Probably proximal pebbles? An outcrop-constrained quantitative analysis of clast transport distances in the lower Triassic Sherwood Sandstone Group, UK

Analysis of the fossil fauna in UK Triassic pebbles has been used to constrain the origin of the pebbles (Salter 1864; Vicary 1864; Matley 1914). Distinct early Ordovician, late Ordovician and Devonian faunas preserved within Triassic pebbles at Budleigh Salterton have a Gondwanan affinity. This is significant because, prior to the late Carboniferous Variscan Orogeny, the Armorican Massif was located south of the Rheic Ocean and was part of the Gondwana supercontinent with a Gondwanan fauna. By contrast, the London–Brabant and Welsh massifs were part of Eurasia, with a very different Eurasian fauna. Consequently, the presence of Gondwanan fossils in Triassic pebbles in Devon is strongly indicative of the northward transport of those pebbles from an Armorican Massif source region (Cocks 1995). Lamont (1946) also ascribes an Armorican Massif source for pebbles found in the Birmingham area based on the Ordovician, Silurian and Devonian fossil content. However, cross-checking against the Paleobiology Database (Uhen et al. 2023) shows that the Llandovery-age fauna, for example, listed in Lamont (1946) occur across a much broader area, extending from the Welsh Massif, through the London–Brabant Massif to Scandinavia and eastern Europe, with only relatively rare occurrences in the Armorican Massif. This suggests that the Armorican Massif was not a unique source of early Triassic pebbles and perhaps not even a volumetrically significant source compared with a more proximal supply from the drainage basins of the Welsh and London–Brabant massifs.

Analyses of heavy minerals and zircon grains from Wessex Basin (Smith and Edwards 1991; Morton et al. 2013) and offshore EISB (Mange et al. 1999) strata also indicate that local sources may have been important. The Wessex Basin heavy mineral and zircon data indicate multiple early Triassic sources. Although some debate exists about the time equivalence of provenance shifts across the southern SSG, they record a dominant Variscan provenance within the central Wessex Basin, with an additional source in the eastern Wessex Basin from the recycling of Old Red Sandstone from the London–Brabant Massif (Morton et al. 2016). The Budleigh Salterton cliff sections, by contrast, record only a pre-Variscan signal, likely derived from the Neoproterozoic of the Armorican Massif. Palaeocurrent and clast analyses from the Budleigh Salterton exposures of the Chester Formation also indicates an eastwards supply from tributary systems bringing fresh feldspars and angular granitic fragments into the otherwise quartz-dominated clast assemblage (Smith and Edwards 1991).

Detrital geochronological studies can be useful in mapping sediment transport routing (e.g. Tyrrell et al. 2006). So far, these studies have focused on data from exploration wells in the lower to middle Triassic of the Wessex Basin, the EISB and elsewhere offshore Ireland (Tyrrell et al. 2012). The studies suggest that the dominant source of lower Triassic fluvial sandstones was Variscan granites, most likely in northern France, with additional sediment input from the recycling of older sedimentary units found within the Welsh and London–Brabant massifs. However, isotopic Pb–Pb data are not yet available from all the other possible granite sources (e.g. the Dartmoor Granite), so it is not possible to eliminate these regions as volumetrically significant sources. There are only very limited data from the central or western SSG, limiting our ability to assess how the provenance evolves along the transport transect.

This study aimed to refine our understanding of sediment supply to the early Triassic southern UK river system that produced lower Triassic pebble-rich sandstones. The high pebble concentrations measured within the central SSG and locally within the western and eastern SSG suggest that local sources may have significantly contributed to the early Triassic fluvial systems. The analysis uses an outcrop-constrained quantitative model of sediment source and transport to test competing hypothesis for either a single dominant southerly sediment source in the Armorican Massif of northern France, or a more complex set of major sediment input points sourced from various more local highland areas, including the London–Brabant and Welsh massifs.

