New insights into the stratigraphic evolution of SW Britain: implications for Triassic salt and hydrocarbon prospectivity

Hydrocarbon exploration in the Western Approaches Trough has been set back by a series of disappointing exploration wells since the late 1970s. Despite the diverse plays targeted, over 30 exploration wells have failed to discover producible hydrocarbons. The absence of discoveries likely reflects the lack of new concepts, patchiness of stratigraphic mapping in the area and understanding of its geodynamic evolution. In this study, newly available compilations of two-dimensional seismic reflection data and wells across the trough are used to map stratigraphy. We pay particular attention to the distribution of salt and in modelling the petroleum systems in prospective locations.

The Jurassic Lias is the most productive petroleum system in the region, responsible for charging discoveries in the North Celtic Sea, including Dragon, Kinsale, Ballycotton, Seven Heads, Barryroe, and the Wytch Farm field in the Wessex Basin (Caston 1995; Underhill and Stoneley 1998). The presence of salt appears to have played a crucial role in determining prospectivity in basins surrounding the trough. In the West European Atlantic margin, halokinesis of Permian and Late Triassic salt is associated with structures such as diapirs and canopies, which provide hydrocarbon trapping mechanisms (Biteau et al. 2006; Ferrer et al. 2012; McKie 2017; O'Sullivan et al. 2021). Permo-Triassic salt is critical for hydrocarbon prospectivity in the Central North Sea, where the salt walls evolving into diapirs created a series of traps above mini-basins (Hodgson et al. 1992). In the Aquitaine Basin >8 Tcf of gas were discovered in the Lacq gas field above a salt anticline (Biteau et al. 2006). Due to the excellent sealing properties of salt, it has been crucial for hydrocarbon prospectivity of pre-salt gas plays in the Netherlands and southern North Sea (Cornford 1998). The plays are sourced by a stratigraphically deeper Carboniferous petroleum system, one such example is the Groningen gas field: 101 Tcf initial recoverable gas reserves (De Jager and Visser 2017).

The occurrence of salt in the Western Approaches has been discussed by many authors including Ruffell (1995), Deronzier et al. (1994); Hillis and Chapman (1992); Chapman (1989); Hillis (1988); Evans (1990); Ziegler (1987). Seismic and well data indicate that it is present in the North Melville sub-basin, South Celtic Sea, North Celtic Sea and Wessex basins (Fig. 1). However, salt structures, to the best of our knowledge, have not hitherto been identified in the nearby South Melville sub-basin or Plymouth Bay Basin. Here, we examine seismic data and provide evidence for salt deposition across the region and discuss how it influences the petroleum systems.

First, we provide an overview of basement to Cenozoic stratigraphic interpretations from across the area based on a compilation of previous studies and published literature. Secondly, we describe the dataset and methodology used in the study. Thirdly, we examine observations that reveal the deposition and deformation of salt in the North and South Melville sub-basins and Plymouth Bay and SW Channel basins. We then examine Lias and Carboniferous source rock potential and characteristics in seismic and well data. The observations and 1D petroleum system models are used to assess hydrocarbon maturation and timing. Finally, we examine and discuss the implications of these results for the hydrocarbon potential of the area.

The following section describes the current state of knowledge of key stratigraphic sequences in the study area. Then we explain the assumptions for the seismic interpretation and petroleum systems observations. Later, we present our seismic interpretations and highlight updates to this pre-existing stratigraphic framework.

Early Paleozoic to Middle Carboniferous: Wells 73/02-1 and 83/24-1 drilled through basement and encountered highly deformed Early Paleozoic to Carboniferous rocks below the Variscan unconformity (Fig. 1; Ruffell 1995). The top of the pre-Variscan basement typically contains undifferentiated and highly deformed Early Paleozoic to Carboniferous rocks (Evans 1990).

High amplitude reflectors are occasionally observed on top of the Variscan unconformity. They have been interpreted as Permian volcanic rocks, analogous to the Exeter Volcanics onshore UK. They have been associated with Early Permian stratigraphy filling the intermontane or transtentional basins (Evans 1990; Ziegler 1990; McCann et al. 2006). Previous interpretations show a large area of volcanic rocks covering the Variscan unconformity in the North Melville sub-basin (Evans 1990; Fig. 2). The presence of volcanic rocks in the basin is not debated. They are drilled in wells 73/12-1A and 74/01-1 and are thought to be of Late Carboniferous to Early Permian age. The occurrence and extent of volcanics in the Melville basins are supported by magnetic data (Evans 1990; Fig. 1). Similar volcanic rocks are thought to occur in the Plymouth Bay Basin (Pinet et al. 1987). In this study, an alternate model will be presented for some of these high amplitude reflectors in the basement.

Upper Permian and Lower Triassic rocks are difficult to differentiate in places due to the lack of conclusive bio-stratigraphic data in the wells, and thus are grouped as the Permo–Triassic (Fig. 3). In seismic data they exhibit syn-sedimentary growth patterns, with southward growth throughout the area of study (Hillis and Chapman 1992). Onshore outcrop analogues are best observed along the east Devon coast, SW England. They include the New Red Sandstone at Oddicombe, while Early Triassic rocks of the Sherwood Sandstone Group can be seen in Sidmouth and Seaton along the coast (Gallois 2014).

The Middle and Upper Triassic stratigraphy have been drilled in the SW British Isles, and exhibit kilometre-scale salt-rich stratigraphy. The salt is supported by 2D seismic reflection data in the North Melville sub-basin, and it indicates that there are widespread salt deposits (Evans 1990). Salt diapirs mapped using seismic reflection data show that evaporites have affected the structural evolution of the basins (Hillis and Chapman 1992; Fig. 2). In the North Melville sub-basin, four wells encountered several thick units of halite ranging between 180–1500 m (Hillis 1988; Fig. 2).

