Virtual outcrop-based analysis of channel and crevasse splay sandstone body architecture in the Middle Jurassic Ravenscar Group, Yorkshire, NE England

Well-exposed fluvio-deltaic deposits of the Middle Jurassic Ravenscar Group (Yorkshire coast) provide a direct analogue for North Sea reservoirs, but previous studies were restricted to a few accessible bays. This study used lidar and drone photogrammetry to create near-continuous virtual outcrops, 21 km long and 30–150 m thick, in an area which is largely inaccessible. Remote sensing data were supplemented by 45 outcrop logs and data from 27 pre-existing, behind-outcrop boreholes to improve understanding of the geometries, architectures and stacking patterns of 10 distinct sandbody types; the most important reservoir analogue units are channels and crevasse splay bodies. Channel bodies are 30–2038 m wide and 2–28 m thick with W/T ratio ranges of 5–105; the majority are multi-storey with an average thickness of 8 m and width of 182 m, while the single-storey channel bodies average 4 m thick and 50 m wide. Crevasse splays are 15–>1285 m wide and 0.3–6.5 m thick with W/T ratio ranges of 50–>541. There is a marked lateral and vertical change in channel body dimensions and facies proportion. The study suggests that the channel architecture and depositional nature of these successions are controlled by base-level fluctuation and floodplain topography. Supplementary material: Fieldwork-based sedimentary logs, borehole core-based sedimentary logs with facies proportions, workflow to prepare virtual outcrop, method to estimate true width of sandbody, correlation panel of crevasse splay bodies, interpreted virtual outcrops, and dimensional data of channel, crevasse splay and crevasse channel bodies are available at https://doi.org/10.6084/m9.figshare.c.5795793

Fluvio-deltaic sandstones act as significant petroleum reservoirs around the world (Tyler and Finley 1991;Keogh et al. 2007;Ford and Pyles 2014). Heterogeneity within these reservoirs exists at a variety of scales from the laminae to the basin fill although the main challenges typically exist at the architectural element scale (Keogh et al. 2007). The fluvial architecture describes the stacking of channel and overbank sandbodies and their geometry and interconnectedness (Allen 1978). Understanding and predicting fluvial architecture is key to understanding reservoir heterogeneity which has important economic implications as these deposits act as petroleum reservoirs, groundwater aquifers and potential sites for CO 2 sequestration (Campbell 1976;Miall 1988;Eschard et al. 1992a;Eaton 2006;Choi et al. 2007;Keogh et al. 2007Keogh et al. , 2014Ambrose et al. 2008;Nichols et al. 2011;Issautier et al. 2013;Owen et al. 2017). To date the majority of studies have focused on the stacking of channel sand bodies (e.g. Friend et al. 1979;Allen 1983;Blakey and Gubitosa 1984;Hirst 1991;Alexander 1992a;Alexander and Gawthorpe 1993;Bridge and Tye 2000;Gouw and Berendsen 2007;Jensen and Pedersen 2010;Sahoo et al. 2020), while overbank crevasse splay deposits have received less attention despite the fact that they can be both economically important and represent a key part of any holistic study of the fluvial system (Bown and Kraus 1987;Farrell 1987;Kraus 1987;Smith and Perez-Arlucea 1994;Willis and Behrensmeyer 1994;Miall 1996;Tooth 2005;Hajek and Edmonds 2014). The sedimentology of crevasse splay deposits has been studied both in modern fluviodeltaic settings (Elliott 1974;O'Brien and Wells 1986;Farrell 1987;Bristow et al. 1999;Li et al. 2015) and ancient sedimentary rocks (Fielding 1984;Platt and Keller 1992;Jones and Hajek 2007;Widera 2016;Burns et al. 2017). However, there are few quantitative and architectural studies of crevasse splays. Notable exceptions include studies by Smith (1993) and Gulliford et al. (2017) in the Lower Beaufort Group,  in the Ravenscar Group, Jorgensen and Fielding (1996) in the Callide Coal Measures, van Toorenenburg et al. (2016) in the Huesca fluvial fan, and Burns et al. (2017) in the Castlegate Sandstone and Neslen Formation.
