Gamma ray profiles for the Staithes S-20 and 42/28-2 boreholes are presented in Figure 3. The correlation used the diagnostic log motifs of lithological units that are known to be persistent over greater distances than the spacing between the two study wells (Bachman et al., 2010). These log motifs are calibrated against core in other wells within the basin. In the Permian Zechstein at the base of 42/28-2, gamma ray values are generally low (∼50 API), punctuated by brief peaks to ∼200 API. After a final low, gamma ray values increase to ∼150 API at the boundary with the Bunter Shale. Only the top of the Zechstein was cored in Staithes S-20. Gamma ray values fall to around ∼0 API at ∼3,500 ft, corresponding to an interval of bedded halite, before increasing to ∼130 API at the base of the Bunter Shale (similar to borehole 42/28-2), enabling correlation between the two boreholes.

Both the Staithes S-20 and 42/28-2 boreholes show relatively stable gamma ray profiles during the lower Bunter Shale (around ∼110 API). Gamma ray profiles exhibit a more erratic response during the upper Bunter Shale, beginning at 3,100 ft in Staithes S-20 and ∼7,700 ft in borehole 42/28-2. This marks the base of the Rogenstein Member, which includes interbedded sandstones that contribute to the erratic behaviour in gamma ray data. In both records, the boundary between the Bunter Shale and Bunter Sandstone is marked by a sharp upwards decrease in gamma. The Bunter Sandstone exhibits relatively stable but low gamma ray values, around ∼60 API, and is particularly distinct in the 42/28-2 borehole record. The Bunter Sandstone terminates with a sharp increase in gamma ray, corresponding to an unconformity at ∼2080 ft in Staithes S-20 and ∼6,200 ft in 42/28-2. Evidence for a hiatus in sedimentation is observed directly in the core of Staithes S-20 but is only inferred in borehole 42/28-2 based on gamma ray records and regional correlation (Bachman et al., 2010). Throughout the overlying MMG, the gamma ray record is slightly elevated in Staithes S-20. This offset may be a product of different tools or calibration and both records remain relatively stable. Brief lower gamma Mercia Mudstone packages likely mark the position of greater evaporite proportion, as the low potassium content of the rock reduces the background gamma API values.

The precipitation of gypsum is associated with a minor fractionation factor (Δ³⁴S_{precipitate-brine}), with experimental studies reporting a range between +1.6 and +2‰ (Thode and Monster, 1965; Holser and Kaplan, 1966; Nielsen, 1978; Raab and Spiro, 1991; Van Driessche et al., 2016). Although the δ³⁴S of both the brine and precipitate decrease progressively with continued evaporation, the results of Raab and Spiro (1991) suggest relatively similar Δ³⁴S_{precipitate-brine} values are maintained until the middle of the halite stability field. Thus, it is generally assumed that the δ³⁴S of a marine Ca-sulphate is indicative of the isotopic composition of the brine from which it precipitated (Schreiber and Tabakh, 2000). Considering that we exclusively sampled calcium-sulphates and halite, it is unlikely that changes in mineralogy and/or sulphur isotope reservoir effects associated with prolonged restriction had a major impact on the δ³⁴S_evap records from boreholes 42/28-2 and Staithes S-20.

The δ³⁴S_evap correlation presented in this study is surprising in that the 42/28-2 sediments have been exposed and saturated with drilling fluids: unfortunately, there is no record of what drilling fluids were used when coring. Samples can be fully covered in fluids during the drilling process and washing them away entirely is difficult (Kubo et al., 2016), thus potentially impacting the integrity of sample geochemistry (e.g., Stuckman et al., 2019; Sanei et al., 2020). δ³⁴S_evap values in drill cuttings are vulnerable to contamination by drilling fluids, which can be characterised by a range of different chemical compositions (Ball et al., 2012). However, the similarity between both the Staithes S-20 and 42/28-2 δ³⁴S_evap records in this study would suggest that the drilling fluids used had little to no effect on the sedimentary sulphate. Despite this, future studies should always be aware that the type of drilling fluid used may impact the sulphur isotope record.

