Rivers, reefs and deltas: geomorphological evolution of the Jurassic of the Farsund Basin, offshore southern Norway

In many petroleum-bearing, data-poor ‘frontier’ basins, source, reservoir and seal distribution is poorly constrained, making it difficult to identify petroleum systems and play models. However, 3D seismic reflection data provide an opportunity to directly map the 3D distribution of key petroleum system elements, thereby supplementing typically sparse, 1D sedimentary facies information available from wells. Here, we examine the Farsund Basin, an underexplored basin offshore southern Norway. Despite lying in the mature North Sea Basin, the Farsund Basin contains only one well; meaning there remains a poor understanding of its hydrocarbon potential. This east-trending basin is anomalous to the north-trending basins present regionally, having experienced a different tectonic, and most likely geomorphological, evolution. We identify a series of east-flowing rivers in the Middle Jurassic, the distribution of which are controlled by salt-detached faults. In the Middle Jurassic, a series of carbonate reefs, expressed as subcircular amplitude anomalies, developed. Within the Upper Jurassic we identify numerous curvilinear features, which correspond to the downlap termination of southwards-prograding deltaic clinoforms. We show how seismic-attribute-driven analysis can determine the geomorphological development of basins, offering insights into both the local and regional tectonostratigraphic evolution of an area, and helping to determine its hydrocarbon potential.

Successful hydrocarbon exploration requires a knowledge of how sedimentary basins evolve in terms of tectonic events and depositional patterns, and how this leads to the development of working petroleum systems and plays (e.g. Johannessen & Andsbjerg 1993;Posamentier 2004;Dreyer et al. 2005;Holgate et al. 2013). Predicting the occurrence of source, reservoir and source rocks is a critical element of petroleum systems analysis, although this is especially challenging in frontier or underexplored basins due to a lack of well data. Even where available, these well data provide only 1D to quasi-3D constraints on the distribution of various rock types in the subsurface. Outcrop studies may provide detailed insights into the typical geometries and spatial relationships between different facies types, but they do not replicate specific subsurface geometries (e.g. Prélat et al. 2009;Romans et al. 2011;Agirrezabala et al. 2013;Holgate et al. 2014;Legler et al. 2014). Therefore, our understanding of the geological complexity of an area, including the distribution of different rock types and constituent petroleum systems, is heavily influenced by well spacing, reducing the accuracy and resolution of derived play maps and tectonostratigraphic models (e.g. Johannessen & Andsbjerg 1993;Dreyer et al. 2005;Holgate et al. 2013;Mannie et al. 2014). These issues are particularly relevant in underexplored areas where a complex area, with a relatively localized geological evolution, may not be incorporated into regional models due to a lack of data and low resolution.
High-quality 3D seismic reflection data do not always suffer from such spatial aliasing, providing relatively high-resolution images (tens of metres) of ancient geomorphic landscapes and seascapes preserved in the Earth's subsurface. By using amplitude-and frequency-derived seismic attributes, such as root mean square (RMS) amplitude, variance, dominant frequency and spectral decomposition, we are able to constrain, to varying degrees of confidence, the geometry and distribution of different rocks types within the subsurface (e.g. Ryseth et al. 1998;Posamentier & Kolla 2003;Posamentier 2004;Colpaert et al. 2007;Chopra & Marfurt 2008;Jackson et al. 2010;Zhuo et al. 2014;Klausen et al. 2016;Eide et al. 2017). Using 3D seismic reflection data, we can therefore examine the stratigraphic evolution of the subsurface over a greater areal extent than borehole data, albeit often at a coarser temporal resolution (e.g. Cartwright & Huuse 2005;Colpaert et al. 2007;Jackson et al. 2010;Klausen et al. 2016;Saqab & Bourget 2016). Seismic geomorphological analysis can therefore also provide additional context to the geological evolution of an area, helping to reduce exploration risk and to update regional tectonostratigraphic models in frontier or underexplored areas.
Here, we conduct a seismic-attribute-driven interpretation of 2D and 3D seismic reflection data located offshore southern Norway, and analyse the 3D facies architecture of the Triassic and Jurassic section preserved in the relatively underexplored Farsund Basin (Fig. 1). Data from well 11/5-1, located along the southern margin of the basin, provides independent constraints on the lithology of the depositional elements imaged within the 3D seismic volume (Fig. 1). Current palaeogeographical models within this area are largely based on regional borehole-correlation studies, with data primarily from the adjacent Egersund and Norwegian-Danish basins, and little within the Farsund Basin; these studies document a clastic-dominated net-transgressive setting throughout the Jurassic (Sørensen et al. 1992;Mannie et al. 2014Mannie et al. , 2016. However, the tectonic evolution of the Farsund Basin differed to that experienced by adjacent basins, being heavily influenced by activity along the underlying lithosphere-scale Tornquist Zone (Phillips et al. 2018). Here, we investigate whether this contrasting tectonic history Red circles represent wells throughout the area; note the high density of well coverage in the Egersund Basin compared to the Farsund Basin. Also shown are the locations of the 3D seismic survey referred to in this study, and the salt depositional limits proposed by ,  and this study. (b) Jurassic stratigraphic column. Lithostratigraphy from Mannie et al. (2016), focused on well data from the Egersund Basin, is contrasted against seismic data around well 11/5-1, located within the Farsund Basin. Note the lack of Triassic and Middle-Lower Jurassic strata in well 11/5-1.
Additional seismic interpretation was undertaken on a series of north-south-orientated 2D seismic sections ( Fig. 1), which image to 7 s TWT (c. 15 km) and allow interpretation of the stratigraphic horizons over a wider area, providing a more regional perspective to our interpretations. Seismic data are displayed as zero phase and follow the SEG normal polarity convention: that is, a downwards increase in acoustic impedance is represented by a peak (black), and a downwards decrease in acoustic impedance is represented by a trough (red) (Fig. 3). The ages of key stratigraphic horizons are constrained through a number of wells regionally (Fig. 1). Well 11/ 5-1, the only well located within the 3D seismic volume (Fig. 2), provides detailed 1D facies information within the study area and is tied to the seismic interpretations through a seismic well tie (Fig. 5). The synthetic seismogram proves a good fit to the seismic data in areas where both sonic and density logs are available, with a good match between the top Rotliegend and top Sandnes horizons, and the lower section of the Egersund Formation (Fig. 5). There is a poor fit throughout the Egersund and Tau formations that is primarily related to a lack of density log information in the interval 1130-1180 m. Areas immediately above and below this data-absent interval, and towards the top of the log, where data recording begins, are influenced by edge effects relating to data-absent areas and therefore poorly correlate with the original seismic trace (Fig. 5). Seismic attributes, such as RMS amplitude, variance, dominant frequency and spectral decomposition, were calculated within windows located above, below or between specific key horizons in order to further interrogate and extract information from the seismic data (see Appendix A for details regarding specific seismic attributes).
