Rift basins within the study area and the wider North Sea have been active during multiple regional tectonic events since at least the Carboniferous–Permian (Deeks and Thomas, 1995; Erlström et al., 1997; Graversen, 2009; Jensen and Schmidt, 1993; Michelsen and Nielsen, 1993; Mogensen, 1995; Mogensen and Korstgård, 2003; Thybo, 2000). Variscan orogenesis drove regional N–S-directed shortening in the Carboniferous–Permian, resulting in the formation of a regional W- to NW-trending dextral strike-slip system, incorporating the W-trending Fjerritslev faults (Skjerven et al., 1983), in addition to several N- to NE-trending structures such as the Varnes and Skagerrak grabens, situated north and east of the study area, respectively (Fig. 1a; Heeremans and Faleide, 2004; Heeremans et al., 2004; Lassen and Thybo, 2012; Ro et al., 1990).

Fault systems within the study area were intermittently active during the Mesozoic in response to several tectonic events affecting the broader North Sea (e.g. Bergerat et al., 2007; Deeks and Thomas, 1995; Erlström et al., 1997; Liboriussen et al., 1987; Mogensen, 1995; Mogensen and Jensen, 1994; Mogensen and Korstgård, 2003). During the Permian–Triassic, E–W continental extension occurred in response to the breakup of Pangaea, leading to the formation of a predominately N-trending rift across the North Sea (e.g. Bell et al., 2014; Færseth, 1996; Ziegler, 1992). In the Horn Graben, south of the Farsund Basin, rifting and associated activity on N–S-striking faults were initiated during the Late Permian (Vejbæk, 1990; Fig. 1a). Rifting migrated northwards into the Norwegian–Danish Basin during the Triassic, towards the southern margin of the study area, where additional N–S-striking faults occur (Fig. 1c). An additional rift phase, centred on the central North Sea and related to the collapse of a Middle Jurassic thermal dome (Rattey and Hayward, 1993; Underhill and Partington, 1993), was initiated in the Late Jurassic, continuing until the Early Cretaceous. The extension direction during this rift phase varies spatially, with NW–SE to E–W extension proposed for areas north of the study area in the northern North Sea (e.g. Bell et al., 2014; Brun and Tron, 1993; Doré et al., 1997; Færseth, 1996) and NE–SW extension proposed south of the study area in the Central Graben and southern North Sea (Coward et al., 2003). During the Late Cretaceous, horizontal shortening, related to far-field tectonic effects from the Alpine Orogeny, drove basin inversion within several North Sea sedimentary basins (Biddle and Rudolph, 1988; Cartwright, 1989; Jackson et al., 2013; Mogensen and Jensen, 1994; Phillips et al., 2016). This inversion was amplified along the upper-crustal expression of the STZ (Thybo, 2000), being associated with regional uplift and the reverse reactivation of basin-bounding faults (e.g. Bergerat et al., 2007; Deeks and Thomas, 1995; Jensen and Schmidt, 1993; Liboriussen et al., 1987; Mogensen and Jensen, 1994). Further uplift and erosion occurred during the Neogene due to uplift of the South Swedish Dome (Japsen et al., 2002; Jensen and Schmidt, 1993), resulting in erosion of Cretaceous strata across much of the study area.

The E-trending Farsund Basin is oriented at an unusually high angle, in many cases almost perpendicular, to largely N-trending structures across the North Sea. To the south, in the Horn Graben, faults primarily strike N–S to NW–SE (Glennie, 1997; Vejbæk, 1990), whereas to the north and west, where they define the Varnes Graben and Lista Nose, they strike N–S (Heeremans et al., 2004; Lewis et al., 2013; Skjerven et al., 1983; Fig. 1a). Furthermore, to the east, NE–SW- and NW–SE-striking faults occur in the Skagerrak Graben and along the eastern part of the STZ, respectively (Mogensen and Jensen, 1994; Ro et al., 1990; Fig. 1a). The E–W geometry of the faults defining the Farsund Basin means that they are not optimal for reactivation during any of the North Sea regional tectonic events outlined above, suggesting they record a hitherto undocumented phase of tectonic activity or, more likely, are influenced by a pre-existing structure such as the STZ.

The Tornquist Zone forms a lineament that spans the lithosphere and extends > 1000 km across central and northern Europe (Alasonati Tašárová et al., 2016; Berthelsen, 1998; Mazur et al., 2015; Pegrum, 1984). The Tornquist Zone comprises two segments: the Teisseyre–Tornquist Zone (TTZ) in the south, extending northwest from the Carpathian orogenic front to the Rønne Graben (e.g. Alasonati Tašárová et al., 2016; Berthelsen, 1998; Grad et al., 1999; Guterch et al., 1986; Pharaoh, 1999; Fig. 1a) and the Sorgenfrei–Tornquist Zone (STZ) in the north, continuing northwest from the Rønne Graben to the Farsund Basin (e.g. Babuška and Plomerová, 2004; Berthelsen, 1998; Pegrum, 1984; Thybo, 2000) and possibly extending further westwards beneath the main North Sea rift (Pegrum, 1984; Fig. 1a).

The Tornquist Zone, including the STZ and TTZ, has been extensively studied using a variety of geological and geophysical methods, including seismic tomography (Cotte and Pedersen, 2002; Voss et al., 2006), seismic refraction (Alasonati Tašárová et al., 2016; Guterch and Grad, 2006; Guterch et al., 1986), seismic anisotropy (Babuška and Plomerová, 2004) and seismic reflection surveying (Grad et al., 1999; Lassen and Thybo, 2012; Thybo, 2000). These data suggest that the lineament separates thick, old cratonic lithosphere of the East European Craton, including Baltica, from the younger, thinner lithosphere associated with assorted Palaeozoic terranes that comprise present-day central and western Europe; the lineament thus represents a major change in lithospheric properties and thickness (e.g. Babuška and Plomerová, 2004; Berthelsen, 1998; Cotte and Pedersen, 2002; Erlström et al., 1997; Kinck et al., 1993; Michelsen and Nielsen, 1993; Pegrum, 1984; Pharaoh, 1999; Voss et al., 2006).