Evaporites form in hydrographically isolated marginal marine basins with restricted circulation (Warren, 2010). Local depositional and diagenetic (i.e., post-depositional) processes can alter a global marine δ³⁴S_evap signal recorded in evaporites (Lu et al., 2001; Crockford et al., 2019), potentially limiting the use of δ³⁴S_evap data for stratigraphic correlation. Local effects include riverine inputs of continental sulphate from weathered pyrite and/or evaporites (Lu et al., 2001; Bottrell and Newton, 2006), reservoir effects associated with evaporite precipitation during periods of prolonged restriction (Raab and Spiro, 1991), and microbial sulphur cycling (e.g., microbial sulphate reduction) and subsequent pyrite burial (Bernasconi et al., 2017). Diagenetic effects include syn-sedimentary dissolution/precipitation of halite, the replacement of halite with gypsum (and vice versa) soon after burial (Schreiber and Tabakh, 2000), microbial sulphate reduction below the sediment-water interface (Jorgensen et al., 2019) and a cycle of diagenesis during burial and uplift (Ortí et al., 2022). The latter involves the formation of anhydrite from the dehydration of primary gypsum at moderate to deep burial depths, followed by the rehydration of anhydrite to form secondary gypsum during uplift (Ortí et al., 2022). The sulphur isotopic effect of the dehydration/rehydration of Ca-sulphates associated with burial and uplift is not well understood. Current evidence suggests it is associated with only minor sulphur isotope fractionation (see Ortí et al., 2022 for details), and is unlikely to have imparted any major effect on our δ³⁴S_evap records. During burial to temperatures exceeding 120°C, thermochemical sulphate reduction can occur, involving the chemical reduction of sulphate to sulphide, and the sulphurisation of organic matter (see review by Cai et al., 2022). This process has been associated with significant sulphur isotope fractionations in metal sulphide deposits (Cai et al., 2022). However, only a minor degree of sulphur isotope fractionation occurs between sulphate and sulphide in most petroleum-related sour gas settings (Worden et al., 1997; Cai et al., 2022), likely due to the complete reduction of any anhydrite that dissolves, with sulphate dissolution being rate limiting (Meshoulum et al., 2016; Cai et al., 2022).

It is understood that diagenetic fluids exhibit a high degree of spatial isotopic heterogeneity and will produce more scattered δ³⁴S values in rocks that have interacted with diagenetic fluids, limiting the stratigraphic reproducibility of δ³⁴S_evap records between different basins (Metzger et al., 2014). Thus, a high degree of stratigraphic reproducibility in δ³⁴S is generally considered to be evidence of a global isotopic signal (Metzger et al., 2014). The δ³⁴S_evap curve from Staithes S-20 has been shown to represent a global record of marine sulphate (Salisbury et al., 2022). The small degree of scatter and/or minor offsets between the δ³⁴S_evap curves of Staithes S-20 and 42/28-2 may reflect local depositional and minor diagenetic effects, but the robust correlation between them (Figure 4), as well as their agreement with the global δ³⁴S_evap record (Salisbury et al., 2022), suggest that the long-term trends in the δ³⁴S_evap data represent a global marine sulphur isotope signal.

The δ³⁴S_evap record from borehole 42/28-2 is presented alongside Staithes S-20 (Salisbury et al., 2022) in Figure 4. Clear stratigraphic trends are apparent in both δ³⁴S_evap records even at different sample resolutions. The principles of sulphur isotope stratigraphy and isotope correlation are based on; (1) the assumption that the sulphur isotope composition of the ocean is isotopically homogenous at any given time, and hence, the absolute δ³⁴S_evap value; (2) the trend in the δ³⁴S_evap is directly comparable (i.e., positive versus negative); and (3) inflection points in δ³⁴S_evap represent a shift from one slope to another. It would be appealing to correlate all the trends and inflections points in Figure 4, however, caution needs to be exercised due to the different sampling resolution between the boreholes. Thus, in Figure 4 we have only chosen to correlate seven points through the latest Permian to Late Triassic time interval. In addition, we have chosen to use one lithostratigraphic correlation which will be discussed below.