The Lower Jurassic has been drilled by 3 wells in the basin, and it comprises a series of marly limestone with a sandstone member drilled by well 73/01-1, the unit has an average porosity of 15% (Hillis 1988). The Lias is a proven source rock in neighbouring basins, including the Wessex, North Celtic Sea, Saxony basins, Dutch Central Graben and West Netherlands Basin, with high total organic carbon (TOC) values up to 11% (Cooles et al. 1986). Several petroleum systems modelling studies have been performed by Ruffell (1995), Deronzier et al. (1994) and Hillis (1988), and they show an immature to early oil window in the North Melville sub-basin. South of the study, on the French side of the Western Approaches Trough, the Lias was drilled and showed good source rock potential with TOC's up to 4% in well Rea Gwenn, and 3% in Travank (Deronzier et al. 1994).

Complex structures and highly deformed stratigraphy caused by the Early Cretaceous event masks the true extent of salt deposition in the southern British Isles. In some areas, salt-bearing Upper Triassic appears to have been completely removed by the erosion that generated the Base Cretaceous Unconformity (Hillis 1988; Fig. 2). Well and seismic data indicate that the unconformity formed between Mid-Jurassic to Albian times (Evans 1990). The uncertainty in its age is due to the lack of stratigraphic record, as such it is unclear if the angular unconformity is due to Early Aptian or Mid-Late Jurassic tectonic activity. Smith (1995) interpreted the unconformity as two separate events in the Cockburn Basin, with the mid-Aptian event being responsible for most of the erosion, but, as we will discuss, this is contrary to new evidence from the study area. Ziegler (1987) suggested that Early Aptian rocks are largely absent in the study area. In contrast, we identify Lower Cretaceous stratigraphy in the hanging walls of faults. The event spans from the Late Jurassic to the Early Cretaceous and is likely related to sea floor spreading in the North Atlantic (Hillis 1988). The unconformity caused significant erosion in the onshore SW UK and the basins surrounding the study area (Western Approaches, SW Channel, Wessex, Paris, South Celtic Sea and the Fastnet basins; Evans 1990). It has also been suggested that thermal upwelling processes or extensional stresses could be the origin of the uplift, where it is equivalent to the Late Cimmerian rifting phase (Ziegler 1988, 1990). McMahon (1995) attributed the event to plate re-arrangement stresses, at the time of the northward propagation of oceanic crust at the Azores-Gibraltar fracture zone (McMahon 1995).

The newly released data was provided by the UK Oil and Gas Authority (OGA), the Irish Department of Communications, Climate Action and Environment (DCCAE), and the United Kingdom Onshore Geophysical Library (UKOGL). The seismic reflection data used in this study consists of pre-stack time migrated data covering the UK offshore basins in the study area. The seismic data used for interpretation consists of 30 surveys with 302 lines of varying quality and depth resolution (Supplementary material). An additional 205 lines from offshore Ireland are used for confirming the regional context of the observations seen in the study.

Seismic lines were tied to wells and key horizons were mapped to constrain the tectono-stratigraphic evolution of the basins. Existing biostratigraphic reports and lithostratigraphic well data was used to determine the age of the horizons. The horizons include top Variscan basement, top Mid-Late Carboniferous, top Permo–Triassic, top Upper Triassic, top Jurassic, BCU, top Lower Cretaceous, top Upper Cretaceous and the seabed (Fig. 3). Permo–Triassic extensional faults were mapped first, as the faults would have likely controlled the Upper Triassic sediment distribution in the basin. Then using the interpreted horizons, stratigraphy sub-cropping beneath the BCU was mapped throughout the basin, to generate a regional BCU sub-crop map. Salt was then outlined where it was directly visible on the seismic data or shows indirect evidence from folding, syn-sedimentary growth, or amplitude anomalies.

Petroleum systems modelling (PSM) allows us to examine the dynamic nature of sedimentary basins to determine what conditions would be suitable for hydrocarbons to be generated in source rocks, and if the hydrocarbons may charge potential reservoirs (Hantschel and Kauerauf 2009). The petroleum systems model simulates the burial history, thermal history and source rock generation processes. In this study, PSM was used to test the maturity and the emplacement timing of source, seal, reservoir and migration. The models focus on the Lias and Carboniferous petroleum systems. To model the prospective locations, an initial model is generated at the location of well 73/13-1. The well contains measured temperature and vitrinite reflectance data. The model was constructed using inputs that include well top depths, formation lithologies, estimated erosion amounts and boundary conditions (heat flow, palaeo-water depth, and sediment-water interface temperature). The heat flow was predicted by calibrating the simulated results to vitrinite reflectance and present-day borehole temperature data. The palaeo-heat flow peaks were estimated based on global heat flow analogues for the different basin tectonic phases (Allen and Allen 2013). The heat flow trend was modelled to peak during the end of the syn-rift, followed by a gradual decrease during the post rift (Jarvis and McKenzie 1980). This same heat-flow trend was then assumed to be applicable to the 1D models generated in nearby prospective areas. The source rock transformation into hydrocarbons was calculated by the model using a kinetic reaction. The reactions used relate to the kerogen type of the source rock. In this study a Pepper and Corvi type 2 reaction was used for the Lias, and a Pepper and Corvi type 3 for the Carboniferous source (Pepper and Corvi 1995).

The pre-Variscan basement is represented by chaotic, low amplitude reflections (Figs 2, 3). The Variscan unconformity is typically marked by a discontinuous high amplitude reflector. In some locations in the seismic data, continuous high amplitude reflectors can be seen below the unconformity. Pre-Variscan rocks have been encountered in wells 83/24-1 and 73/02-1 (Fig. 4a). Well 83/24-1 encountered a possibly reworked section of sandstones and siltstones at 905 m depth. This was followed by a unit dominated by slate until the base of the well. Well 73/02-1 penetrated Paleozoic metamorphic rocks displaying alteration and schistose foliation.