At the junction of the TTZ and STZ, offshore southern Sweden, a zone of NW-diverging splay faults, termed the “Tornquist Fan” occur, demarcated to the north and south by the STZ and a further regional structure, the Trans-European Fault, respectively (Thybo, 2000, 2001). Here, the STZ is still defined as a change in lithospheric thickness at sub-crustal, i.e. upper mantle, levels (e.g. Babuška and Plomerová, 2004; Berthelsen, 1998; Cotte and Pedersen, 2002). Geological evidence, in the form of drilled crystalline basement, shows that basement of Baltica affinity is present on either side of the STZ (Berthelsen, 1998), indicating that the margin of Baltica as identified at sub-crustal depths does not correspond to the same location at upper-crustal levels, with crystalline basement of Baltica affinity continuing south of the STZ (Fig. 1d). At upper-crustal levels, the STZ is defined by a zone of Late Cretaceous inversion (e.g. Bergerat et al., 2007; Deeks and Thomas, 1995; Michelsen and Nielsen, 1993; Mogensen, 1995; Mogensen and Jensen, 1994; Pegrum, 1984), which correlates in plan view to the STZ as defined by the change in lithospheric thickness at sub-crustal levels (Babuška and Plomerová, 2004; Liboriussen et al., 1987; Mogensen and Jensen, 1994). Several authors suggest the STZ represents a weak zone during later tectonic events, acting to accommodate stresses between adjacent crustal blocks (Mogensen and Jensen, 1994; Mogensen and Korstgård, 2003); however, the link between different structural levels and, therefore, how a change in lithospheric thickness may behave kinematically and influence the development of overlying upper-crustal rift systems' later events, has not been established. Understanding how these sub-crustal structures are expressed within rift systems is vital to understanding both the local rift evolution and the causal stress field.

Seismic interpretation focused on a 500 km2, borehole-constrained 3-D seismic reflection dataset covering the southern margin of the Farsund Basin (Fig. 1c). These data image to 4 s two-way time (TWT; ∼ 6 km), having inline and crossline spacings of 18.75 and 12.5 m, respectively. Regional 2-D seismic reflection datasets covering the entire basin were used to provide regional structural context for the area imaged by the 3-D dataset; this 2-D dataset consists of closely spaced (∼ 3 km), N-trending seismic sections that are oriented perpendicular to the dominant E–W structural trend (see Fig. 1c for the locations of sections used in this study and Table S1 in the Supplement for additional information on acquisition and processing parameters). The 2-D seismic reflection data are zero phase and follow the SEG reverse polarity convention whereby a downward increase in acoustic impedance is represented by a trough (red) and a downward decrease in acoustic impedance is represented by a peak (black). The 3-D seismic volume follows the opposite (normal) convention. Image quality is excellent throughout the 2-D and 3-D datasets at shallow levels, although quality deteriorates at depth; the 2-D sections image to deeper structural levels (7 s TWT), with some reaching 12 s TWT (e.g. Fig. 2), and cover a wider area (approximately 2000 km2) than the 3-D dataset. The ages of key horizons are constrained using stratigraphic information from well 11/5-1, located in the area covered by the 3-D dataset, and wells 9/3-1, 10/5-1, 10/7-1, 10/8-1 and 11/9-1, located in the wider region imaged by the 2-D seismic data (Fig. 1a). Checkshot information from these wells was used to convert structural measurements (i.e. specific spot measurements of fault cut-off depths and horizon depths for fault offset and thickness calculations) from the time to the depth domain.

We mapped six key tectonostratigraphic horizons throughout the area, along with two supplementary ones (Figs. 1b, 3). The six main horizons define the main stratigraphic intervals of the basin and correspond to the acoustic basement surface, the base Jurassic unconformity (BJU), the top Jurassic surface, the top Lower Cretaceous, base Cenozoic unconformity and the seabed (Figs. 1b, 2). Where present, the base of the Upper Permian Zechstein Supergroup salt represents the acoustic basement, i.e. the deepest mappable coherent reflection within the study area (Figs. 3, 4). Where Zechstein salt is absent and Triassic strata directly overlie pre-Upper Permian strata, or where erosion by the BJU removes Triassic strata completely, we map these basal reflections (i.e. base Triassic or BJU) as the acoustic basement (Figs. 4, 5). Due to the large thickness of the Lower Cretaceous interval and the associated burial of the tips of the main E–W-striking faults, we mapped an internal surface within the Lower Cretaceous interval, termed the intra-Lower Cretaceous (ILC) horizon (Figs. 4, 5). A further horizon, proposed as corresponding to the base of a Carboniferous–Permian aged interval was mapped locally beneath the acoustic basement horizon. Within the 3-D volume, we generated a series of isochrons (TWT interval thickness maps) between horizons to define the structural style and infer the subsidence history of the basin (Fig. 6). These isochrons correspond to the Triassic interval (between the acoustic basement and base Jurassic unconformity), the Jurassic (between the BJU and top Jurassic) and the Lower Cretaceous (between the top Jurassic and top Lower Cretaceous).

Within these surfaces, faults are represented by black polygons, defined by hangingwall and footwall cut-offs of the horizons against the fault plane (Figs. 3, 6). These cut-offs are used as input for quantitative fault analyses. The width of the polygon (i.e. the horizontal distance between the hangingwall and footwall cut-off) is representative of the fault heave. Due to difficulties in accurately determining the cut-offs and the fact that throw is much larger than heave in extensional settings (reducing potential measurement errors), we measured fault throw rather than heave in our quantitative fault analysis.