During the correlation process between borehole 42/28-2 and Staithes S-20, the stratigraphy is assumed to be proportionally equivalent and each stratigraphic element (apart from truncation by the unconformity) is assumed to expand or contract at a similar rate (see left-hand side Figure 4). If a comparable sample resolution was obtained for borehole 42/28-2, then the correlation between the boreholes (core versus cuttings) would have been stronger. However, this was not possible due to the stratigraphic resolution of the drill cuttings collected and the fact that many cutting samples did not contain any evaporitic minerals for isotopic analysis. Similar concerns (e.g., stratigraphic thickness and time, as well as sample resolution) were highlighted by Gröcke (2020) when performing carbon isotope stratigraphy for correlation. The same issues apply for all types of correlation using isotope stratigraphy.

Cyclic variability in δ³⁴S_evap is observed through the Zechstein and lower Bunter Shale intervals in Staithes S-20 (3,100–3,500 ft: Figure 4). This apparent cyclicity is not observed in the equivalent interval of borehole 42/28-2 (7,700–9,680 ft), although this may be in part due to lower sample resolution (Figures 4A,B). If the cyclicity observed in Staithes S-20 reflects global changes in the sulphate isotopic reservoir, then the cycles may prove very useful in high-resolution correlation of the Zechstein throughout Europe. Future sulphur isotope studies of the Zechstein should be performed at high resolution to assess reproducibility of the apparent cyclicity observed in Staithes S-20. After this period of cyclicity, δ³⁴S_evap gradually increases throughout the Bunter Shale and Bunter Sandstone from approximately +8‰ to +15‰ in both boreholes. This range in δ³⁴S_evap constrains the age of these stratigraphic units to the latest Permian (Claypool et al., 1980; Crockford et al., 2019; Present et al., 2020; Salisbury et al., 2022), suggesting the placement of the PTB in this part of the Southern Permian Basin should be shifted from the upper boundary of the Zechstein evaporites to a point within the Bunter Sandstone (Figure 5). This is significant, due to its’ apparent conflict with biostratigraphic constraints for an Early to Middle Triassic age for Bunter Sandstone equivalents in southern England and Germany (see above). Although, this part of the record is more difficult to accurately correlate based on principles (2) and (3) described above, a late Permian age for the lower Bunter in the SNS Basin suggests that towards the basin margins, the Bunter stratigraphic motif is aliased by an older continental clastic system. By extension, it is also possible that the lower parts of the onshore Sherwood Sandstone Group in northern England also extend into the late Permian.

Due to erosion associated with the Hardegsen unconformity in the Early Triassic (Bachman et al., 2010), the δ³⁴S_evap curves of Staithes S-20 and 42/28-2 fail to record the large positive δ³⁴S_evap excursion characteristic of the Early Triassic. The entire evaporite-based stratigraphic interval of the Staithes S-20 borehole hinges on the constraint of a single palynological age (earliest Anisian), which is obtained from immediately above the unconformity (see Salisbury et al., 2022). δ³⁴S_evap values of > +15‰ above the unconformity in Staithes S-20 also support an Anisian age when compared with the global composite record as discussed in Salisbury et al. (2022). Based on absolute δ³⁴S_evap values at this point it is evident that borehole 42/28-2 records the Anisian recovery to pre-excursion δ³⁴S_evap values in the Olenekian that persisted prior to the negative excursion (Figure 4). δ³⁴S_evap values in borehole 42/28-2 show a declining trend, similar to that which is recorded in Staithes S-20, for the remainder of the Triassic—ending with δ³⁴S_evap values of ∼ +12‰ in the Norian.