We used GeoTeric software in order to calculate the spectral decomposition attribute. To do this, we extracted a frequency spectrum from the data and split this into a series of discrete bins, each corresponding to a range of 10 Hz (Fig. 4). Frequency values centred on 22, 30 and 45 Hz were assigned to the colours red, green and blue, respectively, and blended to produce the spectral decomposition attribute (Fig. 4b).

Regional stratigraphy
We here describe the stratigraphic succession preserved in the Farsund Basin as revealed by well 11/5-1, and compare it to the succession encountered regionally (Figs 1, 2 and 5). The Triassic interval is a non-marine succession across the much of the North Sea (McKie & Williams 2009;Jarsve et al. 2014). Triassic strata are preserved within the footwalls of the NS1 and NS2 faults, although no Triassic strata are encountered in well 11/5-1, with Middle Jurassic strata unconformably overlying the Upper Permian Rotliegend Group due to erosion by the BJU (Figs 3 and 5). The Middle Jurassic (Bajocian-Bathonian) Bryne Formation is present regionally, but is also not present in well 11/5-1 (Fig. 5). The Bryne Formation in the adjacent Egersund Basin comprises non-marine sandstone and siltstone (Vollset & Doré 1984;Mannie et al. 2016). Stratigraphically above the Bryne Formation, and forming the lowermost Jurassic interval penetrated in well 11/5-1, is the Mid-Jurassic (Callovian) Sandnes Formation. In well 11/5-1, this comprises 45 m of predominantly marine sandstones and mudstones, with some intervals containing abundant metre-scale carbonate stringers (Fig. 5) (Vollset & Doré 1984;Mannie et al. 2016). The Late Callovian-Early Tithonian Egersund and Tau formations are situated stratigraphically above the Sandnes Formation. These formations make up the majority of the Jurassic interval in well 11/5-1 (189 m), and consist of organic-rich claystones and shales (Fig. 5). Isolated glauconitic and pyritic layers are present throughout these formations, indicating a low-energy depositional environment that was prone to anoxic conditions (Fig. 5) (Vollset & Doré 1984). Regionally, the uppermost Jurassic (Tithonian)-lower Cretaceous Sauda Formation consists of marine shales. However, in well 11/5-1, the upper 15 m of the 22 m-thick Sauda Formation (incorporating the lowermost Lower Cretaceous interval) is sandstone-dominated (Fig. 5). The Jurassic interval in both the Farsund Basin and regionally is overlain by a large thickness of Lower Cretaceous deep-water claystones and mudstones.

Frequency v. depth
Well 11/5-1 provides direct constraints, albeit only in one dimension, on the stratigraphy of the Triassic and Jurassic succession of the Farsund Basin. Using this stratigraphic framework, we now use a suite of seismic attributes (see Appendix A) to determine the 3D geometry and distribution of different facies types within the basin. In each subsection, we first describe our seismic geomorphological observations, based on seismic reflection, seismic attribute and well data analysis, before posing an interpretation for the likely depositional environment and geomorphological origin of the identified features.
Triassicthe limit of thin-skinned tectonics and the depositional extent of mobile Zechstein salt In the North Permian Basin, Zechstein salt is overlain by Triassic strata (Clark et al. 1998;Lewis et al. 2013). Post-depositional salt mobilization and modification means that the initial depositional limit of salt and salt basins is often uncertain (Clark et al. 1998;. Triassic strata in the Farsund Basin are dominated by south-dipping, salt-detached normal faults related to southwards-directed salt mobilization into the Norwegian-Danish Basin (Fig. 6). The top of the Triassic interval is eroded by the BJU, with Triassic strata largely absent across the footwalls of NS1 and NS2, north of the Fjerritslev South Fault (Figs 3 and 5). The Fjerritslev North and South faults in this area do not show any pre-Cretaceous activity and were not present during the Triassic, with the Fjerritslev South Fault appearing restorable up to the BJU (Figs 3 and 6) (Phillips et al. 2018).
Within the hanging wall of the Fjerritslev South Fault, a series of small-scale (c. 50 ms TWT, 70 m height) clinoforms are identified in the Triassic interval, prograding towards the south (Fig. 6). The relatively small height of these clinoforms, and the sub-aerial regional character of the Triassic (Jarsve et al. 2014), implies deposition within a terrestrial, probably fluvio-deltaic, environment (cf. Patruno et al. 2015a). Immediately south of these clinoforms, now located on the footwall of the Fjerritslev South Fault, thinskinned salt-detached faulting is restricted northwards (Fig. 6). We propose that the transition from the preserved clinoform sequence in the north to where the clinoforms are bisected by thin-skinned saltrelated faults in the south could indicate the initial northern depositional limit of the mobile component of the Zechstein salt (Fig. 6). North of this limit, mobile Zechstein salt, or salt of sufficient thickness to flow, is not present. We propose the following model: prior to salt mobilization, deltaic clinoform sequences prograded southwards over the Zechstein salt basin (Fig. 6). Following the onset of salt mobilization during the Triassic, areas with underlying Zechstein salt were subject to the formation of thin-skinned salt-detached faults, offsetting shallower strata, including the clinoform sequence. However, areas north of the initial depositional limit, where no salt was present or was unable to flow, were unaffected and preserve the southwards-prograding clinoform-bearing sequences (Figs 1 and 6). Using these criteria, we are able to map the original depositional limit of mobile Zechstein salt across the present-day Farsund Basin. The depositional limit strikes east-west across the footwall of the Fjerritslev South Fault, which had not formed at that time ( Fig. 1). Further east, the depositional limit of the Zechstein salt steps northwards across NS2, indicating that some relief may have been present across this fault during salt deposition, before continuing in an east-west orientation to the east (Fig. 1).

Bryne Formationfluvial systems
In the footwall of the Fjerritslev South Fault we identify two highamplitude, laterally discontinuous reflections above the BJU (Fig. 7). Additional features may be present elsewhere in the basin, but are not identifiable due to the overall high reflectivity through much of this succession. The high-amplitude reflections are situated in the hanging walls of thin-skinned, salt-detached faults (Fig. 7). Each feature is around 500 m wide and is imaged as a very high-amplitude, tuned seismic reflection, thus making it difficult to calculate their true thickness (Figs 3 and 7) (Widess 1973;Brown 2011). However, given that the vertical resolution of the data in this interval is c. 24 m (Fig. 4a), this value represents a maximum thickness estimate. Due to the restricted lateral extent of these reflections, and their stratigraphic position below the Sandnes Formation, we interpret that these features are not penetrated by well 11/5-1 and instead correspond to the Bryne Formation; thereby representing the oldest Jurassic strata in the Farsund Basin.