To constrain the geometry and growth of the upper-crustal fault population, we mapped and performed quantitative analysis of the major faults delineating the basin. Horizon cut-offs, fault tip-lines and fault intersections (as defined by branch lines) were mapped to minimise artefacts that may lead to an incorrect assessment of fault kinematics (e.g. Duffy et al., 2015; Walsh et al., 2003; Yielding, 2016). Throw–length (T-x) plots, where throw is measured at regular intervals along the fault for different stratigraphic levels (see Appendix A), were then calculated and were subsequently used as input for fault displacement backstripping (Figs. 7, 8), a technique used to unravel the kinematic and geometric evolution of a fault throughout its history (see Appendix B; see Jackson et al., 2017). These analyses only record the vertical displacement along the fault plane throughout its evolution; we make no inferences as to how this throw accumulated (i.e. oblique slip/dip slip). T-x plots allow us to examine the distribution of throw along a fault and help elucidate its kinematic history. Fault displacement backstripping expands upon this; by systematically removing throw accrued during specific stratigraphic intervals, starting with the youngest, we are able to quantitatively examine the throw distribution along the fault throughout its history and hence determine its geometric evolution.

The E–W-striking Fjerritslev north and Fjerritslev south faults delineate the southern margin of the Farsund Basin; these faults merge laterally to the east to form a single structure, the Fjerritslev Fault system (Fig. 1c). A S-dipping fault, termed the Farsund north fault, bounds the Farsund Basin to the north (Fig. 2), separating it from N–S-striking faults associated with the Varnes Graben (Fig. 1c). Projecting these basin-bounding faults downwards, based on their overall dip and subtle reflection terminations at depth, the N-dipping Fjerritslev north and south faults appear to merge with S-dipping faults on the northern margin of the basin, including the Farsund north fault, potentially at a depth of approximately 7 s TWT (15 km; Fig. 2). Together, these structures thus appear to form a system that, at least superficially, resembles a negative flower structure (Fig. 2). Although the exact location and nature of this apparent geometric link at depth is uncertain due to poor seismic image quality, we infer that the faults defining the basin are geometrically linked, both along strike and down dip (Fig. 2). In plan view, the Fjerritslev north and south faults also merge eastwards (Fig. 1c).

We observe high-amplitude reflections 9–10 s TWT at the basin margins; we interpret these as Moho-related reflections, which are noticeably absent directly beneath the Farsund Basin (Fig. 2). In addition, east of the Farsund Basin, Moho-related reflectivity is observed at 10–12 s TWT (approximately 30 km); a key observation in this location is that this reflectivity is offset by 1–2 s TWT (4–5 km) across the Fjerritslev Fault system (Kind et al., 1997; Lie and Husebye, 1994). Furthermore, again east of the Farsund Basin along the STZ, Deeks and Thomas (1995) observe a zone of high reflectivity within the lower crust beneath the STZ, which they interpret as an area where deformation is transferred between the upper- and sub-crustal lithosphere in a more ductile manner. We propose that the faults defining the Farsund Basin represent the brittle upper-crustal component of the STZ, although the exact nature of the linkage between the upper-crustal faults and sub-crustal lithospheric step is uncertain. One potential mechanism is that the crustal-scale faults offset the Moho, causing it to be poorly imaged below the basin. We speculate that these crustal-scale faults extend deeper below the Moho to link to the lithospheric step associated with the STZ as defined at sub-crustal levels (Figs. 1c, 2; e.g. Babuška and Plomerová, 2004; Berthelsen, 1998; Cotte and Pedersen, 2002). Based on the anomalous, overall E–W strike of the basin-bounding faults compared to the broader northerly trending North Sea rift (Fig. 1a) and the proposed linkage between the upper crust and sub-crustal lithosphere, we argue that, by studying the geometric and kinematic evolution of the well-imaged upper-crustal fault systems, we can gain a better understanding of the kinematic behaviour of the whole lithosphere-scale STZ.

In plan view, across the southern margin of the Farsund Basin, the acoustic basement surface is characterised by the E–W-striking Fjerritslev north and south faults and a N-trending fault population (Figs. 3, 4). The Fjerritslev north fault is approximately 70 km long (32 km within the 3-D volume; Figs. 1a, 3), with a series of relay ramps separating the fault into three 10 km long segments within the area covered by the 3-D volume (Fig. 3). The Fjerritslev south fault is approximately 75 km long (38 km within the 3-D volume; Fig. 1a) and shares several branch lines with N–S-striking faults (Fig. 3).

Two major E-dipping, N–S-striking faults, hereby termed NS1 and NS2 from north to south, respectively, dissect the basin (Figs. 3, 5). NS1 and NS2 lie in the hangingwall and footwall of the Fjerritslev north fault, respectively, and also abut this structure. The two branch lines are laterally offset by approximately 10 km. Further south, NS2 cross-cuts the Fjerritslev south fault (Fig. 3). We observe additional N-striking, E-dipping faults within the hangingwalls of NS1 (i.e. HF1) and NS2 (i.e. HF2; Fig. 5). HF2 links and terminates against the Fjerritslev north and south faults (Fig. 3). Another N–S-striking fault, termed NS3, is located east of HF2 where it abuts the footwall of the Fjerritslev south fault (Fig. 3). A series of minor (∼ 50 ms TWT throw), N–S-striking faults is present across the footwall of NS2; these faults cross-cut the Fjerritslev south fault and are eroded by the BJU to the north, terminating at the acoustic basement surface within the footwall of the Fjerritslev north fault (TF1-4; Figs. 3, 5). Although not observed directly, based on 2-D seismic data located to the south of the 3-D dataset, we propose that these N–S-striking faults may merge, forming a single structure (Fig. 1c).