As shown in Figure 4 the correlation is somewhat more uncertain when there are larger gaps in the δ³⁴S_evap record (e.g., Bunter Sandstone) from borehole 42/28-2. This is to be expected, as lower resolution records will fail to capture the full extent of isotopic variability (especially short-term changes), creating the false impression of greater isotopic heterogeneity between sedimentary basins (Metzger et al., 2014). It is thus unclear whether the lack of cyclicity at the base of 42/28-2 (discussed above) reflects a difference in the evolution of the sulphur cycle between the Cleveland and southern North Sea basins during this time interval. We present two potential correlations (see Figures 4A,B) between Staithes S-20 and borehole 42/28-2 during the latest Permian. Figure 4A assumes that the lithostratigraphic boundary between Z5 and the Bunter Shale is age equivalent in both boreholes (dashed correlation line). In contrast, Figure 4B displays a correlation scheme based entirely upon the δ³⁴S_evap data, thus suggesting a different age for the boundary between Z5 and the Bunter Shale in each borehole, taking account of the possible diachroneity of lithostratigraphic boundaries. Each of these correlations are plausible but highlight the issue when performing sulphur isotope correlation during time intervals with little isotopic variability (akin to strontium isotope stratigraphy). Additional techniques such as evaporite palynology (Gibson & Wellman, 2021) may help to constrain age and sulphur isotope correlation during the late Permian Zechstein. Despite this, the correlation between 42/28-2 and Staithes S-20 looks exceptionally good, even with different sampling resolutions.

It should be noted that data resolution is not the only factor to consider when attempting to derive robust, high-resolution isotope stratigraphic correlations. As briefly discussed above, the rate and magnitude of sulphur isotope variability are of crucial importance (see Yao et al., 2019). Between the depths of 6,760 ft and 5,940 ft in the borehole 42/28-2 dataset, robust correlations can be achieved with the Staithes S-20 record, despite the relatively low sample resolution of 42/28-2. This is due to the abrupt and significant shift in the δ³⁴S_evap record during this depth interval (see Figure 4), which is also reflected in the global δ³⁴S_evap curve (see Salisbury et al., 2022). Therefore, if the time-interval studied exhibits high sulphur isotopic variability, a sample resolution comparable to our Staithes S-20 record may not be essential. This study has thus demonstrated that drill cuttings can be confidently used for δ³⁴S_evap stratigraphic correlation of sedimentary sequences between sedimentary basins.

To compare the gamma ray and sulphur isotope correlations, we used the gamma ray correlation lines shown in Figure 3 to correlate the δ³⁴S_evap curves from Staithes S-20 and borehole 42/28-2 (Figure 6). Seven confident gamma ray motifs were correlated based on their visual similarity. An additional gamma ray correlation was tentatively assigned in the middle Bunter Shale (see Figure 3, dashed line) at the shift from moderate scatter to high scatter in the record. Based on these correlation lines, the δ³⁴S_evap curves are also strikingly similar, however, there are some key differences. For example, based on gamma ray correlation, borehole 42/28-2 extends further back in time in Zechstein cycle Z5. Thus, using gamma ray stratigraphy only the top 4 samples in Z5 from borehole 42/48-2 correlate with the Staithes S-20 δ³⁴S_evap record (Figure 6).

Another key difference is in the correlation of the Röt Halite Member (Triassic). Gamma ray correlation would erroneously imply that the Röt Halite Member extends into the Ladinian, whilst the sulphur isotope correlation constrains it to the Anisian (Figure 7). It is plausible this may reflect errors in the application of lithostratigraphy during the original logging process. In the case of the Staithes S-20 borehole, the lack of clear halite intervals defining both the Röt and Muschelkalk halites, and the facies transition from the limestone-prone Muschelkalk into clastic basin margin facies, has made the log motifs ambiguous. This demonstrates how δ³⁴S_evap records provide a critical constraint on the validity of lithostratigraphy for stratigraphic correlation, particularly where the lithostratigraphic data exhibit greater ambiguity, such as basin margin settings.