The basement is occasionally onlapped by a sequence of high amplitude reflectors that are truncated by an unconformity, the sequence has only been partially drilled in well 86/18-1. The unit contains horizontally bedded coaly shales, cherts, and coals. The coal rich units can be seen as a low gamma ray signature in the well logs (Fig. 4b). The biostratigraphic data of the unit is inconclusive. Currently the well report points to an Early Carboniferous to Early Permian age, but we believe the base unit could be Mid-Late Carboniferous grading into Early Permian stratigraphy. The rocks appear to have experienced minor deformation compared to the underlying pre-Variscan rocks. More importantly, the sediments have not been metamorphosed, in contrast to the drilled pre-Variscan rocks. The presence of coals in horizontally bedded, unmetamorphosed rocks highlight a possible source rock in the basin.

The basement is onlapped by Permo–Triassic stratigraphy and is seen as semi-transparent low amplitude reflections that become more stratified towards the top of the sequence. The sequence has a few regional high amplitude reflections that assist with interpreting the unit across the basin. Well data from 72/10-1A, 73/02-1, 73/06-1, 73/07-1 and 73/12-1A confirm the Permo–Triassic rocks are dominated by sandstones, claystones, volcanic rocks and occasional anhydrites (Fig. 1). Well 73/12-1A encountered basic volcanic lavas K–Ar dated to a youngest age of Late Permian. The following interval is dominated by conglomerates and sandstones rich in acidic volcanogenic sediments. Acidic lavas were penetrated in well 74/01-1A. Both units of acidic volcanic rocks can be clearly seen on well logs by their distinguishing high gamma ray signature. The top of the Permo-Triassic sequence is interpreted based on the increase in sandstone. Well 72/10-1A appears to contain a 400 m unit of conglomerates dominated by quartzites. Similar conglomerates are observed in the Early Triassic Buddleigh Salterton Pebble Beds in East Devon, south of the UK.

The Upper Triassic seismic facies are represented by a series of semi-transparent, discontinuous, and stratified reflections. The reflections are occasionally thickened by acoustically transparent seismic units. The transparent seismic units represent halite and are diapiric in some locations in the North Melville sub-basin. In the south of the basin, the salt is thin or forms smaller isolated pods close to the BCU. The Upper Triassic rocks are best observed in the North Melville sub-basin, in wells 72/10-1A, 73/02-1 and 73/06-1. The sequence is dominated by claystones and occasional siltstones, sandstones and anhydrites. The Upper Triassic salt unit has been dated as Carnian using biostratigraphic data from above and below the salt. The salt mostly consists of pure halite, but sometimes, small units of siltstones and sandstone occur within the salt. The clastic units can be clearly seen in the gamma ray log, where they show a higher gamma ray than the halite baseline.

The Lower Jurassic facies are only preserved in synclines and hangingwall blocks, and are represented as continuous, stratified high amplitude reflectors on the seismic data. In wells 72/10-1A, 73/01-1A and 73-12-1A, the Jurassic facies observed in well cuttings belong to the Lias, and they mainly include calcareous claystones and limestones. Sandstone units can only be seen toward the base of the unit in well 73/01-1A. Most of the Jurassic stratigraphy have been eroded by the BCU.

The BCU is commonly seen as an angular unconformity incising as deep as the pre-Variscan basement. In the syncline north of the basin, the BCU appears to be parallel to the Jurassic stratigraphy, where it is onlapped by Lower Cretaceous sediments. Similar Lower Cretaceous units onlap the faults in the hanging walls in the South Melville sub-basin (Figs 2, 4). Late Lower Cretaceous glauconitic rich sandstones and claystones with occasional units of volcanic rocks are identified in multiple wells. On seismic data the units are seen as high amplitude reflectors draping the BCU in most of the basin.

Discontinuous low amplitude reflections of the Upper Cretaceous chalk appear to conformably overlay the Lower Cretaceous, but a minor unconformity is suggested in some of the wells. The Chalk group contains an unconformity in parts of the basin, where Campanian rocks directly overlie Lower Cretaceous rocks. The extent of the intra-chalk unconformity has not been mapped in this study.

The Paleogene in the Melville basins is present as a series of continuous high amplitude reflectors. The Lower Eocene and Upper Paleocene consist of a shale rich regional marker. The marker shows an increase in gamma ray compared to the underlying chalk group. The Late Paleogene and Neogene is observed as low amplitude semi-continuous reflectors with some reflectors top-lapping at the seabed (Fig. 2). The units are composed of limestones, sandstones and claystones.

In the following section we will present our observations and interpretation of the Upper Triassic salt, and salt related structures in the Melville, Plymouth Bay and SW Channel basins.

In the southern Melville sub-basin, evidence of salt on seismic data is more subtle because deformation related to the Early Cretaceous unconformity masks the presence of the salt. The southern basin is marked by the reappearance of the Lower Jurassic seismic facies preserved in a series of downthrown fault blocks below the BCU (Fig. 5). The Lower Jurassic reflectors are deformed, and show folding vertically above underlying Permo–Triassic normal faults. Most of the folds are truncated by the BCU and are covered by uniformly thick shallower stratigraphy. An example of halokinesis-related folding that also affects Cenozoic stratigraphy can be seen on seismic data in the central part of the section (Fig. 5). In this section an amplitude anomaly is observed above an underlying Permo–Triassic footwall. The structure exhibits continuous folding up until the Eocene, and syntectonic growth units can be observed in the Cenozoic section, north of the structure.