Supra-salt structural levels (i.e. BJU and above) are dominated by the E–W Fjerritslev north and south faults (Fig. 4), with a key observation being that N–S-striking faults, with the exception of NS2, are absent (Figs. 3, 5). At these structural levels, the Fjerritslev south fault terminates at the NW–SE-striking HF2 (Fig. 3). Numerous faults, displaying a range of strikes, are present within the footwall of the Fjerritslev south fault at supra-salt structural levels; these represent thin-skinned, salt-detached faults that are accordingly not expressed at sub-salt structural levels (i.e. the acoustic basement; Figs. 2, 4). No thick-skinned (i.e. basement-involved) faults are present at the top of the Lower Cretaceous; the only faults present are arcuate, broadly E–W-oriented, S-dipping, salt-detached faults located along the footwall of the Fjerritslev south fault (Figs. 3, 4).

Regional 2-D seismic reflection data indicate the acoustic basement is reflective but contains very little in the way of coherent, mappable reflectivity. Discrete reflections likely represent Carboniferous–Permian strata (Figs. 4, 5; e.g. Mogensen and Jensen, 1994; Sørensen and Tangen, 1995). Acoustic basement reflections are truncated at base salt (or its laterally equivalent stratigraphic horizon where salt is absent; Fig. 4). Carboniferous–Permian strata are tabular in both the hangingwalls and footwalls of the Fjerritslev north and south faults, indicating these structures were inactive. Apparent thickening of Late Palaeozoic strata into the hangingwall of the Fjerritslev north fault appears to simply reflect increased updip erosion of strata below the BJU (Fig. 4). Synkinematic strata also appear absent in the hangingwall of NS2, although this may be due to poor at-depth imaging within the 3-D volume (Fig. 5). Instead, Carboniferous–Permian strata appear to thicken regionally to the south (Figs. 2, 4). The seismic–stratigraphic architecture of sub-salt strata suggests E–W-striking faults were inactive during the Carboniferous–Permian (Figs. 4, 5).

The Triassic interval does not thicken into the Fjerritslev south fault along most of its length (Figs. 2, 4), although a depocentre, which may be related to salt withdrawal and not fault movement, may potentially be present within the hangingwall of the eastern segment (Fig. 6a). The Triassic interval also does not thicken towards the eastern part of Fjerritslev north fault (i.e. east of its branch line with NS2), suggesting at least this part of the fault was inactive at this time (Fig. 6a). A critical observation we make is that a thick, tabular, seemingly pre-kinematic package of Triassic strata is preserved within the hangingwall of the central segment of the E–W-striking Fjerritslev north fault (i.e. between its branch lines with NS1 and NS2; Figs. 4, 6a). With regards to activity along the N–S-striking faults, the Triassic interval thickens across NS1 and NS2 (Fig. 5, 6), with Triassic strata also preferentially preserved in the hangingwalls of other N–S faults beneath the BJU (Fig. 6a). Triassic strata are largely eroded across the footwall of NS2 and are completely eroded from the footwall of NS1, with the thickness of missing section increasing northwards (Fig. 4, 6a). Our observation that Triassic thickness changes principally reflect differential preservation of strata beneath the BJU indicates that any activity along depocentre-bounding faults must have occurred prior to BJU erosion in the Early Jurassic (Figs. 5, 6a). Therefore, Triassic activity predominately occurred on N–S-striking faults, with the central segment of the E–W Fjerritslev north fault and potentially the eastern segment of the Fjerritslev south fault also appearing active (Fig. 6a). The other segments of the E–W-striking Fjerritslev north and south faults were inactive at this time (Figs. 2, 4, 6a).

The Jurassic interval is relatively thin across the basin (approximately 200 ms TWT, ∼ 250 m) and of constant thickness within individual fault blocks (Figs. 2, 6b). A subtle, stepwise basinward thickening (approximately 60 ms TWT, 75 m) occurs northwards across the Fjerritslev north and south faults (Fig. 6b), with relatively minor thickness changes (approximately 40 ms TWT, 50 m) also occurring across NS1 and NS2 (Figs. 4, 6b). Based on the lack of obviously fault-driven, short-wavelength changes in Jurassic sediment thickness, we propose the Middle and Late Jurassic (i.e. post BJU-erosion) were a time of relative tectonic quiescence (Figs. 4, 6b).

The Lower Cretaceous interval thickens northwards across the E–W-striking Fjerritslev south and north faults, reaching a maximum thickness of approximately 1400 ms TWT (1800 m) in the centre of the basin (Figs. 4, 6c). Fault-related thickening is observed along the entire length of the Fjerritslev north fault, with strata thickening eastwards towards the largest depocentre occurring next to the eastern segment (Fig. 6c). Three depocentres occur along the Fjerritslev south fault: a minor, E-trending depocentre in the west (approximately 700 ms TWT, 850 m); a major E-trending depocentre situated between the TF1 branch line location and the NS2 branch line (approximately 850 ms TWT, ∼ 1000 m; Fig. 6c); and a further, NE-trending depocentre to the east of the NS2 branch line (approximately 800 ms TWT, ∼ 900 m; Fig. 6c). Lower Cretaceous strata do not appreciably thicken across N–S faults, and this interval is partially eroded to the northwest by the base Cenozoic unconformity (Figs. 4, 6c). Upper Cretaceous strata are largely absent across the basin due to erosion along the base Cenozoic unconformity; strata of this age are only preserved along the southern basin margin (Figs. 2, 4).