Pebble concentration data were collected and collated from the lower Triassic Chester Formation and Helsby Sandstone Formation strata exposed in outcrop across five geographical areas (Fig. 1, Table 1). The data were either measured directly in the field or compiled from outcrop images and descriptions in published sources (Table 1). For all the field locations, the pebble long-axis orientation data were also measured to provide information on the palaeoflow direction (Fig. 2). The field measurements of pebble concentration used between nine and 18 randomly selected 15 cm × 15 cm square measurement windows on well-exposed areas of outcrop at each location (Table 1).

The pebble concentration was calculated by dividing the total area of pebbles exposed in each measurement window of exposure surface and dividing this by the area of the measurement window. A similar method was replicated as much as possible with the published outcrop images. The concentrations from each measurement window show roughly uniform or normal distributions (Fig. 3). The concentrations were averaged using a standard arithmetic mean to give one mean pebble concentration value per outcrop location (Table 1). A simple analysis of how the mean at each location varies as the sub-sampling size is increased (Fig. 3e) suggests that, even with the relatively small sample size, the sampled values span the full range of possible values and the means are an accurate indicator of the relative magnitude of pebble concentrations between locations at different distances along the sediment transport path (Fig. 3f).

The Chester Formation and Helsby Sandstone Formation strata outcrop across the southern SSG region in the Wessex Basin (Fig. 1), with a total maximum thickness of c. 260 m. Pebble concentration data were taken from published descriptions of the Chester Formation at Budleigh Salterton Cliffs and the Helsby Sandstone Formation exposed at Foxenholes (Smith and Edwards 1991). The pebble-size clasts are dominantly quartzite, with accessory populations of tourmaline and hornfels, quartz feldspar porphyry, vein quartz, sandstone and limestone within a feldspathic sandstone matrix (Smith and Edwards 1991). The Ordovician and Devonian age fossils in the pebbles indicate an Armorican Massif source (Cocks 1989). Pebble concentrations from 5 to 90% with an average of 51% (Fig. 3) were measured from field photographs and sedimentary logs in Smith and Edwards (1991).

The central SSG outcrops in the Worcester, Knowle, Hinkley, Staffordshire and Needwood basins are located 300–500 km north of the Armorican Massif and the strata are up to 200 m thick (Radley 2005). Outcrop data were collected at Hawksmoor Wood and Hulme Quarry (Fig. 2, Table 1), where the average recorded pebble concentrations are 57 and 47%, respectively (Fig. 3, Table 1). About 90% of the pebble clasts are quartzite (Ali 1982), with subordinate clasts of tourmaline and granite. The palaeoflow data indicate a predominantly north to NNW drainage direction (Figs 1, 2).

The eastern SSG on the East Midlands Shelf is 500–700 km north of the Armorican Massif and the Chester Formation is up to 400 m thick. Outcrop data were collected from Nottingham Castle, with additional data on this location from Wakefield et al. (2015). The exposed outcrops are medium-grained, well-sorted, well-rounded, red–brown sandstone with sharp bed contacts and massive and trough cross-bedding. The pebbles are well-rounded quartzite, with subordinate tourmaline and granite, with common mudball intraclasts. The pebble concentration ranges from 1 to 12% with an average value of 6.39% (Fig. 3, Table 1). The palaeoflow data indicate drainage to the NE (Figs 1, 2).

The western SSG in the Cheshire Basin is 500–600 km north of the Armorican Massif and the Chester Formation is 90–365 m thick across the region. An exposed outcrop at Chester Canal and Farndon Cliffs (Table 1) contains medium-grained, well-sorted, well-rounded, red–brown sandstone with occasional finely laminated silt layers defined via sharp contacts. The pebble-sized clasts consist of well-rounded quartzite, tourmaline and granite, with mudball intraclasts. The average pebble concentration is low, with values of 4.78 and 1.72% (Fig. 3), respectively, although some conglomeratic beds reach 40%. The palaeoflows are dominantly west to NW (Figs 1, 2).