The clearest evidence for laterally continuous salt is observed in the northern part of the basin, which is bounded by Permo–Triassic extensional faults. (Fig. 2). The series of faults form a graben where the Upper Triassic stratigraphy appears to be thickest in the locations of wells 73/01-1A, 73/02-1 and 73/04-1 (Fig. 6). Thick salt diapirs are observed on the seismic data in the vicinity of the interpreted central graben zone. By mapping the faults and presence of salt, the main salt body in the north appears to be connected through a series of grabens that link it to South Melville sub-basins (Fig. 6). In a local topographic high, dividing the North and South Melville sub-basins, well 73/07-1 did not directly log any salt units. The well recorded salt minerals adhered to clays in the mudlog, and dipmeter data showed a disrupted 75 m zone in the Carnian unit. Well 73/08-1 in the central part of the basin did not encounter any halite and sits on a horst block. The Carnian–Norian facies encountered in this well are siltstones, claystones, marl and traces of anhydrites.

Interpreted Upper Triassic stratigraphy throughout the seismic data reveals that in the eastern part of the basin the salt-bearing Carnian is truncated at the BCU. To emphasize the role of base Cretaceous denudation on the present-day distribution of the salt, a BCU subcrop map was generated (Fig. 6). The western and southwestern boundary of the salt is limited by a series of basin bounding pre-Variscan basement highs (Fig. 6). The highs are truncated by the BCU and are covered by Upper Cretaceous and Cenozoic stratigraphy showing minimal structuration. Seismic data over the pre-Variscan high in the SW of the basin shows uniform Permo–Triassic to Lower Jurassic stratigraphy being abruptly truncated at the BCU, exposing the pre-Variscan basement (Fig. 6). Extrapolating the truncated stratigraphy indicates that 1.8 seconds two-way travel time (TWT) was removed from basement highs. Seismic interpretation of the Upper Triassic interval reveals small pods of Upper Triassic rocks in the footwalls of south dipping faults (Fig. 6). In areas where the Upper Triassic is truncated by the unconformity, thickened Lower Cretaceous units are observed to form in the overburden (Fig. 2).

Evidence of the continuation of the salt 230 km NE of the Melville sub-basin can be seen in the SW Channel Basin (SWCB). Well Kulzenn in the French part of the basin encountered c. 40 m of a Triassic halite member as part of the Mercia Mudstone Group (Menpes 1997). In the Plymouth Bay Basin, north of the SWCB, 8–10 km of Permo–Triassic sediments can be observed in the basin depocentre on seismic data. No wells have been drilled to support direct evidence of the salt in the basin, but interpreted faults, folds, and salt weld observations on the seismic data indicate that salt was widespread. A distinct angular unconformity between 1 and 2 s TWT truncates the Permo–Triassic units (Fig. 7). The unconformity is buried by Triassic sediments. The interpretation of Triassic stratigraphy at the seafloor is based on surface geology maps from the British Geological Survey (BGS 2022). The Intra-Triassic unconformity is mapped throughout the Plymouth Bay Basin and the SWCB, and in many locations it behaves as a detachment for structures formed above the unconformity. On seismic data, folded horizons are seen above the unconformity, and faults below are truncated (Fig. 7).

The salt and unconformity interpretation between the Plymouth Bay and Wessex basins are uncertain due to the lack of seismic data coverage between the two basins. The Intra-Triassic unconformity has been mapped across the Plymouth Bay Basin and the SW Channel Basin. It is used to estimate the extent of the salt (Fig. 8). In the Wessex Basin, 70 km NW of the Plymouth Bay Basin, Upper Triassic salt has been previously described by Underhill and Stoneley (1998) and Harvey and Stewart (1998). Interpretation of seismic data from offshore Wessex Basin indicates that salt continues to the eastern proximity of the Plymouth Bay Basin (Fig. 8). 140 km north of the Plymouth Bay Basin, salt collapse structures east of the Somerset basin have been described by Trude et al. (2012), and thus the interpretation expands the area of previous salt deposition to the SE Bristol Channel. The salt in the South Celtic Sea is clearly observed in the new seismic data, and it creates a series of diapirs analogous to the North Melville sub-basin. The only direct evidence of evaporites between the Plymouth Bay Basin and the Melville basins occurs in well 86/17-1. Several potassium salt units were recorded along with a significant increase of chloride in the mud log.

In the following two sections, we will present data and interpretations of the seismic and stratigraphic characteristics of the two key petroleum systems in the study area. We then create a series of 1D petroleum systems models in the prospective source areas to test the maturity and generation timing of the source rocks.

Lower Jurassic source rocks have been drilled by wells 72/10-1A and 73/01-1A in the North Melville sub-basin. They have an average total organic carbon (TOC) <1%, vitrinite reflectance of 0.5%R_o, and orange spore colouration. The properties reflect immature source rocks with poor source potential in 72/10-1A (OGA 2016). In the South Melville sub-basin, similarly low values of TOC and vitrinite reflectance are seen in well 73/13-1. Vitrinite reflectance values of 0.45–0.7%R_o sampled from well 73/13-1, and spore colouration of medium orange to dark orange, indicate that the unit ranges from immature to early oil window.

In the North Melville sub-basin, the deepest present-day pod of Lias occurs in a syncline in the eastern part of the basin, with the Lias reaching present-day depths up to 4000 m (Fig. 9). To assess the potential for these rocks to be mature, we tested the following geological scenarios: First, we systematically varied the thickness of the missing section below the BCU. Secondly, we tested models in which the lower amplitude seismic unit was assumed to be part of the Middle or Upper Jurassic units. If the section is Middle Jurassic, it reduces the total amount of missing section at the BCU. Finally, we varied the heat flow peak during the syn-rift to examine the sensitivity of the parameter on the generation window.