Overbank deposits record significant information on the environment and palaeoclimate at the time of deposition, especially as they typically contain a more complete stratigraphic record than the associated channel deposits (Bridge 1984;Fielding 1984;Kraus 1987;Demko et al. 2004;Lazar et al. 2015;van Toorenenburg et al. 2018). Overbank systems can also be relatively sand-rich when they are crevasse splay-prone and contribute to net reservoir/aquifer volumes and improve reservoir communication Doyle and Sweet 1995;Anderson 2005;Pranter and Sommer 2011;van Toorenenburg et al. 2016van Toorenenburg et al. , 2018. The character of overbank deposits, the presence and maturity of palaeosols and the vertical stacking of different facies (e.g. fining or coarsening upward) provide insight into the shifting position of the concurrent fluvial channel (Bown and Kraus 1987;Slingerland and Smith 2004;van Toorenenburg et al. 2016;Burns et al. 2019). The nature of overbank deposits is useful for predicting the style of channel avulsion, which has a strong influence on fluvial stratigraphy and the sandbody architecture (Jones and Hajek 2007;Hajek and Edmonds 2014). Detailed study of the channel bodies in conjunction with the associated overbank deposits provides a clear picture of the palaeodeposition systems, sandbody distributions and their architectures (Fielding 1984;Bown and Kraus 1987;Kraus 1987). The present research focuses on a 40 km long, 30-150 m thick series of outcrops which expose the fluviodeltaic deposits of the Middle Jurassic Ravenscar Group along the Yorkshire coast between Kettleness and Scarborough South Bay (Fig. 1). The three fluvial formations within the Ravenscar Group are each between 45-65 m thick and include channel sandstones, crevasse splays and floodplain mudstones which were deposited under a humid, subtropical climate in an alluvial to deltaic and paralic setting (Alexander 1992a;Morgans et al. 2017). These formations have been used as an analogue for hydrocarbon reservoirs, especially the similar aged Ness Formation which is part of the Brent Group in the North Sea (Hancock and Fisher 1981;Dreyer et al. 1990;Alexander 1992a and references therein;Butler et al. 2005).
Whilst these sections have been previously studied from a sedimentological (Hemingway and Knox 1973;Leeder and Nami 1979;Livera and Leeder 1981;Alexander 1992a;Ielpi and Ghinassi 2014) and reservoir analogue (Ravenne et al. 1987;Eschard et al. 1992a;Alexander and Gawthorpe 1993; perspective, the previous work has focused on a limited number of accessible sections based on field observations and photomosaics taken from boats while the majority of the cliffs have been inaccessible. Moreover, outcrops studied by twodimensional photomosaics and/or scaled outcrop sketches, and the correlation between field logs, cannot account for the natural rugosity of exposures and introduces errors in the measurement of sizes and orientations of geobodies (Rittersbacher et al. 2014). In the current study these shortcomings have been addressed by using virtual outcrop data acquisition techniques as an addition to traditional field data collection Hodgetts 2013;Rittersbacher et al. 2014;Buckley et al. 2019).
The current study is designed to holistically address the fluvial/ fluvio-deltaic system including both the channel and overbank deposits to provide a better understanding of the depositional history and controls on sandbody architecture and the associated heterogeneities The study used modern, virtual outcrop methods (Enge et al. 2007;Buckley et al. 2008) to provide additional insight into the geology of the entire depositional system by linking virtual outcrops with sedimentary logs from outcrops and behind outcrop boreholes. Specifically, the present research aims to improve understanding of the fluvio-deltaic sandbody architecture in these sections through (i) examining the fluvial facies in terms of their detailed sedimentology, stacking patterns and proportion; (ii) quantifying the channel and crevasse splay body dimensions; (iii) identifying vertical and lateral trends in sandbody dimensions to  Kirby and Swallow 1987;Milsom and Rawson 1989;Kent 1980). improve understanding on controlling factors; and (iv) analysing the architecture of overbank sandbodies and their role in paralic reservoir.