We extract RMS amplitude, variance and dominant frequency seismic attributes, calculated within a 25 ms TWT window above the top of the BJU horizon in order to encompass the full thickness of the features, highlight their 3D geometry and provide clues as to their geological origin (Fig. 7). In map view, the RMS amplitude attribute highlights two east-west-trending high-amplitude features, with curvilinear channel-like geometries, on the footwall of NS2, termed Channel 1 and Channel 2 from north to south, respectively ( Fig. 8). Channel 1 is c. 8 km long and can be traced westwards across the footwall of the Fjerritslev South Fault. It is not imaged in the hanging wall of the Fjerritslev South Fault, due to the amplitude signal being masked by higher background amplitudes (Fig. 8). Channel 2 originates within the footwall of the Fjerritslev South Fault, has an overall length of c. 9 km and widens eastwards from c. 200 to c. 400 m (Fig. 8). In cross-section, the channels display an asymmetrical geometry, with Channel 2 being thicker towards the south (Fig. 7). Both channels cross, and are not offset by, NS2 to the east (Fig. 8). However, the channels widen as they cross NS2, from c. 500 m in the footwall to c. 2 km in the hanging wall. Furthermore, the channels display a more SE orientation within the footwall of the fault (Fig. 8).
The variance and dominant frequency seismic attributes provide more detailed insights into the channel geometry. The variance attribute highlights a series of minor linear channels orientated perpendicular to, and joining along, both channel 1 and 2. These secondary channels display lengths and widths of c. 400 and c. 150 m, respectively (Fig. 8c). In some instances these minor channels link Channel 1 and Channel 2 ('linking channel' in Fig. 8e). The secondary channels display an asymmetrical distribution with respect to the main channel, being concentrated along one margin, which displays a relatively shallow gradient. The opposite margin of the channel is more sharply defined and is often associated with an underlying salt-detached fault (Figs 7 and 8). The dominant frequency attribute further defines the first-order geometry of the channels within the footwall of NS2, which are Based on the seismic-attribute-derived observations described above, we interpret these channel-like features as originally east-to SE-flowing Middle Jurassic fluvial systems within the Bryne Formation. The high-amplitude character of the channels indicates a different, perhaps more sand-prone, lithology to the Egersund and Sandnes formation mudstone above and potentially fine-grained lithologies below (Figs 5 and 7). The smaller structures merging along the margins of the main channels are interpreted as tributaries (Fig. 8e). The east-west orientation of the channels is partly controlled by underlying thin-skinned, salt-detached faults. The underlying faults appear associated with a more sharply defined channel margin that hosts relatively few tributaries, whereas the adjacent margin is associated with a gentler gradient and hosts numerous tributaries (Fig. 8e). This asymmetrical geometry indicates that, at least in the west, movement along these thinskinned faults influenced the depositional surface and channel physiography. The channels were unaffected by the Fjerritslev South Fault, as this was not present at the time. Towards the east, the channels widen across a topographical gradient located above NS2. This topographical gradient formed due to differential compaction of Triassic strata across the now inactive fault (Fig. 8). The widening of the channels appears to represent an estuarine setting, transitioning from a non-marine to a shallow-marine environment. This indicates that the location of NS2 is likely to have represented the palaeoshoreline during the deposition of the Bryne Formation in the Middle Jurassic.

Sandnes Formationpatch-reef development
Atop the Bryne Formation channel systems, a series of isolated high-amplitude features (IHAFs) are identified within a stratigraphic interval corresponding to the Middle Jurassic Sandnes Formation, which overlies the BJU and, where present, the Bryne Formation. As penetrated in well 11/5-1, the Sandnes Formation consists of sandstone and mudstone, with some isolated carbonate stringers also present. No IHAFs are directly penetrated by the borehole (Fig. 5).
In cross-section, the IHAFs display a double ( peak-trough) reflection character, with a large positive impedance contrast at the top (Fig. 9). Two distinct IHAF morphologies are identified: short, wide structures with heights of 25 ms TWT (c. 30 m), and taller, narrower structures with typical heights of 35 ms TWT (c. 50 m) (Fig. 9). The plan-view morphology of the IHAFs is further highlighted by seismic attributes (Fig. 10). RMS amplitude, variance and dominant frequency attributes were extracted from a 50 ms window above the BJU and also above the Bryne Formation channels, ensuring coverage of the full height of the structures and a lack of input from stratigraphically lower features. In map-view, the IHAFs are expressed as circular to subcircular high-amplitude anomalies, consisting of a highamplitude core and relatively low-amplitude margin (Fig. 10). A total of 333 IHAFs are identified across the area; smaller IHAFs are most accurately delineated using the spectral decomposition attribute (Fig. 10c). The larger structures have a diameter of c. 450 m; whilst the smaller structures have a diameter of c. 150 m (Figs 9 and 10). The tall, narrow IHAFs are predominately situated within the hanging wall of NS2, whereas the short, wide IHAFs are restricted to the footwall (Fig. 10d). Notably, the distributions of the two different morphologies are unaffected by the east-trending Fjerritslev North and Fjerritslev South faults that dominate the present-day basin morphology, but were not present during the deposition of the Sandnes Formation with the wider, shorter IHAFs situated on both the hanging walls and footwalls of the Fjerritslev North and South faults (Fig. 10). The distribution of the IHAFs does not change laterally to the south and west, indicating that the IHAF domain may extend outside of the 3D volume. However, the concentration of the IHAFs does decrease to the NE, in the hanging wall of both NS2 and the Fig. 5. Well-log information and synthetic seismic for the Jurassic interval of well 11/5-1. A synthetic seismic section was created using the RHOB and DT wireline logs of the well. A lack of density data at around 1130-1180 m results in a major discrepancy between the original and synthetic seismic data. Otherwise, the synthetic provides a good match to the original seismic at the Rotliegend Group and Sandnes Formation intervals, as well as the Sauda Formation.
Fjerritslev North Fault, implying that this may represent a limit to the IHAF domain ( Fig. 10).
In some instances the IHAFs are cross-cut by later faults ( Fig. 10d and f ), implying that they are brittle in nature. Some IHAFs display non-rounded, more elongate geometries, which RMS amplitude shows are typically a result of these IHAFs containing multiple high-amplitude nuclei. The typical subrounded morphology of the IHAFs implies a radial mode of growth, with those IHAFs that Fig. 6. Uninterpreted and interpreted seismic section showing the northwards limit of thin-skinned, salt-detached faulting marking the northern depositional limit of mobile salt. Note that an undeformed clinoform interval progrades southwards before being offset by thin-skinned faulting to the south. See Figure 2 for the location. contain multiple nuclei representing IHAFs that have grown radially and since merged (Fig. 10e).