Before modelling the petroleum system in the syncline, a 1D PSM model at the location of a well with good quality borehole temperature and vitrinite reflectance data constrains the heat flow and erosion estimates assumed in the model. The 1D model inputs of well 73/17-1 best calibrate with an erosion estimate of 1200 m at the BCU, and a basal heat flow peak of 70 mW m⁻² during the mid to late Jurassic syn-rift stage. In the second step, a series of 1D petroleum systems models were generated in the syncline location (Fig. 10). The base case (a) assumes the same heat flow, lithology, and erosion, estimates, as constrained by the calibrated model performed on nearby well 73/13-1. To test the impact of erosion on the timing of hydrocarbon generation, two scenarios with less erosion were tested. Scenario (b) uses the same boundary conditions but with 600 m of erosion at the BCU. This value is chosen, as it is the threshold where a secondary hydrocarbon generation window is predicted in the model. Scenario (c) is a minimal case, assuming 200 m erosion at the BCU, this is an arbitrary value to model the effect of the unconformity on the timing and source rock transformation.

The results indicate potential for Late Jurassic oil generation. We note that increasing peak Jurassic heat flow has minimal impact on the overall generation window. A secondary oil generation window is predicted when BCU erosion estimates are lower than 600 m; this value would allow a scenario where the present-day depth of the Lias is much deeper than its depth prior to the Early Cretaceous uplift event.

In the eastern part of the Western Approaches Trough, 70 km west of the Plymouth Bay Basin, well 86/18-1 encountered Carboniferous rocks. At a depth of 2240 m, the well penetrated conglomerates and interbedded sandstones, shales, carbonaceous coaly intervals and beds of highly siliceous and cherty sediments. Further down the well, at a depth of 2408 m, the lithology is dominated by highly carbonaceous shales. Vitrinite reflectance measurement from several units within the well reveal two distinct trends (Table 1). The first is characterized by low-maturity values observed in the Lower Cretaceous and Upper Triassic rocks (0.38%R_o–0.49%R_o). The second trend occurs in Early Permian to Carboniferous rocks and has a range between 2.03 to 2.91%R_o. The vitrinite reflectance measurements are based on a significant number of measured samples. Notably, two samples from the Permian to Carboniferous section (2325.62 and 2444.49 m) show an unexpected decrease in reflectance values. Given the similar range of vitrinite reflectance values from the Permian coaly interval at 1877 m, it is plausible that this interval is part of the underlying Carboniferous section, deposited without a major unconformity. This could suggest that the package constitutes an Early–Permian to Mid–Late Carboniferous unit. However, no additional source rock properties such as rock pyrolysis data and geochemical analysis are available. Moreover, this unit has not been drilled elsewhere in the basin, thus source potential remains unproven. While biostratigraphic evidence from the sampled units is inconclusive, it points towards a possible Early Permian to Early Carboniferous age.

On seismic data, the 13 km wide package appears as a high amplitude seismic unit and it onlaps highly deformed basement (Fig. 11). In the seismic unit, evidence of minor compression can be seen through a series of low angle thrust faults dipping south. The higher amplitude reflectors appear to be dominant in the northern side of the unit. The horizons are subparallel to an overlying unconformity and are then truncated by it. Similar seismic facies occur in the North Melville sub-basin, where high amplitude reflections subparallel to the Variscan unconformity can be seen below the Permo–Triassic unit (Fig. 2). The reflections are c. 20 km wide and become less apparent towards the south of the section and below the salt diapirs in the north. The reflections are truncated by an unconformity and then onlapped by horizons above. Continuous reflections continue to depths over 4 s TWT in the central part of the section. In the Plymouth Bay Basin, a similar seismic package can be seen at the Variscan basement boundary (Fig. 7). The package is a series of high amplitude reflections that thin on both sides and average around 15 km in width. The seismic units can be traced on multiple seismic lines and occur between 25 to 40 km south of both the Melville basin-bounding fault and the Variscan thrusts in the Plymouth Bay Basin area (Figs 6, 7).

A 1D petroleum system model was generated in the location of high amplitude reflectors, at the top of the basement in the North Melville sub-basin (Figs 2, 12). The model assumes the same heat flow and erosion history as well 73/13-1. The first scenario was simulated with 2200 m of erosion at the BCU. This estimate includes 1200 m erosion, as assumed for the Lias 1D base case model (Fig. 10) and an additional 1000 m for the missing Upper Triassic and Lower Lias sections. The second scenario is the minimal erosion case, 900 m of erosion at the BCU, this assumes that there are no additional Jurassic rocks deposited above the Upper Triassic stratigraphy. The results indicate Early Triassic to Late Jurassic hydrocarbon generation with most generation occurring after deposition of the Upper Triassic salt.

Figures 2 and 5 show that salt halokinesis likely created traps that would benefit the plays sourced from the Lias. The figures also indicate that the salt could play an important role as a regional seal for Carboniferous sourced hydrocarbons, trapped in Permo-Triassic rocks. Therefore, we discuss the seismic observations made earlier and their implications for understanding Late Triassic salt distribution in the region.

In the South Melville sub-basin, seismic evidence of a growth package affecting the Cenozoic stratigraphy (Fig. 6), and the seismic anomaly atop a Permo-Triassic footwall, indicates a salt diapir that could have developed by withdrawal of salt from the north. In other areas of the South Melville sub-basin, the stratigraphy below the BCU is still folded and smaller thickness variations in the Cenozoic stratigraphy indicate the salt is more likely to be thinner and less reactive to the deposition of overburden.