It should be noted that this distinction between the eastern and western regions of Triassic deposition separated by the Pennine Hills (Fig. 1) may not reflect the early Triassic palaeogeography. These two regions are now separated by the Pennine uplift and there is evidence that this was originally a Variscan structure or older (Evans et al. 2002). However, there is also strong evidence for >1 km of Mesozoic cover across much of the Pennine area, including several hundred metres of Triassic strata, all removed by post-Cretaceous uplift and erosion (Pearson and Russell 2000; Smith et al. 2005). This implies a continuous east to west zone of Triassic deposition, with little or no source of Triassic sediment in this area, such that the east–west distinction in the Triassic outcrop is useful for description, but is only a relatively recent feature.

All the code and data used to generate the models described here are available from GitHub at https://github.com/Burgesski/SIPATA

The lower Triassic sandstone lithology was modelled using a simple exponential decay recurrence relationship to represent the transport of varying input volumes of pebbles and sand. This method is generally similar to commonly applied diffusion models of sediment transport and deposition with a characteristic transport distance for each lithology implied in the decay coefficient (e.g. Paola 2000; Armitage et al. 2016; Allen 2017). Sediment input and transport distance were calculated from one or more points along a one-dimensional 700 km transect representing the main axis of sediment transport from the Armorican Massif generally northwards between the London–Brabant and Welsh massifs towards the Irish Sea Basin, following the palaeogeographical model suggested most recently by Newell (2018).

The values of E calculated using each method are the criteria used to select the best-fit model from 2.0 × 10⁵realizations calculated using a Monte Carlo process to randomly vary: (1) the sediment input point position across the 700 km length of the model grid; (2) the sediment source input volume up to twice the initial guess value for volume and location; and (3) the proportion of sediment deposition D up to ten times the initial guess value. These ranges are selected to generate realistic pebble proportions along the model profile based on the initial test calculations compared with the data.

Selecting the number of Monte Carlo realizations required to calculate a reasonable best-fit is a compromise between the computation time and the likelihood of finding error minima in the model parameter space. Each model realization is computationally inexpensive, so 2.0 × 10⁵ iterations can be computed in c. 120 s with the MATLAB script running on a high-end desktop computer. Testing with 1.0 × 10⁶ and 2.0 × 10⁶ realizations has a substantially higher computational cost, generates a much larger sample of the model parameter space, yet produces very similar best-fit models with only marginally lower errors, suggesting that 2.0 × 10⁵realizations is the optimum choice.

Because the purpose of this modelling was to test a hypothesis of multiple sediment sources against a null hypothesis of a single sediment source, the initial analysis focused on finding a best-fit single sediment source model. A minimum-error best-fit model for all the outcrop data (Fig. 4a) shows a good match with the most proximal outcrop data points (<400 km), but a substantial mismatch with most of the intermediate distance (400–500 km) and distal (>500 km) outcrops (Table 2). Fitting the model to the proximal and intermediate distance outcrop data only (Fig. 4b) reduces the error (Table 2), but cannot then explain the most distal data points. Fitting the model to the proximal and distal outcrop data (Fig. 4c) produces the lowest error of the options calculated (Table 2). Because the lowest error model matches the proximal and distal points well, but not the pebble proportions observed at intermediate distances in outcrops at x = 400–500 km, this suggests a hypothesis that the pebbles observed in the intermediate distance outcrops may have originated from a different source: not the Armorican Massif, but a more local source, such as the London–Brabant Massif.

To explore and test the hypothesis of multiple, more proximal, pebble sources, multiple best-fit models were calculated with two, three and four sediment sources located along the modelled profile. In each case, the first sediment source has the parameters from the best-fit single source model that generated the lowest error matching only the proximal and distal outcrops (Fig. 4c, Table 3). Additional sediment sources were then introduced to try to generate a similarly low error from good matches with the intermediate outcrops also. The errors were calculated using the two different calculation methods described here, representing either outcrops recording transport along multiple independent axes (Fig. 5, Table 3) or outcrops recording transport along a single axis of sediment transport (Fig. 6, Table 3).