Published data provide further contrasting age estimates. Previous lithostratigraphic correlations suggest an Olenekian age for the Röt Halite Member (Warrington et al., 1980), while more recent work suggests an early Anisian age, in broad agreement with our δ³⁴S_evap record (Bachmann et al., 2010). Isotopic analysis of fluid inclusions suggests a late Early Triassic age for the Röt in the Netherlands, Germany and Poland, reporting δ³⁴S values between +27.1‰ and +32‰ (Kovalevych et al., 2002). In contrast, we only record δ³⁴S_evap values between +18.5‰ and +25.4‰ (Figure 4). The reason for this difference is currently unclear, however, it should be noted that Kovalevych et al. (2002) used a different analytical method for sulphur isotope analysis (e.g., offline SO₂ generation versus online continuous flow IRMS at SIBL). Furthermore, the Röt may have formed earlier to the east of the UK and SNS. Additional sulphur isotope analyses of the Röt Halite Member are required to constrain its’ age and determine if this lithostratigraphic unit is diachronous or synchronous across basins.

Correlations using gamma ray and δ³⁴S_evap data suggest that the successions in each well are broadly comparable across the Cleveland and SNS basins (Figure 7). This can be clearly seen in the Anisian interval (Figure 7). The position of the Hardegsen unconformity can be located using the gamma ray logs (Figure 3) and the Anisian exhibits a high rate and magnitude of sulphur isotopic variability (Figure 4), enabling a robust correlation to be made. Interestingly, the correlation schemes exhibit greatest disagreement during the Ladinian, Carnian and Norian (Figure 7). In particular, the sulphur isotope correlation suggests the top of the Mercia Mudstone in borehole 42/28-2 extends into the Carnian–Norian, while the gamma ray correlation suggests it extends until just below the Norian–Rhaetian boundary, possibly reflecting issues with depth averaging between the gamma ray and δ³⁴S_evap records. In this case the sulphur isotope correlation scheme would be favoured, demonstrating the capacity for sulphur isotope stratigraphy to capture lithofacies diachroneity.

The δ³⁴S_evap correlation based on gamma ray is quite robust but has limitations when compared to the sulphur isotope correlation, which is independent of lithostratigraphy (Figure 7). Due to the ‘layer cake’ appearance of the Triassic southern North Sea, lithostratigraphy assumes that the deposition of one facies is “time-equivalent” to the same facies in another basin. When using gamma ray to correlate δ³⁴S_evap one is forcing it to fit in line with the principles of lithostratigraphy. The deposition of an evaporite in one basin may not be time equivalent to the deposition of an evaporite in a nearby basin, and hence, wireline log correlations of apparently similar log motifs could cross time boundaries. The stability and residence time of sulphur in the ocean precludes crossing time boundaries as at any given time the ocean will have a homogenous δ³⁴S_evap signature (Paytan et al., 2012). Therefore, as time changes, so will the δ³⁴S_evap signature of the ocean. In Figure 6 it is assumed that the gamma ray correlation is a timeline and hence, only four δ³⁴S_evap data from borehole 42/28-2 are correlatable to the Staithes S-20 record for the Z5 lithology (see Figure 4A). However, if we consider that the Z5 evaporite sequence in borehole 42/28-2 is “time-equivalent” to the base of the Bunter Shale in the Staithes S-20 core/basin, then this would create a very different correlation curve (see Figure 4B). Thus, a significant strength of sulphur isotope stratigraphy is that it provides independent validation on whether straight lithostratigraphic correlations are accurate, differentiating between log motifs that appear similar but represent strata of distinct ages.