Evidence of the salt being more laterally extensive before the BCU event can be seen on seismic data, as an abrupt truncation of the salt-bearing Upper Triassic (Fig. 2). In the area where the Carnian unit is truncated, the folding and local thickening of the Lower Cretaceous units can be associated with compensatory subsidence because of salt withdrawal or dissolution. The missing Upper Jurassic and lowermost Cretaceous stratigraphy in the basin give a 35-million-year window where sub-areal exposure and erosion could have occurred (Fig. 3). The angular nature of the unconformity and the large window of exposure could have exposed the salt to near-surface dissolution or withdrawal resulting in discontinuous residual salt at the BCU subcrop.

South of the Variscan front, 180 km east of the Melville sub-basin, and 70 km west of the Plymouth Bay Basin, well 86/17-1 encountered c. 30 potassium salt bearing units from 1188 m to T.D (Fig. 11). The units can be interpreted from an increase in gamma ray signature. The thickness of the units approximately ranges from 1 to 5 meters. The stratigraphy bearing the salt units have been previously interpreted as Permian age (OGA 2016). But the stratigraphy occurs much higher compared to nearby well 86/18-1 and instead could be part of the Triassic Carnian unit above. If this is the case, then it could represent, sabkha facies, which signify the lateral facies equivalent of the Carnian halite in the Melville basins.

In the Plymouth Bay Basin, the folds caused by interpreted salt anticlines occur above an intra-Permo–Triassic unconformity (Fig. 7). The salt is likely Carnian as it was encountered in well Kulzenn in the SWCB, however it the first time it occurs atop a large unconformity. There can be several explanations for the unconformity. The interpreted Permo-Triassic section could be Early Permian basin fill that is onlapped by Late Triassic stratigraphy. A combined Altmark-Hardegsen unconformity could separate the Permian and Triassic stratigraphy. The Altmark unconformity is thought to originate from thermal uplift that preceded the opening of Tethys (Stollhofen et al. 2008). The Hardegsen is a wide-spread angular unconformity within Triassic stratigraphy, which can be observed in the Dutch Central Graben and Step Graben. In the two basins, during the late Early Triassic, the Hardegsen unconformity separates the Main Buntersandstein Subgroup from the Solling Formation (Geluk and Rohling 1997; Kortekaas et al. 2018). It is thought that the Hardegsen unconformity (Fig. 3) marks a rifting pulse (Radies et al. 2005; Bachmann et al. 2010) or caused by intra-plate stresses and warping (Pharaoh et al. 2010). A second explanation for the intra-Permo–Triassic is that it separates Late Triassic from Permo–Triassic stratigraphy, and it represents a merged Hardegsen and Early Cimmerian unconformity. The Early Cimmerian unconformity has been linked to intraplate stresses exerted on the Central European Basin during the closure of Paleo-Tethys (Ziegler 1990; Stampfli and Kozur 2006). But this is unlikely because the Early Cimmerian unconformity typically occurs at the base of the Norian, and the presence of Carnian salt above the unconformity is not compatible with this hypothesis. A third explanation for the intra-Permo–Triassic unconformity could be related to uplift preceding the increased intensity of rifting during the Carnian and the extension at the Biscay Fracture Zone. In conjugate basins and south of the Western Approaches, the Triassic can be observed to onlap basement in the Algarve, Lusitanian, Jeanne d ’Arc basins (Sinclair 1994; Kullberg et al. 2014; Ramos et al. 2016), and Permian stratigraphy in the Fundy Basin (Sues and Olsen 2015).

The chaotic basement reflections seen on the seismic data reflect the highly structurally deformed nature of the stratigraphy. A good analogue is observed in outcrop at northwestern Cornwall, where east–west striking thrusts and chevron folding are observed within Carboniferous rocks (Leveridge and Hartley 2006). The interpreted post-orogenic Mid-Late Carboniferous basins onlapping the highly deformed basement represent the geometry of the area of study before the deposition of the main Permo–Triassic sequence (Fig. 3). The breccias, volcanic rocks and coarse clastic rocks observed in the wells most likely are the sediments deposited when the Variscan highlands were being eroded. Finer sediments and salt comprising the thick sequences of Upper Triassic rocks are part of the syn rift sequence, where 2 to 8 km of sediment accumulated in the developing Permo–Triassic basins (Figs 2, 6). The multiple zones of halite seen in wells and seismic, capture the likely intermittent marine incursions that would have spilled over into the formed rift basins through connections or sills to the Tethys epeiric seas (Fig. 13). Local topographic highs and marine connections would likely have facies that include continental evaporites, sabkhas and shallow marine facies as observed in well 73/08-1.

Salt has been described and interpreted in several basins south of the study area (Fig. 13). These include the Parentis Basin, Basque-Cantabrian (Ferrer et al. 2012), the Aquitaine Basin (Biteau et al. 2006), and northwestern Iberia (Ford and Vergés 2020). The Upper Triassic salt in these basins is of Carnian age, and its presence in these basins indicate a seaway linking the area of study to the Tethys Ocean through the Bay of Biscay Fracture Zone (McKie 2017). North of the study area, along the Cornubian Ridge between the Western Approaches and the South Celtic Sea, the Upper Triassic is not fully preserved below the BCU. Therefore, it is uncertain if the deposition of the Upper Triassic salt continued north of the area of study in the Haig-Fraas Basin (Figs 1, 12). Northwards in the South Celtic Sea Basin, the thick salt is likely to have covered a larger area before the Early Cretaceous uplift event. Van Hoorn (1987) Interpreted the Triassic basin continues in the Bristol Channel Basin, but it is unclear if the salt deposition is linked with the Somerset Basin. Halite deposition continued north and eastward to the Wessex Basin, Somerset (Trude et al. 2012), Staffordshire and Cheshire (Evans and Holloway 2009), East Irish Sea and North Celtic Sea (McKie 2017), the Slyne and Erris basins west of Ireland (O'Sullivan et al. 2021).