Palaeozoic–Mesozoic deformation in the Farsund Basin was accommodated by both E–W- and N–S-striking faults, although fault activity was strongly partitioned in time and space. Triassic faulting primarily occurred along N–S-striking faults (Figs. 5, 6a), along with isolated segments of E–W-striking faults (Figs. 4, 6a). In contrast, in the Early Cretaceous, following Early–Middle Jurassic regional uplift and erosion and a period of relative tectonic quiescence during the Middle and Late Jurassic, activity preferentially occurred on E–W-striking Fjerritslev faults (Figs. 2, 4, 6c).

Triassic extension was concentrated on N–S-striking faults, as well as relatively short, discrete segments of the E–W-striking Fjerritslev north and south faults (Fig. 6a). The central segment of the Fjerritslev north fault, between its branch lines with NS1 and NS2, corresponds to a marked increase in throw (approximately 1000 ms TWT, 1300 m) as measured along the acoustic basement horizon (Fig. 7a). Throw increases sharply at the branch lines with NS1 and NS2, with significantly larger throws (approximately 1500 ms TWT, 2000 m) observed along the central segment compared to the western and eastern segments (approximately 500 ms TWT, 650 m; Fig. 7a). Throw on NS1 is ∼ 950 ms TWT (1700 m) at the branch line with the Fjerritslev north fault, decreasing northwards to ∼ 500 ms TWT (∼ 800 m; Fig. 7a); throw on NS2 at its branch line with the Fjerritslev north fault is ∼ 1200 ms TWT (1700 m), decreasing southwards to ∼ 600 ms TWT (850 m; Fig. 7a). NS1 and NS2 are eroded along the BJU; therefore, throw on these faults as calculated across the acoustic basement horizon largely represents throw accrued during the Triassic. NS2 shows relatively minor, post-Triassic activity (Figs. 5, 9).

Due to BJU erosion and the related absence of Triassic strata, the western segment of the Fjerritslev north fault only records post-BJU activity (Fig. 6a). Along the eastern segment of the fault, where Triassic strata are preserved, the same amount of throw occurs as along the western segment (approximately 500 ms TWT, 650 m), indicating that both segments were only active post-BJU, likely during the Early Cretaceous (Fig. 7). Only the central segment of the Fjerritslev north fault shows any pre-Early Cretaceous activity, with throw backstripping showing that the central segment of the Fjerritslev north fault, along with NS1 and NS2, accrued ∼ 1000 ms TWT (1300 m) during the Triassic (Fig. 7b).

A further, discrete Triassic throw increase is observed along the eastern segment of the Fjerritslev south fault, located between HF2 in the west and NS3 to the east (Figs. 3, 8). To the west of HF2, throw along the acoustic basement, BJU and top Jurassic horizons remains relatively constant at ∼ 500 ms TWT (∼ 800 m), indicating a lack of pre-Early Cretaceous activity at this time (Fig. 8). However, between the HF2 and NS3 branch lines, throw increases to ∼ 700 ms TWT (1300 m), a 200 ms TWT increase above the BJU in this area, seemingly representing ∼ 200 ms TWT of Triassic throw in this area. Throw along the abutting HF2 and NS3 is similar, ∼ 200 ms TWT (500 m; Figs. 3, 5). Fault displacement backstripping indicates that, west of the branch line with HF2, the Fjerritslev south fault was initiated during the Early Cretaceous, and that only the eastern segment was active during the Triassic (Fig. 8). An Early Cretaceous age for initiation of faulting along the western part of the Fjerritslev south fault is further supported by our prior observation that Carboniferous–Permian, Triassic and Jurassic strata do not thicken across it (Figs. 2, 4).

Segments of the Fjerritslev north and south faults, located between branch lines with N–S-striking faults, appear to have been active during the Triassic, at the same time as the N–S-striking faults (Figs. 7b, 8). The main phase of activity along other parts of the E–W-striking faults occurred later, during the Early Cretaceous (Figs. 7b, 8b). One possible explanation for this relatively early, discrete Triassic activity could be that local segments of pre-existing E–W-striking faults are reactivated between the N–S-striking faults in response to an oblique stress field (“trailing segment reactivation” of Nixon et al., 2014). Although this model may explain the observed increase in throw along discrete E–W fault segments (Fig. 7a), with both N–S and E–W faults active simultaneously and accruing throw as a geometrically and kinematically linked system, it requires a pre-existing E–W fault that was subsequently reactivated as the trailing segment. This requirement is inconsistent with our seismic–stratigraphic evidence, which indicates that such a fault was not present prior to the Triassic (Figs. 2, 4).

We suggest that a more feasible hypothesis is that the distinctive, discrete throw increases relate to passive, post-formation lateral offset of the N–S-striking faults due to sinistral strike-slip motion along E–W-striking faults. Such a model envisages juxtaposition of the hangingwall and footwall of a N–S-striking fault across the E–W fault, with a geometric consequence being an increase in throw along the E–W fault in the absence of any dip-slip extension. This model does not require Triassic activity along E–W-striking faults, in agreement with the Early Cretaceous timing of initiation for the Fjerritslev north and south faults (Figs. 7b, 8b). The local increases in throw would correspond to the difference in elevation between the hangingwall and footwall of the N–S-striking faults, which is essentially the throw accumulated on the N–S-striking faults during the Triassic (Figs. 7a, 10). This model is supported by three observations. First, NS1 and NS2, and their respective hangingwall faults (HF1 and HF2), are geometrically similar, with HF1 and HF2 both showing folding and only minor offset of the acoustic basement surface (Figs. 5, 9). Second, NS1 and NS2, along with HF1 and HF2 located within their respective hangingwalls, are laterally offset by approximately 10 km (Figs. 5, 9). Finally, NS1 and NS2 display similar values of throw at their branch lines with the Fjerritslev north fault (∼ 1000 ms TWT) to the discrete increase in throw that occurs between the intersections (Fig. 7a). This is consistent with the increase in throw being equivalent to the throw accrued along N–S faults during the Triassic.