Geological setting and stratigraphy
The Ravenscar Group was deposited in the Cleveland Basin during the Middle Jurassic (Aalenian-Bathonian) (Hemingway and Knox 1973;Lott and Humphreys 1992;Cox and Sumbler 2002). The basin formed in the Jurassic as a part of a system of shallow epeiric seas and small extensional basins which linked to the North Sea Basin via the Sole Pit Trough (Fig. 1a). The Cleveland Basin was relatively small and was bounded to the NE by the Mid-North Sea High and the Pennine High to the west. The East Midland Shelf lay to the south of the basin, the northern part of which was comprised of the Market Weighton High (Kent 1955). The basin formation began in the Late Triassic as a result of differential subsidence and continued into the Cretaceous; by the Late Cretaceous active subsidence had halted and inversion began in the Early Neogene (Kent 1980;Kirby and Swallow 1987;Rawson and Wright 1992).
The Cleveland Basin lies to the south of the Mid-North Sea High and the major Early to Middle Jurassic thermal dome which uplifted the central North Sea (Sellwood and Hallam 1974;Ziegler 1982;Underhill and Partington 1993). Middle Jurassic sedimentation in the Cleveland Basin was characterized by progradation of fluviodeltaic systems away from the thermal dome region toward the south and marine transgressions advanced in a north to northwesterly direction across the Market Weighton High (Hemingway 1974).
A number of regional-scale extensional and probable strike-slip fault complexes affected the Cleveland Basin (Hemingway 1974;Kirby and Swallow 1987;Fig. 1c). The east-west-trending Asenby-Coxwold-Gilling Graben, the Helmsley-Filey Fault Belt and the Howardian-Flamborough Fault Belt (defining the Vale of Pickering) were intermittently active to the south of the basin before the Market Weighton High formed during Mid-to Late Jurassic times (Kirby and Swallow 1987;Wright 2009). Northsouth-trending structures, such as the Peak Trough and Peak Fault (Milsom and Rawson 1989), the Cayton Bay Fault and the Whitby Fault cut the eastern margin of the basin. The Peak Trough contains thick Lower Jurassic rocks, suggesting that faults were at least active then (Hemingway 1974;Milsom and Rawson 1989). Late Toarcian regional uplift initiated by volcanic activity in the Central North Sea was associated with the development of the Mid-Cimmerian Unconformity which removed the upper part of the Lias Group across the basin (Black 1934;Hemingway 1974;Underhill and Partington 1993).
A number of the major bounding faults, especially the east-westtrending Coxwold-Gilling and Howardian-Flamborough fault belts, are known to have been active during the Cimmerian Orogeny which showed extension in the Oxfordian (Wright 2009) and renewed movements in post-Cretaceous times (Kirby and Swallow 1987;Starmer 1995). Petrographic and fission-track analysis suggest that prior to the inversion and north-south compression during the latest Cretaceous-Neogene, the Middle Jurassic sediments were buried to a depth of about 2-3 km (Hemingway and Riddler 1982;Green 1986;Bray et al. 1992). The inversion and north-south compression formed the complex east-west-trending Cleveland Anticline (Fig. 1c) and subsidiary folds including the Lockton Anticline, Goathland Syncline and Robin Hood's Bay Dome (Kent 1980). The Ravenscar Group is comprised of marginal to non-marine clastics, thin coals and ironstones which are intercalated with relatively thin marine units of variable lithology. The maximum thickness is about 230 m in the Ravenscar to Scarborough section, thinning westward to 114 m in the Hambleton Hills and southwards to 57 m in the Fordon Borehole ( Fig. 1c; Hemingway and Knox 1973;Powell 2010). The Ravenscar Group is composed of five formations: the Saltwick, Eller Beck, Cloughton, Scarborough and Scalby formations. The Saltwick, most of Cloughton, and Scalby formations are generally considered to be delta plain deposits (Hancock and Fisher 1981;Livera and Leeder 1981;Alexander 1992a). The delta plain deposits are separated by three distinct transgressive marine units: the Eller Beck Formation, the Lebberston Member of the Cloughton Formation, and the Scarborough Formation (Hemingway and Knox 1973;Livera and Leeder 1981;Fig. 2).