A variety of different processes can lead to the formation of subcircular structures in seismic reflection data (Stewart 1999), including volcanic edifices (both igneous and mud-related) (Davies & Stewart 2005), hydrothermal vent systems (Magee et al. 2016), gas accumulations and pockmarks (Hovland et al. 1987;Fichler et al. 2005 (Jackson & Talbot 1986). Based on the relatively small (hundreds of metres) scale of the structures, coupled with a lack of igneous activity within this area at this time, we discount an igneous-/volcanic-edifice-related origin for the IHAFs. Similarly, the small scale of the IHAFs and the lack of regional igneous activity also discount an origin as hydrothermal vent systems (Magee et al. 2016). In addition, we do not consider an evaporate-related origin based on the IHAFs being located stratigraphically above the Upper Permian Zechstein salt, as there is no Jurassic salt present in this area of the North Sea . Furthermore, the IHAFs are also present north of the aforementioned depositional limit of the Zechstein salt (Figs 1 and 6). The relatively small-scale nature of the structures would be consistent with an origin as pockmarks; however, these structures are associated with positive relief, whereas pockmarks would typically form cavities infilled with material from overlying strata (Hovland et al. 1987;Agirrezabala et al. 2013;Kluesner et al. 2013;Marcon et al. 2014).
Therefore, based on: (i) their radial growth mode; (ii) the positive impedance contrast at the top of the structures; (iii) their overall size and morphology, along with the binary nature of the size distribution potentially reflecting different growth conditions; and (iv) their brittle nature, we interpret that the IHAFs represent a series of carbonate patch reefs. Carbonate is present locally, as demonstrated by the carbonate-rich intervals penetrated in well 11/ 5-1 (Fig. 5). Modern-day carbonate patch reefs are typically found within shallow-marine environments and are often associated with sheltered low-energy environments. Modern patch reefs have diameters of c. 200 m, and heights of c. 10 m, similar to those within the study area (e.g. Brock et al. 2008;Purkis et al. 2015). The binary size distribution of the IHAFs may reflect growth conditions in slightly different, albeit still shallow, water depths. The IHAFs in slightly deeper water grow preferentially upwards towards the sea surface, compared to the wider, shorter patch reefs located in shallower waters, which have limited accommodation space and therefore grow laterally (Kendall & Schlager 1981;Schlager 1981

Egersund and Tau formationsdeposition of anoxic shales
The Sandnes Formation is overlain by the Upper Jurassic Egersund and Tau formations (Figs 3 and 5). As determined from boreholes regionally, including well 11/5-1 in the Farsund Basin, these formations typically comprise organic-rich shales (Fig. 5). Within the Farsund Basin, the Tau Formation has a slightly elevated gamma-ray value (c. 120 API) when compared to the underlying Egersund Formation (c. 110 API) (Fig. 5). Both formations are associated with a poorly reflective seismic facies within the basin (Figs 3 and 5). Based on the observations outlined above, we interpret that the deposition of both the Egersund and Tau formations occurred in a low-energy environment, with the presence of pyritic and glauconitic horizons suggesting periodic anoxic conditions (Fig. 5). Such an environment may indicate a sea-level rise and marine transgression since the deposition of the Sandnes Formation, with deposition occurring in a deep-marine environment, or may alternatively indicate deposition within a restricted, low-energy shallow-water environment (Van Der Zwaan & Jorissen 1991).

Sauda Formationdelta progradation
A package of high-amplitude reflections is present at the top of the Jurassic interval, corresponding to the Sauda Formation in well 11/ 5-1 (Figs 3 and 11). This reflection package is c. 40 ms TWT (c. 50 m) thick, thins southwards, and is associated with lateral changes in amplitude and low-angle clinoform sequences that downlap towards the south (Fig. 11). These downlap terminations often correspond to the areas of amplitude brightening. The top and base of the high-amplitude reflection package was mapped throughout the 3D volume, along with individual internal horizons. Seismic attributes were extracted from between these top and base horizons (Fig. 11).
RMS amplitude, spectral decomposition and dip azimuth seismic attributes highlight a series of divergent, curvilinear lineations in plan-view, defining high-and low-amplitude packages of varying frequency (Fig. 12). Each band is c. 400 m wide, diverges westwards and displays a concave-to-the-south planform geometry. The bands are arranged into a series of discordant sets and thus truncate each other, at either low (i.e. sets 1-3 in Fig. 12d) or high angles (i.e. Set 4 in Fig. 12d). Additional internal sets and  truncations may tentatively be present, although these are not clear and accurately delineated within the data (Fig. 12). The lineations are ubiquitous across the whole of the study area, apart from across the footwall of the present-day Fjerritslev South Fault where the Sauda Formation is absent due to erosion across the footwall of the fault (Figs 2 and 12). The lineations are seemingly unaffected by the Fjerritslev North Fault, which cross-cuts but does not noticeably offset the lineations (Fig. 12). Both the Fjerritslev North and Fjerritslev South faults were not present at this time, becoming active in post-Sauda times. A set of north-south-striking lineations (Set 4) are present to the NW of the area and appear to correspond to the footwall of NS1, which was inactive at this time (Fig. 12). These lineations are concave to the east and diverge to the SW. These north-south-striking lineations also appear unaffected by the Fjerritslev North Fault, although they align with and are located along the footwall of NS1, suggesting this Triassic fault may have had some topographical expression at this time (Fig. 12d). Faint lineations are identifiable in the SE (Fig. 12); these lineations are not associated with the Sauda Formation, which thins NW of the area, but instead are related to subcrops of truncated underlying strata associated with salt mobilization in this area. In some instances there appear to be mutual cross-cutting relationships between individual sets of lineations (i.e. Set 2 and Set 4 in Fig. 12d), rather than truncations against one another. However, these cross-cutting relationships are most likely to be due to signal mixing within the attribute extraction window: that is, closely superposed sets produce cross-cutting relationships, meaning that their relative ages cannot be distinguished in plan-view. The prominent lineations observed in plan-view (Fig. 12) appear to correspond to the downlap terminations of clinoform sequences, or where the clinoform sequences thin below seismic resolution in cross-section (Fig. 11).