Further evidence for sinistral strike-slip activity is observed at the northern margin of the basin, along the E–W-striking Farsund north fault, implying offset of the NS1 and west Varnes Graben (WVG) fault (Figs. 1c, 10). A shallow footwall and deeper hangingwall, related to Early Cretaceous extensional activity, are identified along the Farsund north fault, despite it being associated with a complex zone of deformation (Fig. 11). A key observation we make is that Carboniferous–Permian and Triassic strata are preserved in the footwall of the Farsund north fault, yet are eroded by the BJU in its hangingwall (Fig. 11). The footwall of the E–W Farsund north fault also corresponds to the hangingwall of the N–S-striking WVG fault, whereas the hangingwall of the Farsund north fault corresponds to the footwall of the NS1 fault (Figs. 10c, 11). We propose that, in the same way as we interpret to the south, the N–S-striking WVG and NS1 faults initially represented a single, through-going structure, with erosion along its footwall occurring due to the BJU. This fault was then offset by ∼ 10 km of sinistral strike-slip motion. This strike-slip activity juxtaposed the original hangingwall and footwall of the N–S fault across the Farsund north fault, resulting in the complex and perhaps somewhat unusual structural and stratal geometries we observe (Fig. 11).

Having presented geometric observations showing strike-slip motions along E–W-striking faults within the Farsund Basin, we now determine the timing of this activity. Based on the geometric and kinematic evidence outlined above (Figs. 5, 7, 9, 11), we propose that the WVG, NS1 and NS2 faults initially formed a singular N–S-striking fault during the Triassic, with HF1 and HF2, and HF2 and NS3 forming further, through-going, N–S-striking faults at this time. These faults were then offset along a series of E–W-striking, sinistral strike-slip faults, which appear to, in places, follow the location of, and may represent precursors to, the present-day E–W-striking faults (Fig. 10b). This strike-slip activity must have occurred after the extensional activity along the N–S-striking faults during the Triassic (Fig. 6). Relatively smooth throw profiles and low displacement gradients along the Fjerritslev north and south faults on supra-BJU horizons (top Jurassic and ILC; Figs. 7, 8), and the relatively isopachous Jurassic interval across the Farsund north fault (Fig. 11), indicates that any strike-slip activity must have occurred prior to the deposition and preservation of Jurassic strata (Figs. 7, 8), most likely during the Early–Middle Jurassic, a period of either non-deposition or erosion by the BJU. More precise constraints on this age and evidence for any deformation associated with the activity are not possible given erosion along the BJU.

Following Triassic extensional activity along N–S faults and strike-slip activity during the Early to Middle Jurassic, E–W-striking faults were active during the Late Jurassic – Early Cretaceous (Figs. 3, 6). Here, we detail the geometric and kinematic behaviour of the Fjerritslev north and south faults in response to this tectonic event.

We have determined the geometric and kinematic evolution of upper-crustal faults within the Farsund Basin. Triassic N–S-striking faults were offset by a series of E–W-striking, sinistral strike-slip faults during the Early–Middle Jurassic (Figs. 10a, b, 15a). Strike-slip systems in nature and in experimental models often display anastomosing, duplex geometries, where strain is distributed along a series of discrete structures that link in both strike and dip directions (e.g. Chamberlain et al., 2014; Cheng et al., 2017; Corti and Dooley, 2015; Dooley and Schreurs, 2012; Mann, 2007; Naylor et al., 1986; Richard et al., 1995; Scholz et al., 2010; Schreurs, 2003; Vauchez and Tommasi, 2003; Wu et al., 2009). However, due to BJU erosion, we can only confirm prior strike-slip activity between the laterally offset N–S-striking faults, i.e. NS1 and NS2 (Figs. 7, 10), HF2 and NS3 (Figs. 8, 10), and WVG and NS1 (Figs. 10c, 11). Where we know strike-slip faulting occurred, i.e. between the offset N–S faults, the strike-slip faults correspond to the same location as Early Cretaceous normal faults, i.e. the Fjerritslev north and Farsund north faults (Fig. 15b, c), observed at the present day, potentially indicating that these later extensional faults reactivated the prior strike-slip ones. However, the Fjerritslev south fault west of HF2 cannot have reactivated an older strike-slip fault. The strike-slip fault offsetting HF2 and NS3 does not share the same location as the Fjerritslev south fault west of HF2, as the lack of offset of NS2 would imply an unrealistically steep displacement gradient (Peacock and Sanderson, 1991; Fig. 15b). Instead, we propose that, in this location, the strike-slip fault strikes NW–SE, following the location of HF2 (Fig. 15b), and joining with the strike-slip fault between NS1 and NS2 (later reactivated as the Fjerritslev north fault).

Fault displacement backstripping (Figs. 7, 8) and seismic stratigraphic observations (Figs. 2, 4) indicate an Early Cretaceous onset of extension for the E–W-striking faults. This is consistent with our interpretation of Early–Middle Jurassic strike-slip activity along E–W-striking faults, with these faults experiencing no dip-slip motions prior to the Early Cretaceous. Apparent Triassic throw along the central segment of the Fjerritslev north fault and the eastern segment of the Fjerritslev south fault occurred through the passive juxtaposition of different structural levels across the fault, as opposed to active dip-slip faulting. This study shows that, particularly in areas of non-collinear faulting and oblique stress fields, care needs to be taken when interpreting fault kinematics from T-x profiles alone. In such instances, further lines of evidence, such as the presence of synkinematic strata, are required to confirm extensional fault activity.