The above discussion reflects the importance of sample resolution and isotopic variability in deriving robust stratigraphic correlations (see Yao et al., 2019 for further discussion). The δ³⁴S_evap data from the Zechstein and the base of the Bunter Shale lack the sample resolution and isotopic variability to facilitate high-resolution correlations. In contrast, although the sample resolution is comparable in the Mercia Mudstone Group, the greater isotopic variability enables high-resolution correlations to be made using δ³⁴S_evap alone (Figures 4, 7). For intervals where high sampling resolution cannot be achieved, or where the δ³⁴S record exhibits relative stability through time, we suggest a multidisciplinary approach, integrated with wireline log stratigraphy, to produce a global δ³⁴S evaporite curve for correlation. With further research, it is anticipated that the δ³⁴S curve from evaporites may become comparable to how strontium isotope stratigraphy is used for global correlation of marine sediments. More importantly, placing evaporite-bearing strata into more constrained stratigraphic timelines will increase our knowledge of these extreme environments in Earth’s history.

As discussed, our δ³⁴S_evap records from the Staithes S-20 and 42/28-2 boreholes provide further chronostratigraphic constraint for the late Permian–Triassic strata of the Cleveland and UK SNS basins. In particular, the base of the Bunter Sandstone has been assigned to the latest Permian, suggesting the PTB may occur within the Bunter Sandstone at these locations (Figure 5). This enables the degree of Hardegsen erosion to be better constrained stratigraphically (i.e., when it initiated and stopped in a basin). This will help to develop more accurate palaeogeographic maps for the latest Permian and Triassic of the United Kingdom. This is of particular significance due to the interest in the Bunter Sandstone as a potential reservoir for carbon capture and storage (CCS) (Holloway et al., 2006; Newell and Shariatipour, 2016; Alshakri et al., 2023, in press).

The use of geological reservoirs for the storage of sequestered carbon dioxide (CO₂) as a super-critical fluid is thought to be a key technology for addressing anthropogenic climate change (Bickle, 2009). The Bunter Sandstone and other equivalents within the Sherwood Sandstone Group are considered suitable candidates for CO₂ storage (Brook et al., 2003; Holloway et al., 2006). This is due to the Bunter Sandstone’s favourable reservoir properties (Chadwick et al., 2008; Dobbs et al., 2018), with fair to good porosity and permeability (Noy et al., 2012), and suitable stratigraphic position, whereby it is sealed by a durable caprock, the overlying saliferous Mercia Mudstone Group (Armitage et al., 2013). In addition, its supposed sheet-like geometry is predicted to be conducive to effective pressure dissipation and reliable injectivity.

It should be noted, however, that an accurate understanding of the regional stratigraphic framework is essential for reliably estimating the suitability of a reservoir for CO₂ storage. For example, the spatial variability in properties (Ringrose, 2020) such as sand:shale ratios, porosity, and permeability within the Bunter Sandstone can be mapped for assessing its’ suitability for CO₂ storage. These assessments are likely erroneous if our revised stratigraphic framework for the western margin SNS is correct, as authors of previous research will have contoured the properties from sandstones of different ages. In addition, although drillcore can provide useful insights, it is common for analogous facies to be studied at outcrop to provide a broader view of the heterogeneities present within reservoir zones (e.g., Newell and Shariatipour, 2016). However, such studies commonly focus on the onshore Sherwood Sandstone outcrops, which were deposited by large river systems during the Early Triassic, a time interval characterised by a warmer and wetter climate than the late Permian (Trotter et al., 2015). Thus, if the base of the Bunter Sandstone in the western margin SNS is truly latest Permian in age then the depositing rivers would have experienced lower volume discharges leading to different reservoir architectures and heterogeneities (Issautier et al., 2014). As a result, alternative analogue facies may need to be found from more ephemeral fluvial systems. These issues will impact the reliability of dynamic models used to predict the outcome of different scenarios, such as injection rates, well counts and locations (e.g., Noy et al., 2012; Williams et al., 2013; Tucker, 2018; Ringrose, 2020). This could lead to suboptimal developments, financial implications and a potential to fail contractual obligations.