Discordant sets of high-amplitude lineations identified in seismic reflection data, superficially similar to those observed here, have previously been interpreted as ancient shoreface beach ridge environments (Jackson et al. 2010;Klausen et al. 2015Klausen et al. , 2016. Such beach ridge systems typically comprise sand-rich ridges separated by elongate, typically lower-energy, depressions (Otvos 2000), and form cuspate, concave-to-coastline morphologies comprising multiple discordant sets, formed through longshore drift (Billy et al. 2014;Vespremeanu-Stroe et al. 2016). Although geometrically similar, based on the lines of evidence outlined below, we discount a beach ridge origin for the lineations within the Farsund Basin. Firstly, the amplitude changes associated with the upper Jurassic curvilinear features within the Farsund Basin are located downdip, or below the small-scale clinoform sequences (Fig. 11), rather than being associated with the topsets as would be expected with a beach ridge interpretation (e.g. Jackson et al. 2010;Billy et al. 2014). Instead, the observed amplitude brightening may represent the tuned seismic response of the breakpoint (Dreyer et al. 2005) or downdip terminations of the clinoforms themselves (Eide et al. 2017). Secondly, the concave-to-south planform geometries of the lineations would suggest that the palaeoshoreline was north-facing; an interpretation largely incompatible with the regional setting of the basin during the Late Jurassic, which was open to the south (Figs 1  and 12). An additional and final argument against a beach ridge origin for these lineations lies in their regional context. Using regional 2D seismic data, the high-amplitude lineations can be traced outside of the 3D volume. Here, they correspond to the lateral terminations of larger, lobate, high-amplitude packages that thicken northwards (Figs  13 and 14). Overall, these packages are retrogradationally stacked, with clinoforms in younger packages downlapping onto the tops of older, underlying clinoform packages (Fig. 13), causing the overall Fig. 13. Uninterpreted and interpreted north-south-orientated regional 2D seismic section across the Farsund Basin. See Figure 1 for the location. The area also imaged in the 3D volume is situated to the south. A series of deltaic systems are identified prograding southwards, and aggrading and stacking atop one another, as evidenced by downlap terminations. The base of the Sauda Formation is marked as a reference horizon by a blue dashed line on the uninterpreted section. thickness of the Sauda Formation to increase northwards. This retrogradational stacking pattern suggests deposition during a period of relative sea-level rise, at a time when the rate of accommodation generation outpaced sediment accumulation rates.
Individual clinoform-bearing packages are also partitioned laterally, forming a series of discrete lobes that are identified via concordant lineations and terminal downlapping reflections around their margins. The lateral terminations are also associated with an increase in amplitude as the package thins below seismic resolution and, we suggest, constructively tune (Fig. 14). Based on these terminations we identify four main lobes, with the easternmost margin of the eastern lobe defined by an area of amplitude brightening (Fig. 14).
The lobes form an arcuate geometry in plan-view (Fig. 15). Individual lobes prograde southwards, are c. 100 ms TWT (c. 140 m) thick and are 20-40 km wide (Figs 14 and 15). They continue northwards into the Varnes Graben, which may represent a sediment fairway from the mainland (Fig. 13), although a lack of data coverage in this area means their geometry cannot be constrained. As with the lineations identified in the 3D volume ( Fig. 12), individual lobes identified on the 2D data display discordant relationships with one another as younger lobes overlap and stack above and adjacent to older lobes (Figs 14 and 15). The central lobe (Lobe 3) corresponds to the major lineations observed within the 3D volume (sets 1 and 2: Fig. 12). This lobe is situated at shallower stratigraphic levels and appears to overlap Lobe 2 situated to the east (Fig. 15). Set 3 within the 3D volume may also represent an older, stratigraphically deeper lobe (Lobe 1: Fig. 15) that has been covered by lineation sets 1 and 2 of Lobe 3 (Figs 12 and 15). Lobe 3 appears to be overlain by Lobe 4 to the west, which incorporates the lineations observed along the footwall of NS1 (Set 4: Fig. 12d), as the thickness of the Upper Jurassic Sauda Formation increases westwards (Fig. 14). This thickness change is representative of increased lobe stacking to the north and west (Figs 13 and  15). We suggest that clinoform downlap terminations within individual lobes may give rise to the lineation sets observed in the 3D data (Fig. 12). Furthermore, the overprinting and vertical stacking of different lobes may give rise to the discordant Uninterpreted and interpreted east-west-orientated regional 2D seismic section. See Figure 1 for location. Three discrete deltaic lobes can be identified across the area, with lateral downlap terminations observed either side. These lobes appear to thicken towards the west. The base of the Sauda Formation is marked as a reference horizon by a blue dashed line on the uninterpreted section. truncations of the lineation sets observed in the 3D data, with older lobes being partially overlapped by younger ones (Fig. 15).
Based on this regional information and in conjunction with the evidence outlined previously, we interpret that the lineations within the Upper Jurassic Sauda Formation correspond to the downlap termination of clinoforms within a series of deltaic lobes. The lobes prograded into an unconfined basin setting with lobe avulsion causing them to stack laterally and vertically (Bridge & Leeder 1979;Jones & Schumm 1999). This interpretation is based on: (i) the lobate geometry of the individual sequences (Figs 14 and 15); (ii) lateral downlap terminations at the margins of individual lobes (Figs 13 and 15); (iii) small-scale clinoforms indicative of deposition within a relatively shallow-water environment (Patruno et al. 2015a;Eide et al. 2017) (Figs 11 and 13); (iv) retrogradational stacking of individual lobes indicating that the generation of accommodation outpaced the sediment accumulation rate, potentially during a period of relative sea-level rise (Fig. 13); and (v) progressive landwards onlap of the Sauda Formation by Lower Cretaceous strata (Figs 13 and 14). The Sauda Formation overlies the Egersund and Tau formations (Fig. 5), which were deposited within an anoxic environment. A corollary of the interpretation here is that these anoxic shales were likely to have been deposited within a shallow-marine environment (Van Der Zwaan & Jorissen 1991), rather than in deep water, as the latter would require a drastic shallowing between the deposition of the two formations.

Discussion
Our model for the geomorphological evolution of the Farsund Basin, and our interpretations of the different facies types present (Fig. 16), differ drastically from regional tectonostratigraphic models of the area (Fig. 1b) (Johannessen & Andsbjerg 1993;Andsbjerg 2003;Mannie et al. 2014Mannie et al. , 2016. Here, we first compare and contrast our model for the geomorphological evolution of the Farsund Basin outlined above to that more regionally, before discussing the implications for the structural evolution of the basin and regional tectonic activity, and the viability of petroleum systems in the area.

Regional palaeogeographical setting
The North Sea represented a predominately non-marine environment during the Permian, as recorded by deposition of the Rotliegend Group (Glennie 1997;van Wees et al. 2000;Glennie et al. 2003). Deposition of Zechstein salt in the Upper Permian occurred during a marine transgression and basin flooding (Glennie 1997;Glennie et al. 2003). East of the Farsund Basin,  defined the depositional limit of mobile Zechstein salt striking ESE across the Lista Nose fault blocks (Fig. 1). We find that this limit continues eastwards along-strike across the southern margin of the Farsund Basin, and did not extend north as far as previously described   (Fig. 1). The local occurrence of small-scale, probably fluvio-deltaic, clinoforms (Patruno et al. 2015a) (Fig. 6), and the non-marine Smith Bank and Skagerrak formations regionally (e.g. Goldsmith et al. 1995;McKie & Williams 2009;Jarsve et al. 2014), indicate a return to a subaerial environment during the Triassic. The Triassic clinoforms were likely to have been sourced from mainland Scandinavia to the north, and probably prograded southwards through the Varnes Graben (Fig. 1).