En-echelon fault segmentation, such as that observed along the Fjerritslev north fault at the ILC structural level (Fig. 12), may form due to the oblique reactivation of a pre-existing fault (Giba et al., 2012; Grant and Kattenhorn, 2004; Lăpădat et al., 2016; Withjack et al., 2017). In the case of the Fjerritslev north fault, this is likely to be the pre-existing strike-slip fault (Figs. 10c, 15). Conversely, the Fjerritslev south fault, which was also active during the Early Cretaceous, is more linear and not obviously segmented. In addition, its western segment does not have the same strike as the pre-existing strike-slip fault (Fig. 15b). The Fjerritslev south fault instead appears to propagate westwards away from the prior strike-slip fault in the east as a newly formed structure, rather than simply reactivating a pre-existing one – hence the difference in structural style between it and the Fjerritslev north fault (Figs. 8, 15). The Farsund north fault, which also shows only Early Cretaceous extensional activity (Figs. 2, 11), appears to show a complementary eastwards propagation along the northern margin of the basin (Fig. 15c). Relay ramps along the Fjerritslev south fault appear to correspond to the intersections with the N–S-striking faults (TF1-4). These N–S-striking faults were active during the Triassic (Fig. 6a) and were not offset during Early Jurassic sinistral strike-slip activity (Fig. 15b). Therefore, as the Fjerritslev south fault was initiated as a new structure and propagated westwards from HF2 during the Early Cretaceous (Fig. 15c), the presence of the pre-existing TF1-4 may have segmented the newly forming Fjerritslev south fault (Fig. 12). The geometry and kinematic history of the Fjerritslev North fault during the Early Cretaceous, in particular the development of en-echelon segmentation and formation of a major depocentre in the more optimally oriented eastern fault segment, indicate that although dominant fault motion within the basin is extensional during the Early Cretaceous, there is a significant component of oblique slip (e.g. de Paola et al., 2005; Grant and Kattenhorn, 2004; Naylor et al., 1986; Richard et al., 1995; Fig. 15c). In the east of the basin, rapid subsidence may locally have been accentuated by salt mobilisation (Christensen and Korstgård, 1994).

We have determined the kinematic history of upper-crustal faults during multiple tectonic events, which we propose link to, and reflect activity along, the STZ at sub-crustal depths. Here, we examine the driving forces behind these tectonic events. Due to the fixed E–W orientation of the upper-crustal expression of the STZ relative to later tectonic events, we are able to use the observed reactivation style to infer the prevailing regional stress field at that time.

During the Carboniferous–Permian, dextral transtension and transpression occurred on a series of NW-trending faults along the STZ and TTZ (Fig. 16a; e.g. Erlström et al., 1997; Liboriussen et al., 1987; Michelsen and Nielsen, 1993; Mogensen, 1994), including along the Fjerritslev fault system to the east of the basin (Hamar et al., 1983; Mogensen, 1994; Skjerven et al., 1983). We document no such activity in the Farsund Basin (Figs. 2, 4). Sinistral motion is also proposed along the Skagerrak Graben at this time (Fanavoll and Lippard, 1994; Lie and Husebye, 1994; Mogensen, 1994), which, when combined with dextral motion along the STZ, results in net E–W extension, accommodating N–S compression from the Variscan Orogeny to the south (Fig. 16a). As such, we may not expect any extensional activity along the E–W faults within the Farsund Basin. N–S-striking faults in and around the Farsund Basin do not appear to be related to the STZ; instead, these faults appear to form the northern continuation of the Horn Graben, which formed in response to Permian–Triassic E–W-directed extension (Nielsen, 2003; Vejbæk, 1990).

Throughout the Mesozoic, both sinistral (Jones et al., 1999; Liboriussen et al., 1987; Norling and Bergström, 1987; Pegrum, 1984; Sivhed, 1991) and dextral (Bergerat et al., 2007; Erlström et al., 1997; Graversen, 2009; Hansen et al., 2000; Michelsen and Nielsen, 1993; Mogensen, 1995; Mogensen and Korstgård, 2003) strike-slip and oblique motions have been documented at various locations along the STZ. Sinistral strike-slip activity along an E–W-trending structure, such as observed during the Early–Middle Jurassic (Fig. 16c), could be driven by regional E–W- to NW–SE-oriented extensional stress fields. Potential regional driving mechanisms during the Early–Middle Jurassic include (i) localised extension north of the STZ within the Skagerrak Graben (Ro et al., 1990; Vejbæk, 1990), (ii) regional stresses relating to the inflation of the middle North Sea thermal dome (Rattey and Hayward, 1993; Underhill and Partington, 1993) or (iii) far-field stresses relating to the opening of the Atlantic and Tethyan oceans to the south (Vissers et al., 2013; Ziegler, 1990; Ziegler and Stampfli, 2001). The Skagerrak Graben was inactive throughout the Jurassic and was furthermore decoupled from the STZ at this time (Ro et al., 1990); accordingly, localised extension along the Skagerrak Graben could not have driven the observed strike-slip activity. The influence of the middle North Sea thermal dome was felt in the Farsund Basin during the Early–Middle Jurassic and associated stresses would be localised south of, and potentially buffered by, the STZ (Underhill and Partington, 1993; Fig. 16c). Although it is difficult to envisage how stresses arising from broad, regional inflation result in strike-slip activity, this still may represent a possible explanation, particularly if the uplift is buffered at the STZ (Fig. 16c). A further possibility is that the strike-slip activity is related to far-field stresses transmitted along the Tornquist Zone associated with ocean formation to the south (e.g. Coward et al., 2003; Vissers et al., 2013; Ziegler and Stampfli, 2001).

