).

Several prominent salt structures of varying maturities populate the current-day mobile-salt domains (Figs. 5f, 6, 8, 9). Salt stocks of different sizes populate the Salt Dome Province, including the Edna, Rolf, Dagmar, John, and Skjold structures. These all penetrate Cretaceous strata (Fig. 11c) with the John structure reaching the shallowest level around 0.6 km below sea level in the borehole John-1 (Fig. 1; Chevron Petroleum Company of Denmark, 1983). Other intrusive diapirs (sensu Jackson and Talbot, 1986) include the Gorm, South Arne, and Dan structures. At a glance, their 3D shapes are elliptical in map view and parallel to the underlying faults (Fig. 5f). Although their influence is evident at base Cretaceous levels (Fig. 11e), they have not penetrated into Cretaceous strata. The Dan structure is remarkable, as Zechstein salt has delaminated the Triassic succession laterally along weak Triassic evaporite layers to form wings or salt wedges (sensu Kockel, 2003) that extend laterally for several kilometres (Rank-Friend and Elders, 2004; Fig. 7). A handful of concordant (sensu Jackson and Talbot, 1986; e.g. pillows, anticlines, and rollers) salt structures with more than 1000 vertical metres of salt are present in the modern-day mobile-salt domains (Fig. 5f). One of them may be considered part of the Dagmar structure, whereas the four others are isolated structures. These include the named structures Kraka (ca. 1800 vertical metres of salt) and Lola, as well as a previously unnamed structure near the Jens-1 well, which we refer to as the Jens Pillow. A network of smaller salt rollers connect the larger structures in the southern part of the study area (Fig. 5f). The elongated axes of all salt structures appear to align with the underlying basement fault trends (Fig. 5f). This is especially evident to the east of the Gorm–Tyra fault. In the northeastern part of the Tail End Graben, a local Zechstein thickness maximum may represent a significant volume of non-mobilized salt considering its depth, although the data are noisy in this area (Fig. 5). Korstgård et al. (1993) interpreted such remnant Zechstein units to be present in the immediate hanging wall of the Coffee Soil Fault just north of our study area, which lends supports to this interpretation.

Beyond the mobile-salt pinch-out, we infer the presence of mechanically weak Zechstein units from the observation that the Top Zechstein reflection (base of the Triassic carapace) is smooth and continuous, while the Top pre-Zechstein is more uneven and discontinuous. In parts of the Tail End Graben, this contrast in morphology is not apparent (marked “A” on Fig. 5c). Zechstein units may be absent here or only present in thicknesses near the limit of seismic resolution. Due to low data quality at depth in this area, the exact position of the western pinch-out of the Zechstein units and their thicknesses are unclear along the southern Arne–Elin trend (Figs. 1, 5f).

Our following descriptions of Late Cretaceous inversion structures agree overall with those of Vejbæk and Andersen (1987, 2002) and Cartwright (1989). Unlike these works, we do not take into account the diachronous and gradual development of many of the structures. Instead, we address the structural connections between the upper inversion structures and the Zechstein units and pre-Zechstein basement below.

Across our study area, Late Cretaceous inversion is evident from the geometry of the open growth folds and their relations to reverse faults along the outer margins of the uplifted basins (Figs. 6–10). Relative uplift is evident especially from Lower Cretaceous units (Cromer Knoll Group) that sit at shallower levels compared to adjacent areas with no preserved Lower Cretaceous units (Figs. 6, 7, 9, 10). The distribution of major growth folds, characterized by onlapping Chalk Group strata, is evident from the isochore map of the syn-inversion Chalk Group (Fig. 11e–f). Thinning of Chalk Group units and truncations are also evident on the fold crests (marked with ∗ in Figs. 6, 8, 9, 10). Major folds generated by inversion are mostly recognizable as monoclines and asymmetric anticlines with steeper forelimbs facing the marginal troughs and more gently sloping backlimbs (Figs. 6, 8, 9, 10c, e.g. Koopman et al., 1987; Badley et al., 1989). In the most extreme cases (e.g. Bo–Jens Ridge, Fig. 6), inversion ridges express a relief of ca. 1000 m relative to the adjacent marginal basins at Base Chalk Group level.