Following Early-Mid-Jurassic uplift, erosion and eventual deflation associated with the Mid North Sea thermal dome (Underhill & Partington 1993), the first formation to be preserved regionally was the Middle Jurassic Bryne Formation. This formation is encountered in the Egersund Basin, Norwegian-Danish Basin and the Danish Central Graben, where it consists of stacked fluvial and floodplain deposits deposited in a coastal-plain environment (Sørensen et al. 1992;Johannessen & Andsbjerg 1993;Andsbjerg 2003;Michelsen et al. 2003;Mannie et al. 2014Mannie et al. , 2016. Within the Farsund Basin, the Bryne Formation is represented by two east-trending channels deposited within a fluvio-deltaic environment (Fig. 8e). At a regional scale, the eastwest orientation of these channels may be influenced by the Mid-Jurassic thermal dome (Underhill & Partington 1993), flowing away from the site of maximum uplift. Alternatively, and in the authors view more likely, channel orientation may be controlled by more local uplift related to the Lista Nose fault blocks and Stavanger Platform to the west (Fig. 1).
The Middle-Upper Jurassic (Callovian) Sandnes Formation documents a basin-wide marine transgression, transitioning from a subaerial to shallow-marine depositional environment, as observed elsewhere within the North Sea (Michelsen et al. 2003;Mannie et al. 2014Mannie et al. , 2016. This transgression was driven by a eustatic sea-level rise (Vail & Todd 1981;Sørensen et al. 1992), and may have been further augmented by rift-related thermal subsidence relating to a Permian-Triassic rift phase (Ziegler 1992). Within the Farsund Basin, the Sandnes Formation is manifest as a series of carbonate patch reefs (Fig. 10), whereas elsewhere, including in the adjacent Egersund Basin, it contains only siliciclastic sediments (Figs 1b and 16) (e.g. Sørensen et al. 1992;Mannie et al. 2014Mannie et al. , 2016. The formation of carbonate patch reefs requires a lack of clastic sedimentation within a relatively sediment-starved basin. The lack of sediment within the Farsund Basin at this time, compared to the Egersund Basin (Mannie et al. 2014(Mannie et al. , 2016, may reflect differences in their respective onshore source areas and suggests that they were not linked at this time. In the Egersund Basin, facies shallow eastwards, from a fully marine environment to a shoreface setting along the margin of the Stavanger Platform (Mannie et al. 2014). A similar waterdepth change occurs within the Farsund Basin, although still largely within a shallow-marine domain. Geomorphological features suggest relatively deeper water depths in the east, represented by taller and, we infer, deeper-water patch reefs (Figs 9, 10 and 16), and a relatively shallow water depth towards the west which resulted in a dominance of shorter and wider patch reefs. These complementary east-and west-facing shorefaces, and their associated distinct facies belts (i.e. carbonate in the east and siliciclastic in the west), indicate a relative high between the Egersund and Farsund basins, potentially represented by the Stavanger Platform and Lista Nose fault blocks (Hamar et al. 1983;Sørensen et al. 1992). One such topographical high, the Eigerøy Horst, continues northwards, as a series of bathymetric highs termed the Hidra Mountains, to the Norwegian mainland where it may reflect an onshore drainage divide (Rise et al. 2008). A further relative sea-level increase is recorded by the deposition of the anoxic shales of the Egersund and Tau formations (Vollset & Doré 1984;Sørensen et al. 1992). Cessation of carbonate patch-reef growth may have occurred due to a basin-wide transgression or the added input of the anoxic shales (Fig. 16). Fine-grained siltstones and claystones comprising the Sauda Formation were deposited during the Upper Jurassic (Vollset & Doré 1984;Mannie et al. 2014), with a series of southwards-prograding and avulsing deltaic lobes identified within the Farsund Basin (Figs 15 and 16). This deltaic system has an extra-basinal source, probably the Norwegian mainland, and was transported through the Carboniferous-Permianaged Varnes Graben, a sediment pathway during the Triassic Jarsve et al. 2014). Lobe 4, however, may have a local sediment source related to degradation of the Eigerøy Horst, which in this area forms the footwall of the Farsund North Fault (Fig. 15). As with the stratigraphically older patch reefs, the deltaic lobes are also restricted westwards of, and are not present on, the Stavanger Platform (Fig. 17). West of the Stavanger Platform, additional deltaic sequences, the Hardangerfjord and Sognefjord units, are present within the Sauda Formation, sourced from and draining western Norway (Dreyer et al. 2005;Somme et al. 2013;Patruno et al. 2015b) (Fig. 17). Wells penetrating the proximal part of the Hardangerfjord unit penetrate a mudstone-dominated unit (Somme et al. 2013), whereas the Sognefjord unit has been shown to be more sandstone dominated (Patruno et al. 2015b). Within the Farsund Basin, well 11/5-1 penetrates a silty sandstone-sandstone interval corresponding to the distal part of the delta sequence; furthermore, the surrounding sediments are largely dominated by mudstones and siltstones (Fig. 5), so a more sandstone-rich interval would produce the observed large impedance contrast (Figs 11 and 13).
The partitioning between the Farsund deltaic sequence described here and the Hardangerfjord unit located west of the Stavanger Platform appear to reflect the location of the drainage divide onshore Norway. Sediments sourced from west of the divide are deposited into the Hardangerfjord unit (Somme et al. 2013), and those east of the divide being deposited into the Farsund Basin and Skagerrak Sea (Somme et al. 2013;Jarsve et al. 2014). The offshore continuation of this divide may be represented by the highs of the Eigerøy Horst, Stavanger Platform and Lista Nose fault blocks (Hamar et al. 1983;Skjerven et al. 1983) (Fig. 17).
We have shown that the Farsund Basin contains different facies associations and experienced a markedly different tectonostratigraphic evolution to basins to the west, separated by the Stavanger Platform and Lista Nose fault blocks. This may correspond to a boundary between different structural domains, between Caledonian Orogeny and post-orogenic collapse-dominated tectonics to the west (Phillips et al. 2016), and an evolution dominated by the Sorgenfrei-Tornquist Zone and the Tornquist trend to the east (Mogensen & Jensen 1994;Thybo 2000;Mogensen & Korstgård 2003;Phillips et al. 2018) (Fig. 17).

Implications for tectonic activity
The Mesozoic structural evolution of this area is relatively understudied (Jensen & Schmidt 1993;Phillips et al. 2018). Here we use inferences from our proposed geomorphological evolution of the Farsund Basin to place additional constraints on its tectonostratigraphic evolution along with more regional tectonics.