Around the Early–Middle Jurassic, the incipient Piemont–Liguria Ocean was situated south of the STZ, between the northwards subducting Tethys to the east and the central Atlantic to the west (Coward et al., 2003; Vissers et al., 2013; Ziegler, 1990; Ziegler and Stampfli, 2001). Seafloor spreading within the Piemont–Liguria Ocean was initiated at around 170 Ma, during the Middle Jurassic, with the ocean basin and prior rifting buttressed to the east at the TTZ (Vissers et al., 2013). Continental extension occurred during the Early–Middle Jurassic, prior to seafloor spreading, and was associated with sinistral strike-slip motion along the TTZ (Vissers et al., 2013; Fig. 16). Compressional forces arising from the subduction of the Tethys to the west may have also led to sinistral strike-slip motions being transmitted along the NW-trending Tornquist Zone, potentially reaching the Farsund Basin (see Fig. 7 in Ziegler, 1990; Fig. 16c). The Tornquist Zone may have acted as a proto-transform structure, representing an ultimately failed link between the North Atlantic rift system to the northwest and the Tethys to the southeast, similar to the transform fault between the Iberian Peninsula and central Europe (see Fig. 7 in Vissers et al., 2013; Fig. 16c). One implication of this model is the occurrence of sinistral motion along the whole of the Tornquist Zone at this time, a requirement not necessitated by the thermal dome hypothesis (Fig. 16c). Based on our data, we are unable to comment on whether this activity occurred along the structure to the east, although some sinistral activity is observed to the west (Jones et al., 1999; Pegrum, 1984; Skjerven et al., 1983).

Dextral transtensional activity occurred within the Farsund Basin during the Early Cretaceous (Figs. 12, 13, 15), as observed elsewhere along the STZ at this time (Bergerat et al., 2007; Michelsen and Nielsen, 1993; Mogensen and Jensen, 1994). Based on the easterly trend of the STZ in this location, the driving regional stress field could be extension and orientated approximately E–W to NE–SW (Fig. 15). Regional extension occurred across the North Sea during the Late Jurassic – Early Cretaceous and was oriented E–W to NE–SW in the central and southern North Sea (Coward et al., 2003; Rattey and Hayward, 1993; Underhill and Partington, 1993), and E–W to NW–SE in the northern North Sea (Bartholomew et al., 1993; Bell et al., 2014; Brun and Tron, 1993). Oblique dextral reactivation of E–W-striking faults indicates that NE–SW-to-E–W extension, associated with activity within the Central Graben, was prevalent at this time within the Farsund Basin (Fig. 16d).

We have shown that the upper-crustal part of the STZ was repeatedly reactivated in a range of different tectonic styles, which we have linked to regional tectonic events, prevailing stress fields, and potentially broader geodynamic context (Fig. 16). Here, we discuss how such a contrast in lithospheric thickness and properties at depth (e.g. Cotte and Pedersen, 2002; Hossein Shomali et al., 2006; Kind et al., 1997) is able to influence rift development within the upper crust.

Changes in lithospheric thickness associated with prior phases of rifting or different continental blocks have previously been shown to influence the development of rift systems (Autin et al., 2013; Brune et al., 2017; Corti, 2009). In addition, numerous studies show that strike-slip and oblique fault systems can dissect the whole lithosphere and are often associated with pervasive fabrics within mantle lithosphere (Tommasi and Vauchez, 2001; Vauchez and Tommasi, 2003; Wylegalla et al., 1999). Examples from the Great Glen Fault, UK (Klemperer and Hobbs, 1991; McBride, 1995), the Transbrasiliano fault zone, onshore Brazil (Daly et al., 2014), and the San Andreas fault system, USA (Chamberlain et al., 2014), represent crustal-terrane separating fault systems that extend down to at least the base of the crust and, in many cases, into the sub-crustal lithosphere (Vauchez and Tommasi, 2003). These lithosphere-scale structures are often oriented oblique to regional tectonic events and are thus subjected to oblique stresses; as such, they are often reactivated in a transpressional or transtensional manner, resulting in complex rifts at upper-crustal levels (e.g. Bergerat et al., 2007; Calignano et al., 2017; Cheng et al., 2017; Corti and Dooley, 2015; Holdsworth et al., 2001; Le Breton et al., 2013; Underhill and Brodie, 1993; Vauchez and Tommasi, 2003).

The E-trending Farsund Basin and STZ represent a further example of a lithosphere-scale structure (Fig. 2). Lassen and Thybo (2012) propose the STZ formed as an Ediacaran rift system during the breakup of Rodinia. Montalbano et al. (2016) note a sinistral phase of motion along the present-day trend of the STZ around this time, indicating that the structure must have existed in some form at this time. Although the STZ appears to have been present in some form since the Late Proterozoic (Lassen and Thybo, 2012; Montalbano et al., 2016), its upper-crustal component appeared to have been established during the Carboniferous–Permian as a lithosphere-scale strike-slip system, exploiting the change in lithospheric thickness and properties at depth (Erlström et al., 1997; Lassen and Thybo, 2012). The localised reactivation of this structure appears to buffer the cratonic lithosphere of Baltica against the regional tectonic events that affect the amalgamated terranes of central Europe (see Berthelsen, 1998; Mogensen and Korstgård, 2003). Berthelsen (1998) proposed that the Fjerritslev Fault system forms a NE-dipping detachment that shields the cratonic lithosphere of Baltica. Although our data are unable to confirm details of this hypothesis, we do note a connection between the sub-crustal and upper-crustal components (Fig. 2), which may concentrate deformation within the STZ, thus leaving the cratonic lithosphere of Baltica relatively undeformed. As the orientation of this lithosphere-scale structure is fixed relative to the later tectonic events (Babuška and Plomerová, 2004; Cotte and Pedersen, 2002), yet is weak enough to be repeatedly reactivated throughout these events, seemingly regardless of orientation, the style of reactivation of the upper-crustal component can be inverted to better understand the regional stress field that has affected the North Sea throughout these multiple tectonic events.