The faults related to inversion are mostly inverted extensional faults that grew slightly upwards during shortening, giving them the typical expression with reverse offsets only in their highest parts (e.g. Figs. 6, 8, 10). Relative to folding, reverse faulting is significantly less prominent at Base Chalk Group level and in the syn-inversion strata above (Fig. 11b), as seen from e.g. the Bo–Jens Ridge and the Adda Ridge (Fig. 6). Much of the shallow reverse faulting only reached into the pre-inversion Lower Cretaceous units, while fault propagation folding probably accounted for the strain at seafloor level. Faults offsetting the base Cretaceous unconformity (Fig. 12b) reveal that reverse reactivation of normal faults was selective, as not all faults with similar strike directions inverted, as can be observed by comparing Figs. 5f, 6, and 12b, where, e.g. the Gorm–Tyra Fault does not invert in contrast to the Coffee Soil Fault. All inversion folds and faults show strike directions within WNW–ESE to N–S, which concurs with the structural trends of the sub-salt basement. From our observations, we outlined the margins of the inverted areas (indicated on Fig. 11f) outside of which no significant signs of Late Cretaceous inversion are apparent. The zone within the margins spans the entire southern study area and becomes narrower towards the north.

A number of smaller extensional growth faults occur locally in the Chalk Group, which formed and were active in Late Cretaceous times, possibly during inversion. Some likely formed due to differential compaction across large but inactive extensional faults (Fig. 11b). Others populate the crests of inversion ridges (e.g. Bo–Jens Ridge, Fig. 6, and Gorm–Lola Ridge, Fig. 9) or appear near the inversion margins (Fig. 11b and d). Local extension, e.g. due to outer-arc stretching or slope instabilities, could have generated these faults (see, e.g. Back et al., 2011).

Post-Danian inversion is evident from the wide and gentle Tyra–Igor Ridge (Fig. 6), which is expressed in the Top Chalk Group topography (Fig. 11d; see also Vejbæk and Andersen, 1987, 2002). Compared to the structural styles of Late Cretaceous inversion, this is markedly different. Instead of prominent folds along the margins of the uplifted areas, a broad and low-amplitude anticline, which is centred above the Late Jurassic rift depocentres, characterizes this inversion. In addition, no associated reactivation of extensional faults in the cover units is evident.

The thickness of the Lower Cretaceous units in the study area is small compared to that of Upper Jurassic units (Figs. 7–10). Locally pronounced thickness maxima indicate areas of increased Early Cretaceous subsidence, namely in the Arne–Elin Graben, Iris, Gulnare, and Roar basins (Fig. 12d). Of these, only the Gulnare Basin and the Arne–Elin Graben show fault-controlled subsidence as indicated by the thickness changes across adjacent faults. Along segments 2 and 3 of the Coffee Soil Fault, the Lower Cretaceous thins and onlaps towards the fault trace, suggesting no syn-depositional activity of the fault (marked as “D” in Fig. 6). In the Salt Dome Province adjacent to segments 2 and 3 of the Coffee Soil Fault (CSF), we lack a basal sub-salt slope dipping towards the master fault (Coffee Soil Fault) along with a weakened (thin and faulted) cover above the upper part of this slope. This configuration of weak zones seems to have provided ideal conditions for the thin-skinned inversion ridges along the western inversion margin. This is because the detachment and inverted cover faults, both antithetical to the relevant major basement fault (CSF 1 or the Gorm–Tyra Fault), approximate a plane dipping ca. 20–30∘, which is ideal for reverse slip. Vejbæk (1986) showed that the depocentre of the Roar Basin shows characteristic onlaps onto the flanks of a large syncline in the units below, which, along with a lack of remaining Zechstein units below the Roar Basin (marked as “A” in Fig. 5e), indicates that it formed as salt moved towards the neighbouring salt structures (see Vejbæk, 1986). The Gulnare and Iris basins may also have seen increased subsidence due to on-going salt movements towards the west and northwest after Late Jurassic rifting ceased (see also Korstgård et al., 1993).