The depositional limit of the Zechstein salt reflects relative structural highs present at the time of deposition (Fig. 1). Onlapping of the salt onto the southern margin of the Farsund Basin indicates that, at that time, the area to the north formed part of the Stavanger Platform, prior to later activity along the Fjerritslev North and South faults. This is in agreement with structural observations that the Farsund Basin did not exist in its present form until the Early Cretaceous (see Phillips et al. 2018). An abrupt step of c. 7 km is observed in the limit across NS2 (Fig. 1). This step may reflect a preexisting topography within the basin at the time of deposition, postdepositional modification and translation of the boundary due to later fault activity, or a combination of both. NS2, along with other north-south-striking faults, may have been active during the Carboniferous-Permian extensional event , although due to a lack of imaging at depth within our seismic data we are unable to confirm this. Preexisting fault-related relief could cause such a step in the depositional limit of mobile salt (Clark et al. 1998); similar steps are observed along strike relating to the Stavanger Fault System (Fig. 1)  . Additionally, the limit of the salt basin may have been modified post-deposition, perhaps relating to Early Jurassic sinistral strike-slip activity (Phillips et al. 2018).
East-trending fluvial channels within the Bryne Formation were, at least in part, controlled by the presence of thin-skinned, saltdetached faults. A key observation is that these east-trending channels are not influenced by the major east-west-striking faults, in particular the Fjerritslev South Fault, that delineate the presentday morphology of the basin. This concurs with structural observations (i.e. the lack of synkinematic pre-Cretaceous strata: Figs 3 and 6) that the east-west faults were not active and had no surface expression at this time. The widening of fluvial channels across NS2 occurs across a subtle topographical gradient, interpreted as the palaeoshoreline. This topographical gradient may be related to differential compaction of the underlying Triassic strata across NS2. East-trending structures also have a negligible influence on the formation and morphology of features within the Sandnes Formation. Patch-reef morphology is unchanged across the eastwest-striking Fjerritslev North and South faults, indicating that they grew in similarly shallow water depths (Fig. 10) (Kendall & Schlager 1981). Conversely, patch-reef morphology changes markedly across north-south-striking faults, from short, wide reefs on the footwall to tall, narrow reefs on the hanging wall (Figs 9 and 10). Water depth has previously been shown to be a key factor in determining carbonate facies and patch-reef morphology, with shallower water depths associated with shorter, wider reef morphologies (Brock et al. 2008), and favouring the formation of patch reefs over more continuous ridges (Colpaert et al. 2007;Purkis et al. 2015). Thus, we infer that this change in reef morphology represents a change in water depth associated with the aforementioned topographical gradient across NS2 (Fig. 9). Those patch reefs that grow in the slightly deeper-water environment (i.e. the hanging wall of NS2) exhibit catch-up growth as they attempt to reach shallower depths, forming tall, narrow structures (Kendall & Schlager 1981;Schlager 1981;Saqab & Bourget 2016). On the other hand, the wider patch reefs situated at shallow water levels (i.e. the footwall of NS2) have no requirement for this catch-up growth and undergo keep-up growth, preferentially growing laterally, forming shorter, wider structures (Brock et al. 2008;Saqab & Bourget 2016). The occurrence of this catch-up/keep-up growth mechanism indicates that the growth of these structures was sensitive to water depth, and therefore that they formed as tropical carbonate reefs, as opposed to cool-water carbonates or carbonate mud mounds (Schlager 2000). Late Jurassic ocean temperatures were relatively equilibrated across the Tethyan Ocean, allowing tropical reefs to form across a large latitude range, including the Farsund Basin (Leinfelder 1994).
Following the deposition of the Egersund and Tau formations, a series of southwards-prograding deltaic lobes were deposited in the Farsund Basin, forming part of the Upper Jurassic Sauda Formation. Lobe geometry appears unaffected by any underlying relief, with the Fjerritslev North and South faults now cross-cutting the lobes. As they are now offset, this implies that no fault-related topography was present at the time of deposition and that the lobes were deposited in a relatively unconfined setting (Somme et al. 2013;Zhang et al. 2016). The lack of Upper Jurassic deltaic systems across the footwall of the Fjerritslev South Fault may be due to erosion following post-depositional fault activity and subaerial exposure of the footwall. Based on this, we infer that activity along the east-west Fjerritslev North and South faults in the Farsund Basin began in the Early Cretaceous, following the deposition of the Sauda Formation deltaic system. Conversely, fault activity within the Egersund Basin started earlier, in the latest Jurassic, affecting the thickness and distribution of different facies (Mannie et al. 2014(Mannie et al. , 2016. In the Farsund Basin, this is likely to correspond to the same extensional event, with the age of the deltaic system straddling the Jurassic-Cretaceous boundary. The Sauda Formation represents an input of sandstone that is likely to have been deposited during an overall net marine transgression (Mannie et al. 2016). This input of clastic material, both in the Farsund Basin and offshore west Norway (Somme et al. 2013), corresponded to the late pre-rift to peak synrift stage of Late Jurassic-Early Cretaceous extension (Brun & Tron 1993;Bell et al. 2014). The Early Cretaceous succession within the Farsund Basin predominately consists of relatively deep-marine sediments. Deposition of these sediments was associated with a deepening of the Farsund Basin relating to Early Cretaceous tectonic activity (Mogensen & Jensen 1994). This is likely to represent the same regional rift event documented to the west, responsible for the deposition of the Hardangerfjord Delta sequence, although this event may be regionally diachronous (Somme et al. 2013;Mannie et al. 2016).

Implications for petroleum systems development of the Farsund Basin
In constraining the geomorphological evolution of the Farsund Basin, we have also identified a series of potential carbonate and clastic reservoirs that may form part of viable petroleum systems. Channels identified within the Bryne Formation (Fig. 7) are likely to be composed of fluvial sandstones (Ryseth et al. 1998). In addition, carbonates, including patch reefs such as those identified within the Sandnes Formation (Figs 9 and 10), and offshore deltaic systems akin to those within the Sauda Formation (Figs 11, 12 and 15) have previously been shown to represent viable petroleum reservoirs (Montgomery 1996;Moore 2001;Saller et al. 2008).
The reservoir potential of the Sandnes Formation patch reefs is complicated due to distinguishing between primary and secondary porosity within the reefs themselves (Enos & Sawatsky 1981). Original porosity within carbonates can be enhanced through dissolution of the host material or, alternatively, may be destroyed and infilled by secondary cementation. This secondary porosity is dependent on a number of different factors. Typically, patch reefs consist of a cemented core containing negligible porosity, with a less cemented, more porous surrounding framework (Enos & Sawatsky 1981). The high-amplitude core and lower-amplitude rim of the carbonate patch reefs as imaged in seismic data (Figs 9 and 10a) may potentially reflect such a change in the degree of cementation, from a relatively compacted and cemented core to a more porous rim. Alternatively, this may correspond to a thinning of the reefs below seismic resolution around their margins. These complications notwithstanding, working patch-reef plays have been discovered, such as the Lime Valley Pinnacle Reef Play in Texas, USA (e.g. Montgomery 1996).