Considering the multiple independent axes models first, all the multiple source models generated a better match with outcrop data (Fig. 5) than the single source model (Fig. 4), with errors ranging from 0.3508 to 0.4028 compared with 1.4194 for the single source model, all outcrop data best fit (Table 3). The lowest error best-fit model (total error 0.3508) had three sediment sources (at 0, 367 and 493 km along the profile), generating a reasonable match with all the pebble proportions observed in outcrop (Fig. 5b). The two-source model (Fig. 5a) has a slightly higher error, indicating that the three-source model is preferable and the four-source model (Fig. 5c) is perhaps an ‘over fit’ to the data, with a sediment input point modelled at 293 km where there is no outcrop data.

The single transport axis best-fit models have a substantially higher error for the two-source scenario (Fig. 6a, Table 3), but best-fit errors of a comparable magnitude for the three- and four-source models (Fig. 6b, c, Table 3). These connected-axis models are also comparable with the independent axes models in terms of the location of sediment input points, the relative sediment input volumes and the depositional decay constants. This shows that the assumption of connected v. independent transport does not make much difference to the model results, suggesting that either the independent transport axes model or a single transport axis model can explain the data equally well.

These modelling results strongly suggest that multiple sediment inputs along the early Triassic fluvial transport system (Fig. 7) are a better explanation of the observed distribution of pebble-sized clasts than a single source of pebbles all coming from a single source in the Armorican Massif.

This very simple sediment transport model provides a useful new tool to help analyse and understand important details of sediment transport systems, specifically the most likely points of significant sediment input along a source-to-sink transport system. For lower Triassic strata in the southern UK, the sediment transport distance calculations suggest that a multiple source model best explains observed outcrop data. The results from single-axis and independent-axis fluvial transport system models indicate a best fit for a three-source model, with input from the Armorican Massif, but also additional sources at 370 and 493 km along the transport profile. The 370 km source indicates sediments derived from the southeastern area of the Welsh Massif (e.g. Powys) and/or the mid-western margin of the London–Brabant Massif, and within the northern end of the London–Brabant Massif, potentially the Midlands Craton (Fig. 7). This result is also supported by the outcrop geology, with pebble concentrations of 29, 56 and 46% observed from 480–500 km long the source-to-sink system in the Hinkley, Needwood and Staffordshire basins.

The multiple source model proposed by the inverse pebble modelling is consistent with existing Triassic palaeogeographical interpretations: the Welsh and London–Brabant Massif formed emergent highs during the Triassic, with particular significance in the early Triassic, and progressive onlap occurred during the Mesozoic (Newell 2018; Pharaoh 2018).

Although the palaeocurrent data show a clearly dominant northward trend in sediment transport (e.g. Thompson 1970), there is also evidence of more complex transport directions indicating multiple sediment source areas. For example, pebble-bearing strata in the lowermost part of the SSG succession in part of the Churnet Valley, Staffordshire locally indicate a palaeoflow from west to east, perhaps indicating a minor sediment source from the southernmost Pennine Hills (Mountney, pers. comm.).

Further north, 600–700 km north of the Armorican Massif, SSG strata are present in the EISB to the Vale of Eden, and the Chester Formation and the Helsby Sandstone Formation are up to 600 m thick (Barnes et al. 1994; Jones and Ambrose 1994; Holliday et al. 2004), with similar lithologies to those described earlier. For example, the exposed outcrops of the Chester Formation at Cove Quarry and Kirtle Water Cliff within the Carlisle Basin consist of fine- to very fine-grained, moderate to well-rounded, quartz-rich red sand with interbedded very fine-grained red silty mudstones, often with abundant mud cracks and occasional angular gravel beds (Brookfield 2004). Volcanic clasts up to 25 mm in diameter are observed within the early strata of this formation, thought to be derived from the Lake District (Barnes et al. 1994). The palaeoflow orientations range from SE through south to NW across the region, as measured via small- and large-scale cross-bedding within thinly bedded sandstone units (Brookfield 2004). This example indicates southwards transport from a local northern sediment source, such as the Lake District, for at least some of the northern SSG strata, which lends further support to the multiple local sediment source hypothesis.