Significant Early Cretaceous faulting occurred in the Arne–Elin Graben, focusing subsidence along its northeastern flank, which is defined by a southwest-dipping fault (Fig. 12d). Thinned units towards the northwest (marked “A” in Fig. 12c) indicate that the South Arne structure (Fig. 1a) may have been growing at this point. As mentioned, remnants of a salt ridge may extend southeast below the Arne–Elin Graben. Salt movements from this ridge towards the South Arne structure could have focused and enhanced the tectonically induced subsidence of the narrow graben, which commenced in the Late Jurassic (Møller and Rasmussen, 2003).

Continued halokinesis in the Late Cretaceous is evident from the Chalk Group isochore map (Fig. 11e–f). Here, the South Arne structure is even more pronounced as indicated by thickness variations (marked as ”A” in Figs. 11e, 12c), suggesting that it accentuated contemporaneously with the inversion of the Arne–Elin Graben. Distinct synclinal minibasins occur parallel to inversion anticlines and monoclines above the Tail End Graben and in the southern study area (Figs. 6, 8, 9), indicating that the lower part of the Chalk Group was folded and subsequently onlapped by younger units. The synclines widen with depth in the units below until they reach the basin floor. We interpret these structures as the results of combined salt movement and buckling due to inversion. Comparing the isochore maps of the Cromer Knoll Group (Fig. 12c) and the Chalk Group (Fig. 11e) shows a southward migration of the depocentre in the Early Cretaceous Roar Basin (marked as ”A” in Figs. 11e, 12c). This may indicate the progressive welding of the Triassic units to the basin floor as mobile salt migrated south along the Gorm–Tyra Fault. Additionally, Late Cretaceous synclinal minibasins formed adjacent to the Dagmar and Skjold structures (see Fig. 5f for salt structure locations) due to salt evacuation into these structures (Fig. 11e–f). As indicated by post-Danian subsidence around them (marked as ”A” in Fig. 11c), the South Arne and Edna structures continued their growth into Cenozoic times. Thickening around the Dagmar and (to some degree) John structures point towards continued growth as well (Fig. 11).

In addition to the correlation between the major Late Jurassic rift depocentres and the Late Cretaceous inversion zones (Vejbæk and Andersen, 1987, 2002; Cartwright, 1989), our results show a strong correlation between the extent of the Zechstein units and salt structures and the areas affected by Late Cretaceous basin inversion along the western inversion margin (Fig. 13). In the southern part of the study area where mobile Zechstein salt is abundant, the outer margins sit far apart. The more prominent inversion folds are limited to the margins, while the effects of shortening are relatively subdued in between. In the northern part, the inversion zone is narrower, mimicking the extent of Zechstein units below, and much more pronounced across its width (Fig. 10). Thick-skinned inversion structures along segments 2 and 3 of the Coffee Soil Fault show typical geometries related to inversion of master rift faults as outlined in Williams et al. (1989, their Fig. 1).

The physical models of Letouzey et al. (1995) illustrate kinematics that are directly applicable to the geometries observed along the western inversion margin. As known from fold and thrust belts, evaporite pinch-outs impose a major control on deformation geometries, regardless of whether they are of structural or depositional origin. This is because the pinch-outs limit the possible extent of detachment horizons for thin-skinned faults and thus localize the formation of compressional structures in the sedimentary overburden. Pre-existing salt structures, e.g. ridges, pillows, and reactive diapirs, also localize folding, reverse faulting, and thrusting in the cover during inversion (Letouzey at al., 1995). According to our interpretation (Fig. 14a), the Zechstein detachment and salt rollers along the western margin (Fig. 13) decoupled the thin-skinned inversion structures from the basement faults below. The only potential exception to this pattern (marked B in Fig. 10) is below the confluence of the Arne and Bo–Jens ridges. Due to the limited visibility here and possible absence of the Zechstein detachment, thick-skinned faults may be present. To the northwest along the base of the Arne–Elin Graben (Fig. 8) and to the south along the base of the Bo–Jens Ridge (Fig. 6), small salt ridges localized shortening, and thin-skinned faulting and folding formed the prominent inversion structures in the thinned and weakened cover units above.