Stratigraphic and structural traps and seals are present within the Farsund Basin. The Zechstein salt would act as a regional seal throughout large parts of the area. However, areas to the north of the depositional limit of the salt may allow vertical migration into the Jurassic section (Fig. 1). The carbonate patch reefs and fluvial channel systems are largely overlain by and encased in shales of the Egersund and Tau formations (Figs 3 and 5). In addition, the isolated nature of the fluvial systems and the patch reefs allow them to represent discrete volumes. The lateral terminations of the Upper Jurassic deltaic lobes would also be expected to have stratigraphic traps at their margins, and are sealed by overlying Lower Cretaceous claystones and siltstones. Variable reservoir quality may be expected within the deltaic lobes; sandier material would be expected in the topsets and foresets of the clinoforms compared to the bottomsets, with a potential reduction in reservoir quality also expected around the distal margins of the lobes (Patruno et al. 2015a, b). Due to the main period of faulting along the east-west faults occurring following the deposition of the Jurassic interval, including these potential reservoir units (Figs 3 and 6), a number of structural traps may also be present. Early Cretaceous faulting offsets and partitions the Upper Jurassic deltaic system into a series of discrete potential reservoir units (Fig. 16).
In addition to these reservoirs, a number of potential source rocks are present throughout the area, each of varying maturity and likelihood of viability. The organic-rich shales of the Tau Formation may be oil-mature in the centre of the Farsund Basin (Skjerven et al. 1983;Sørensen & Tangen 1995;Petersen et al. 2008). These correspond regionally to the Kimmeridge Clay and Draupne shales in the UK and Norwegian North Sea, respectively, which represent key source-rock intervals in each area. The Tau Formation shales may be able to act as a local source rock for the identified Jurassic reservoirs. Regionally, the Cambrian-aged Alum shales, situated to the east of the area (Petersen et al. 2008), may be mature and could act as a potential source rock in this region, although potential migration pathways into the Jurassic interval in this area seem far-fetched. Through constraining its geomorphological evolution, we have identified and mapped key components of the petroleum system within the Farsund Basin, and have shown how seismicattribute-driven interpretation can aid the imaging and mapping of petroleum systems in frontier areas.

Conclusions
In this study we have used a seismic-attribute-driven approach to determine the geomorphological evolution of the Triassic-Jurassic succession in the Farsund Basin, offshore south Norway. Having established this local evolution, we link this to the tectonostratigraphic evolution of the wider area and assess the viability of any potential petroleum systems. Overall, we find that: • The depositional limit of mobile Zechstein salt trends eastwest across the southern margin of the Farsund Basin, onlapping the edge of the Stavanger Platform at the time of its deposition. A step in the depositional limit is likely to reflect base-salt relief relating to a pre-existing fault scarp, but may also be a result of post-depositional modification of the basin. • The geomorphological evolution of the Farsund Basin reflects an overall marine transgression, documented through the identification of fluvial systems of the Middle Jurassic Bryne Formation, shallow-marine patch reefs developed within the Sandnes Formation, and Late Jurassic delta lobes within the Sauda Formation. • The morphology of the identified geomorphological features offers insights into the palaeogeographical setting of the basin. Palaeobathymetry, formed as a result of differential compaction across previously active faults, represents the palaeoshoreline and reflects a change in water depth throughout the deposition of the Bryne and Sandnes formations. The Upper Jurassic deltaic systems are unaffected by, and were therefore deposited prior to, the onset of faulting within the Farsund Basin. • The tectonostratigraphic evolution of the Farsund Basin differs markedly to that of the Egersund Basin to the west, due to the presence of a partition between the two areas. This partition is formed of structural highs corresponding to the Stavanger Platform and Lista Nose fault blocks offshore, and potentially the drainage divide onshore. The differing tectonostratigraphic evolutions between the two areas reflect a difference in their regional tectonic settings: the evolution of the Egersund Basin area is controlled by Caledonian Orogeny and orogenic collapse-related structures, and the Farsund Basin by the underlying Sorgenfrei-Tornquist Zone. • Through this seismic geomorphological analysis we have identified a series of potential reservoirs, including fluvial systems, carbonate patch reefs and offshore deltaic lobes, which along with seals and local sources within the Tau Formation may form parts of working petroleum systems within the area.
This study showcases how seismic attributes and seismic geomorphological analysis can be used to determine the tectonostratigraphic evolution of rift basins. These techniques are able to identify potential petroleum systems, representing a vital tool for the exploration of frontier basins, and also offer insights into the structural evolution and wider tectonic settings of relatively underexplored basins.
Seismic attribute analysis was used to gain more information about the facies present across different stratigraphic levels within the Farsund Basin. RMS amplitude, variance, amplitude contrast and spectral decomposition were calculated along a series of key stratigraphic horizons, with the attributes calculated across a window bounding the interval of interest specified in the text. Root mean squared (RMS) amplitude is a measure of the absolute strength of the reflection, regardless of its polarity; stronger, brighter reflections are represented by higher RMS amplitude values.
Variance represents a measure of discordance between individual traces within the seismic data. The larger the difference between adjacent traces, the higher the variance value; therefore, this seismic attribute typically highlights the edges of discrete structures.
Amplitude contrast works in a similar way to the variance attribute, quantifying the lateral contrast in amplitude between adjacent traces.
Dip azimuth quantifies the dip magnitude and dip direction of the reflections, with the azimuth represented by a colour within a spectrum and the dip by the shade of the colour. This attribute is calculated for each voxel within the data.
In addition to amplitude-derived attributes, we also investigated a number of frequency-derived attributes, to further interrogate the seismic reflection data. The frequency of the data with depth is shown in Figure 4, with an average frequency within the Jurassic interval of c. 35 Hz. Frequency is inherently linked to unit thickness, higher frequency values are able to resolve thinner units (Slatt 2006;Brown 2011).
Dominant frequency is a measure of frequency that accounts for both the instantaneous bandwidth and the frequency, representing a relatively smoothed frequency value.
Spectral decomposition represents a colour blend of discrete frequency values extracted from the frequency spectrum of the data. The frequency spectrum of the data (Fig. 4) is partitioned into a series of bins (each 20 Hz wide) which correspond to a range of frequencies. We then assign three of these frequency values to red, green and blue, representing low-, medium-and high-frequency values, respectively; the brightness of each colour represents the power that frequency component contributes: that is, red colours indicate a preponderance of lower frequencies, whereas structures with a response from all three frequency windows are represented by white. In this study, we assign frequency values of 22, 30 and 45 Hz to red, green and blue colours, respectively, to produce our RGB spectral decomposition colour blend. Hydrocarbon plumbing systems of salt minibasins offshore Angola revealed