In this study two main source rocks were evaluated, the proven Lower Lias oil-prone marine source rock, and the speculative

Carboniferous gas-prone source rock. In this section we summarize the impact of the salt and discuss the types of plays in the Melville Basin (Fig. 14). We investigate the evidence for whether the two petroleum systems could have plausibly charged Permo–Triassic to Cenozoic reservoirs.

Salt typically has higher thermal conductivity than most rocks, this property will create lower temperatures below the salt unit, and higher temperatures will be transmitted to sediments above the salt (Mello et al. 1995). The temperature change can have implications on source rocks below the salt, resulting in overestimated maturity values. Additionally entrapped hydrocarbons below a diapir could have delayed thermal cracking. However, the salt in the Western Approaches is not consistently thick and laterally extensive at present day, thus since the formation of the diapirs, any significant thermal impact would be restricted to the area directly around the two diapirs in the North Melville sub-basin. In terms of sealing potential, the salt's tendency to deform by flowing rather than fracturing following burial would resist it being breached by movements caused by events such as the Early Cretaceous event and the earlier Cenozoic extension. The Cenozoic extension has not been analysed in this study, but evidence of the extension can be seen on Figure 9, where normal faults can be clearly seen in the Upper Cretaceous and Cenozoic. Some of the normal faults detach directly on the salt anticlines in the Upper Triassic. South of the seismic section of Figure 5 shows how Early Cretaceous related faults detach on the interpreted salt layers in the South Melville sub- basin. These two examples of how salt can help shelter potential hydrocarbon accumulations from destructive events.

Hydrocarbon traps in the North Melville sub-basin include fault-related collapsed diapers atop salt anticlines, basin shoulder traps and unconformity subcrop traps. Collapse structures affect some of the reactive diapirs which displace the BCU, and extensional faults can be observed affecting the shallower succussion (Figs 2, 5). Basin shoulder traps can form in the northern part of the basin (Figs 2, 5) and occasionally have Permo–Triassic stratigraphy juxtaposed to the Lias source. The late early Eocene and mid Eocene show potential sandstone reservoirs in the Melville Basin (Fig. 4b). Migration to structures in the Eocene can be facilitated by extensional faults in the Upper Cretaceous above reactive diapirs (Figs 2, 8). Localized pods of Lower Cretaceous sandstones accumulated between the salt anticlines can become good stratigraphic plays if enhanced by differential movements of the salt diapirs. Such an example can be seen in Figure 2 where halokinesis of the north salt diapir after the deposition of the Lower Cretaceous has resulted in enhanced trap closure. In the area, the Permo–Triassic sandstones have good reservoir properties. However, for these reservoirs to be charged, they must be juxtaposed to the Lias, and the salt absent or welded. The thinner discontinuous salt in the south gives way to a new type of play, a diapir flank play; the salt provides the seal, (Fig. 5). The Lower Lias Hettangian sandstones could act as a carrier and reservoir, migrating any hydrocarbons produced in the Lias source updip to the trap. In areas with welded salt, fault blocks can juxtapose Lower Lias and Permo–Triassic rock, which can create migration pathways and fault traps.

There is an abundance of traps, and deep pods of Liassic source rock in the synclines which would appear to be one of the main areas of hydrocarbon generation (Figs 5, 8). However, modelling the deepest pod of source rock in the North Melville sub-basin highlights a timing problem for the Liassic petroleum system in the region (Fig. 10). The modelling results reveal the main window of hydrocarbon generation is likely to be in the Late Jurassic, and afterwards, the Early Cretaceous uplift event could have caused breaching of traps in Jurassic or Permo–Triassic reservoirs. The significant tilting of the southern side of the North Melville sub-basin observed on the seismic data appears to be a high risk for remigration of hydrocarbons entrapped during the Late Jurassic (Figs 2, 5). In the south Melville sub-basin, the Permo–Triassic and Jurassic structures are dominated by half-grabens (Fig. 6), and the salt appears to be thinner. Although the area has a higher likelihood of fault traps, observations of the folded stratigraphy and truncations at the BCU of the top of the fault blocks suggest a lot of movement during the BCU event (Fig. 5). The fault movement after the main Late Jurassic hydrocarbon generation window adds to the preservation risk of any Lias-sourced traps.

In the Western Approaches Basin, the seismic reflection truncations at the basement and well evidence from 86/18-1 (Fig. 11) suggest that Mid-Late Carboniferous, post-orogenic basins occur as part of the basement. The high amplitude reflectors could represent coal-bearing facies deposited in calmer environments, distal to the active Variscan thrusts to the south (Figs 2, 6 and 10). The interpreted Carboniferous unit could be part of an isolated mid to late Carboniferous basin developed as an intermontane basin onlapping a newly forming Early Carboniferous thrust in the north. An alternate interpretation would be a narrow strike slip basin with growth strata dipping towards the south. The sub-parallel units are likely to be onlapping highly deformed Lower Carboniferous or Devonian rocks. An analogous onshore example can be seen in NW France, where Stephanian aged rocks in the Vouvant Basin (Fig. 15) onlap highly deformed Namurian rocks (Praeg 2004).