We propose that the salt structures along the western inversion margin formed along the pinch-out of the Zechstein units, as mobile salt flowed up-dip on the hanging wall away from the main fault (i.e. westward) due to the tilting Top pre-Zechstein surface and the differential loading (e.g. Korstgård et al., 1993 and Geil, 1991). Late Jurassic onlaps onto the salt-ridge cover below the Bo–Jens Ridge support this (Fig. 6). In the case of the Arne–Elin Graben, passive diapirism likely occurred beneath during the Early Cretaceous.

These localizations of shortening by salt structures imply that the inverted cover units along the western margin were pushed up-dip along the Zechstein detachment relative to the basement below during shortening (Figs. 13, 14a). Segment 1 of the Coffee Soil Fault and the Gorm–Tyra Fault lie parallel to the inverted structures along the western margin and can be considered the master faults of the respective half-graben in their hanging walls. Reverse reactivation of these faults potentially provided the basement shortening needed to balance the observed thin-skinned shortening higher on the half-graben dip-slopes. If so, it seems curious that the typical folds indicating such reactivation and thick-skinned inversion do not occur directly above segment 1 of the Coffee Soil Fault (Fig. 8) and the Gorm–Tyra Fault (Fig. 6), as is the case along segments 2 and 3 of the Coffee Soil Fault (Fig. 6: Adda Ridge; Fig. 13: Igor–Emma Ridge). In the following, we apply a conceptual model described by Stewart (2014; Fig. 15) to explain this geometry. It balances thin-skinned extension and subsequent shortening along the western margin to normal and reverse slip on the master basement faults without generating inversion structures directly above them. Figure 15 shows the development of an initial half-graben with a weak basal detachment (corresponding to the Zechstein salt) that eventually sees thin-skinned inversion focused in the hanging wall opposite to the basement fault. There are no direct signs of inversion above the main graben-bounding fault. This geometry occurs in several inverted basins in the North Sea and elsewhere (Stewart, 2014). The evolution in the study area can thus be subdivided into the following phases.

Extension. A thick-skinned master fault (corresponding to the Coffee Soil Fault segment 1 and the Gorm–Tyra Fault) bounding a half-graben initially offsets both basement and cover. During continued extension, domino-block rotation gradually reduces the dip angle of the master fault while increasing the detachment dip angle towards it. The resultant increase in normal stress along the plane of the master fault means that friction eventually inhibits slip on the upper part of it. Instead, it becomes more mechanically feasible to accommodate slip along the now sloping detachment. This causes a fault weld to form as the cover starts rolling down the plane of the master fault with continued extension. An extensional triangle zone sensu Stewart (2014; Figs. 14a, 15) is now apparent in the hanging wall basement. Higher on the dip-slope, thin-skinned normal faults soling out in the detachment now form in the thinner and weaker cover here (corresponding to the Arne–Elin Graben and pre-existing structure at the Bo–Jens Ridge). Along the master fault, a symmetrical syncline becomes apparent directly above the triangle zone (Fig. 15). The formation of salt rollers at the top of a rotated “domino block” is enhanced by the increasing dip of the basal salt. These later become important during shortening (Ferrer et al., 2014).

The rate of displacement during rifting and inversion must have been relatively low to allow the up-dip push of the cover units along the thin Zechstein detachments during shortening. Additionally, the precise areal extent of slip on the Zechstein detachments remains unknown. The absence of resolvable Zechstein units and the consequent unlikelihood of a basal detachment prove to be a problem for the triangle zone model in part of the Tail End Graben (marked “A” in Fig. 5c). This area without salt underlies the intersection of the Arne and Bo–Jens ridges (Fig. 10) that are both decoupled by salt to the northwest (Fig. 8) and south (Fig. 6). At the same time, a higher intensity of folding in the cover units is apparent across this part of the inverted Tail End Graben. In those adjacent areas, folding is more localized along the western inversion margin above salt structures. We suggest that the broader zone of inversion-related folding in part of the Tail End Graben followed as a consequence of the local lack of a basal salt detachment.