This is the first study across the southern UK lower Triassic to apply a quantitative source-to-sink model to identify likely source regions and constrain the scale and nature of the sediment transport system that produced the SSG strata. Careful observations, measurements and calculations of the pebble concentrations with a small, but adequate, sample size was central to this method. However, the sampling perhaps does not yet span a broad enough stratigraphic range within the lower Triassic strata; elevated concentrations of pebbles within specific facies such as channel lags may still skew these data enough to affect the inverse modelling results and interpretation. Expanding the dataset with full grain size profiles on more exposures in more locations would address this possible issue and would also perhaps facilitate the development of more sophisticated forward numerical models of drainage basin denudation to refine the sediment transport model.

In particular, the addition of subsurface pebble concentration data from the c. 200 km gap between the southern SSG in Devon and the central SSG in the southern Midlands, where there is little or no exposed outcrop (Figs 1, 7), would allow the further testing of models, either this simple model or something more complex. Additional pebble concentration data in this gap would be a particularly important further constraint on the prediction of a sediment source area at 370 km along the modelled transport pathway (Fig. 7).

Heavy mineral, detrital zircon and detrital feldspar analysis from the onshore western, central and eastern SSG would also be useful to refine source-to-sink models. Recent provenance analysis has focused on the south coast of England (Morton et al. 2013) and the offshore EISB (e.g. Tyrrell et al. 2012). The observed compositional variation of heavy minerals in the Sherwood Sandstone strata from the south coast basins indicates complex Triassic transport systems with multiple distinct sub-catchments, rather than one Armorican source with associated long-distance fluvial transport (Newell 2018).

Expanding the stratigraphic study interval to cover the middle and upper Triassic would help constrain the evolution of the Triassic Budleighensis river system and better understand and predict the likely heterogeneity of younger Triassic strata – for example, the Mercia Mudstone Group strata, which are increasingly important as a potential subsurface storage host. Understanding exactly what caused the general fining-upwards trend observed within the UK and European Triassic would be very useful. Did the sediment source areas change through time with increasing amounts of basin linkage and hinterland denudation, or is the fining trend mostly the result of regional climate change (e.g. Radley and Coram 2016)? Also, were local sediment sources maintained, such that the poorly exposed and therefore relatively poorly constrained Mercia Mudstone strata may yet prove to be more heterogenous than commonly assumed? An expanded sediment transport analysis of the whole Triassic would help address these important questions.

This study shows how integrating quantitative outcrop data with simple numerical models provides new insights, tests established palaeogeographical models, and highlights questions and gaps in our data and knowledge that can be targeted by future studies.

The sediment transport model predicts tributaries to the Triassic Budleighensis river system at c. 370 km and c. 490 km along the modelled source-to-sink transect. These distances correspond to source regions in the Welsh Massif and the northern London–Brabant Massif, respectively. Although these predicted locations of tributary sediment input are approximate and may change with more data, the indications of additional source regions and tributary systems robustly suggest that a simple, long-distance transport model from the Armorican Massif and the French Massif Central to the UK Triassic basins is an oversimplification of a likely more complex early Triassic drainage network and palaeogeography. This has important implications for the prediction of the heterogeneity of lower Triassic strata.

If the predicted multiple sediment input points along the transport path were persistent into late Triassic time, heterogeneity in the younger Triassic rocks (e.g. the Mercia Mudstone strata) might also be substantially higher than is commonly assumed. Such a greater than anticipated heterogeneity could potentially have important implications for practical subsurface prediction – for example, related to proposed carbon and nuclear waste storage sealed by upper Triassic strata.

PMB: conceptualization (lead), formal analysis (lead), investigation (lead), methodology (lead), software (lead), supervision (lead), validation (lead), visualization (lead), writing – original draft (lead), writing – review and editing (lead); ED: data curation (equal), investigation (equal), writing – original draft (supporting); JL-K: data curation (supporting), investigation (supporting), writing – original draft (supporting).

This work was funded by the University of Liverpool.

We have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper

All data available from authors on request. Key data and source code are available from GitHub at https://github.com/Burgesski/SIPATA