The basins appear to be discontinuous and isolated and the distance they occur from the interpreted Variscan thrusts is analogous to intra-Variscan basins seen in northwestern France and along the Variscan trend in NW Europe (Fig. 15). Similar to the Carboniferous Basins in NW France, the interpreted basins in the study area could have developed as part of the offshore extension of the Armorican shear belt (Gumiaux et al. 2004). The Carboniferous coal-bearing basins are well described in Brittany, onshore France. Carboniferous coals can be seen in the Ancenis, Laval, and Vouvant strike slip basins (Houlgatte et al. 1988; Fig. 15). In SW Germany, the Saar-Lorraine Basin formed during the Mid-Late Carboniferous as a narrow strike slip basin filled with continental sediments. It contains numerous coalfields, several conventional hydrocarbon accumulations, and potential for coal bed methane (Pryvalov et al. 2015). Because of the constantly developing thrust sheets and associated strike slip faults during the end of the Variscan Orogeny, many of the Mid-Late Carboniferous basins develop their own unique stratigraphy (Houlgatte et al. 1988). For example, the central part of the section in the North Melville sub-basin (Fig. 2), the reflectors observed between 3 and 4 s TWT can be interpreted as Namurian-Westphalian coal bearing sediments. The reflections continue deeper than 4 s TWT, suggesting a possible Mid-Late Carboniferous strike slip basin with thicker sequences of Carboniferous than that is observed in the Plymouth Bay Basin area (Figs 7, 14). This type of basin is analogous to the Chatoaulin, Laval and Saar Lorraine basins (Fig. 15), where thicknesses of sedimentary rock range between 1–3 km.

The Culm Basin, north of Cornwall, is an example where metamorphosed and unmetamorphosed sections of the Upper Carboniferous can be compared. The Culm Basin formed in the Late Carboniferous (Hecht 1992). The Carboniferous Crackington Formation in Hartland Quay contains rocks with a vitrinite reflectance values of 4.4%R_o (Cornford et al. 1987). The sampled material is thus overmature and of anthracitic coal rank. The material is inert and has no gas generation potential. 25 km NE of Hartland Quay, the Upper Carboniferous Bideford Formation contains coal seams and carbargillites. Mean vitrinite reflectance of the coal-rich units, range between 1.66 and 2.58%R_o, with 3.81%R_o at Greencliff (Cornford et al. 2011). The lower end of the vitrinite puts the gas-prone coals in the peak gas to dry gas window at present day. The Bideford Formation in the Culm Basin is an example of how gas generation potential can exist in unmetamorphosed source rocks in the Upper Carboniferous.

Seismic evidence of the Carboniferous petroleum system possibly generating gas at present day can be observed on the seismic section of Figure 2. In the section, following the base of the Late Triassic Mercia mudstone to where it is truncated by the BCU, a gas chimney effect is imaged in the Upper Cretaceous units above the unconformity. The gas chimney could be gas being produced from the Carboniferous source travelling into the Permo–Triassic carrier beds updip until the Mercia mudstone seal is truncated by the unconformity. Additional hydrocarbon indications can be seen as gas shows in the Permo–Triassic of well 72/10-1A (Fig. 9). The gas shows occur in the Permo-Triassic below the salt, and this indicates that a source deeper than the Lias is required. Such a source could be a seismically unresolved part of the interpreted Mid-Late Carboniferous basin in the North Melville sub-basin (Fig. 15).

In this study, a new compilation of seismic reflection and well data allowed us to map the distribution of Triassic salt into the South Melville and Plymouth Bay basins, where salt was not previously described. Salt halokinesis has created traps for the Liassic petroleum system but hindered its migration to Permo–Triassic reservoirs in tilted fault blocks. The salt is thick and continuous in several parts in the basin and could provide regional seals for traps sourced from the Carboniferous petroleum system. Hydrocarbons accumulated in such pre-salt traps would likely be sheltered from some of the effects of the Early Cretaceous uplift and later Cenozoic uplift events, due to the ability of the salt to deform without breaching the traps.

Petroleum systems analysis and modelling revealed that the Lias has not proven to be mature in any drilled parts of the basin, and in areas where the Liassic source rock is deepest. Models show that there are high risks on hydrocarbon generation timing. The timing of generation for hydrocarbons produced from the Lias occurred during the Late Jurassic, earlier than the deposition of Cretaceous and Cenozoic reservoirs, and the establishment of adequate seals. The Early Cretaceous uplift event would have introduced a preservation risk due to remigration of hydrocarbons and breaching of seals. The Mid-Late Carboniferous petroleum system is sporadic and will require higher quality 3D seismic data to image pre-salt traps and map the extent of the Carboniferous intermontane basins.

Even with the improved imaging and the advance in modelling petroleum systems, there are significant exploration risks because of source generation timing, structural complexity and the lack of obvious undrilled pre-salt tilted fault blocks. Thus this study highlights why exploration has been so far unsuccessful in the UK part of the Western Approaches. However, we believe that the Carboniferous petroleum system has been overlooked in previous studies. Our results predict that, although very high risk, the Mid-Late Carboniferous petroleum system could have potential for unlocking a new pre-salt gas play in the Western Approaches Trough.

The authors would like to thank the Oil and Gas Authority (OGA) and/or other third parties for providing the seismic and well data for the offshore UK. We would also like to thank the United Kingdom Onshore Geophysical Library (UKOGL) for providing the onshore well and well top data libraries. We would also like to acknowledge that the metadata and Digital Terrain Model have been derived from the portal http://www.emodnet-bathymetry.eu.

We thank the Petroleum Affairs Division (PAD) of the Department of Communications, Climate Action and Environment (DCCAE), Ireland, for providing access to released well and seismic data.

SSH: conceptualization (lead), data curation (lead), formal analysis (lead), investigation (lead), methodology (lead), writing – original draft (lead); AF: methodology (supporting), project administration (supporting), supervision (lead), writing – review & editing (equal); GGR: supervision (supporting), writing – review & editing (equal); RB: methodology (supporting), supervision (supporting), writing – review & editing (equal)

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

All data generated or analysed during this study are included in this published article and supplementary information files.