Even when salt is present in its hanging wall, thick-skinned inversion styles occur along segment 3 of the Coffee Soil Fault (Igor–Emma Ridge, Fig. 7). This is consistent with the observations of Jackson et al. (2013) from the Egersund Basin (offshore Norway). We suggest that the relatively horizontal basin floor in the Salt Dome Province does not provide a geometry enabling initial gravity-driven thin-skinned extension along faults detaching along the Top Zechstein and consequently no thin-skinned inversion. The present dip may however be limited precisely because of the inversion along the Coffee Soil Fault, but there are no signs of basement inversion of a magnitude justifying such an interpretation. On the contrary, the main inverted faults atop the Kraka structure (Fig. 7) and the reverse fault of the Gorm–Lola Ridge (Fig. 9) both sole out into thin Triassic detachments with significant dip angles. Along with the interpretations of thin-skinned structures mentioned above, this demonstrates that even thin evaporite layers and apparent welds (sensu Jackson et al., 2014) can be activated as detachments during inversion if their orientations are favourable. Additionally, if mobile evaporites are present in sufficient amounts to form ridges or reactive diapirs, thin-skinned folds and faults in the overburden will initiate from these or from dipping detachment horizons above their flanks. Figure 14b illustrates the pre-inversion Kraka structure and the reactivation of thin-skinned faults soling out into Triassic evaporites on its flank in the Late Cretaceous.

In central areas with potentially mobile Zechstein units at the time of inversion, the effects of shortening are not as evident as along the inversion margins. As mentioned, we interpret the narrow Late Cretaceous synclines as resulting from an interplay of inversion-related buckling and salt evacuation. Aside from the subtle reactivation along the crests of the Kraka structure and Gorm–Lola Ridge, we are able to infer basin inversion only from the Igor–Emma Ridge in the southern part of the study area. Still, the mobile-salt structures striking WNW–ESE to N–S are likely to have taken up a significant amount of shortening via lateral squeezing of the salt bodies. This would have tightened the cover folds above, explaining, e.g. the slightly anticlinal expression of the Dan structure on the Chalk Group isochore map (Fig. 11e), and made the shape of the structures elongate perpendicular to the shortening direction. Aside from the Gorm–Lola Ridge reverse fault (Fig. 9), no structures directly indicate shortening below the uppermost syn-rift deposits. This implies that if cover structures recording shortening cannot be identified, regional markers are absent, or the salt budget is unconstrained, then the thick mobile evaporites can mask the effects of shortening in mildly inverted rift basins.

Other inverted basins in the North Sea area show a similar strong control imposed by the Zechstein salt, e.g. the Broad Fourteens Basin (offshore Netherlands; e.g. Nalpas et al., 1995) and the Sole Pit Basin (offshore UK; Glennie and Boegner, 1981; van Hoorn, 1987). Where present in these basins, its ability to decouple deformation in basement and cover strongly controlled inversion, which resulted in structural styles comparable to those documented in our study area.

As explained above, we infer from the Late Cretaceous inversion structures that the Gorm–Tyra Fault and all three segments of the Coffee Soil Fault experienced reverse slip during inversion (Fig. 13). As also mentioned, all basement faults have conserved a net normal offset in spite of basement shortening, and there are no indications of reverse slip on any of the smaller basement faults. Reactivation in the basement during inversion therefore seems to be restricted to only the largest faults, which agrees with other studies of inverted basins (e.g. Grimaldi and Dorobek, 2011; Reilly et al., 2017). As indicated by the continuous inversion margins, the Gorm–Tyra Fault and segments 2 and 3 of the Coffee Soil Fault probably reactivated along their full lengths within the visible area (see below; see, e.g. Reilly et al., 2017). Following the eastern inversion margin, the Adda and Igor–Emma Ridges indicate higher degrees of shortening relative to the area above the Poul Plateau. Here, only minor reverse offsets in the upper parts of the bounding faults occur. Below the Adda and Igor–Emma Ridges (Figs. 6, 7), the normal throw of the Coffee Soil Fault is significantly larger than along the Poul Plateau area, as indicated by the depth to the Top pre-Zechstein surface in the immediate hanging wall (Fig. 5a). This agrees with earlier studies into the variations of reverse displacement along the upper parts of mildly inverted normal faults. These found that the magnitude of reverse displacement along the upper parts of the faults mimic that of their deeper normal-displacement variations (Jackson et al., 2013; Reilly et al., 2017).

The triangle zone concept of Stewart (2014) mentioned above offers an explanation as to why no reverse displacement is evident from the upper parts of segment 1 of the Coffee Soil Fault and the Gorm–Tyra Fault (Fig. 14a). Roberts et al. (1990) provided a similar explanation of the inversion structure, Lindesnes Ridge, in the Norwegian part of the Feda Graben (Fig. 1a). Here, the application of the triangle zone concept provided a simple basin geometry, free of inferred and unlikely basement geometries or basement faults that coincidentally inverted to net zero displacement. Likewise, in our case, the concept resolves the need to include large antithetic basement faults below the western inversion margin, where they are not indicated by the seismic data, e.g. below the Arne–Elin Graben (Fig. 8; cf. Vejbæk and Andersen, 1987; Cartwright, 1989).

To explain the Late Cretaceous inversion structures of the DCG, Vejbæk and Andersen (1987, 2002) proposed a NNE–SSW-directed shortening, which induced a mildly dextral transpressional component along NNW–SSE-striking structural lineaments in the area. They note, however, that N–S-striking folds (e.g. Bo–Jens Ridge and Gorm–Lola Ridge) are not consistent with this tectonic regime, and suggest that Zechstein-salt movements played a role in their formation (Vejbæk and Andersen, 2002). Our results strongly support this hypothesis. Cartwright (1989) discarded the need for a strike-slip component and argued for a simple NE–SW-directed shortening to explain the range of strike orientations seen in inversion folds (WNW–ESE to N–S, Fig. 11f). We tend to agree with Cartwright (1989), as we have found only few indications of strike-slip components related to inversion. The most significant are a group of small normal faults atop the crest of the Bo–Jens Ridge (Fig. 11b and d). These strike approximately NE–SW, which is compatible with a dextral shear along the N–S-striking axis of the fold according to the oblique-inversion models of Letouzey et al. (1995). However, we propose that a few minor faults cannot justify the idea that the entire DCG underwent transpression.

The average NW–SE strike of the inversion-related structures indicate an apparent overall NE–SW shortening. As casually pointed out by Cartwright (1989), the total apparent shortening of the rift during Late Cretaceous inversion amounted to only “a few percent” along this axis. We have performed no quantitative analyses to test this claim as, e.g. Jackson et al. (2013) did on data from the Egersund Basin. We concur by referring to our observation that inversion-related folds are open and that the degree of inversion (sensu Cooper et al., 1989) is low, as evident from the interpreted seismic sections and the net extensional offset on basement structures. Still, standard methods such as line-length balancing have been shown to grossly underestimate the amount of shortening in physical models using wet clay (Eisenstadt and Withjack, 1995). If shortening occurred without producing significant reverse movement on basement faults, the magnitude of shortening due to inversion of the DCG may be greater than expected.

Here, “ductile” refers to deformation that is apparently continuous at the seismic scale but may be brittle at a smaller scale. A significant amount of cover folding occurred during Late Cretaceous inversion, especially in the narrow northern inverted area (Figs. 10 and 11f), even though Zechstein salt may not have been present at depth to form a thrust detachment for this thin-skinned deformation. An overall impression of a diffusely distributed ductile style of shortening is apparent from the cover units in this zone, especially when also considering the gentle basin-wide flexure with no significant seismic-scale faults associated with the Paleogene inversion at the Tyra–Igor Ridge (Figs. 6, 10). Cartwright (1989) suggested that the Late Jurassic shales in the DCG (see, e.g. Michelsen et al., 2003) could have retained anomalously high pore pressures at the time of inversion and that tectonic compaction of these units could have taken up a significant amount of shortening during inversion. We concur, as this explains the high degree of folding and ductile thickening of the cover units in the mentioned area, while reverse faults are less abundant. This implies that the degree of shortening could be greater than anticipated from quantitative analyses, as explained above.

Additionally, flexural slip (Tanner, 1989) probably occurred to a high degree across the inverted zones in the study area to allow the significant folding of kilometre-thick units and often limited faulting as resolved on the seismic data, e.g. in the Bo–Jens Ridge (Fig. 6). Again, overpressure could have greatly contributed to the implied shearing parallel to bedding. This mechanism may further explain the often high degree of folding relative to faulting in the Triassic to Lower Jurassic carapace in the southern part of the study area (e.g. Fig. 9), which resulted from both tectonic extension and compression.