The lowermost part of the Ravenscar Group, the Aalenian Saltwick Formation, unconformably overlies the marine Dogger Formation. In places, where the Dogger Formation is absent, it rests unconformably on the Alum Shale of the Lias Group. The Saltwick Formation is comprised of channel sandstones, crevasse splays and floodplain mudstones (Alexander 1992a). The Saltwick Formation is overlain by shallow marine deposits of the Eller Beck Formation . This unit has a conformable base and coarsens upward. It comprises thin beds of ironstone near the base to mudstone and finally fine-to medium-grained planar to low-angle cross-stratified, and wave-rippled sandstone Hemingway 1974;Livera and Leeder 1981). The presence of rootlets at the uppermost part of the sandstone suggests emergence and the unit passes gradationally into the overlying fluvio-deltaic deposits of the Cloughton Formation (Livera and Leeder 1981). The Cloughton Formation is lithologically similar to the Saltwick Formation except for the presence of the marine limestone/ sandstone unit of the Lebberston Member in the south of the basin, which divides the formation into the lower Sycarham Member and upper Gristhorpe Member (Powell 2010). The Scarborough Formation conformably overlies the Cloughton Formation and comprises a wide range of deposits including sandstones, mudstones, argillaceous limestone and different types of mixed siliciclastic/carbonate sediments (Hemingway and Knox 1973;Cope et al. 1980;Gowland and Riding 1991).
Overlying the Scarborough Formation is a succession of channel sandstones, crevasse splay sandstones and floodplain mudstones of the Scalby Formation which demonstrate the return of fluvio-deltaic and paralic depositional settings (Leeder and Nami 1979). The lowermost c. 9 m of the formation consists of fine-to mediumgrained, locally pebbly, cross-bedded channel sandstones termed the Moor Grit Member, which rests unconformably on the Scarborough Formation (Nami 1976;Nami and Leeder 1978;Leeder and Nami 1979;Ielpi and Ghinassi 2014). The Moor Grit is overlain by the exhumed meander belt deposits, and delta plain channel and overbank deposits of the Long Nab Member (Nami 1976;Alexander 1992b;Ielpi and Ghinassi 2014).

Data and methods
Data were collected by traditional field observations, terrestrial lidar scanning, and RPV (remotely piloted vehicle, drone) photogrammetry (Howell et al. 2021) from a 40 km section of the coast. Sections are between 30 and 150 m thick. The sections are subdivided into nine outcrops (Fig. 1b), which are separated by areas of no-exposure, or in the case of Robin Hood's Bay, by an anticline which brings older Dogger Formation deposits to the surface. A total of 45 sedimentary logs were measured at accessible outcrop localities (at least one log per outcrop except Ravenscar; Appendix A) to calibrate the virtual outcrops and to document the lithological variation, sedimentary structures, ichnofacies and palaeocurrent. Detailed sandbody architectures were sketched on photo-panels in the field and then on virtual outcrops in the lab using the LIME software (Buckley et al. 2019). In addition, more than 800 m of core from 27 boreholes were also logged in the National Geosciences Data Centre and the British Geological Survey, UK. These behind-outcrop boreholes are clustered near the Long Nab outcrop at Cloughton Bay and were drilled by Institut Francais du Petrole in the late 1980s. The boreholes are 30-50 m deep and sampled the Scalby Formation and uppermost part of the Scarborough Formation (Eschard et al. 1992a). Sedimentary logs from the borehole cores were prepared at the scale of 1:50 with detailed descriptions of grain size (Wentworth Scale), texture, colour, compaction, grain sorting, bed thickness, sedimentary structures and bioturbation index with characteristic trace fossils (Appendix B).
Ground-based lidar scanning was performed at the Whitby East and Cloughton Wyke sections following the methodologies outlined in Buckley et al. (2008) and Enge et al. (2010). RPV (drone) photogrammetry was used at Kettleness, Whitby West, Beacon Hill, Ravenscar, Long Nab and Scarborough South Bay cliff sections using the methodology outlined in Howell et al. (2021). Lidar and RPV data were used to build scaled photorealistic models (virtual outcrops; Appendix C) from which sandstone body geometry (e.g. width, thickness), stacking pattern, distribution and average lithofacies were studied. The virtual outcrops (VOs) provide a spatially accurate three-dimensional representation of the outcrop (Fig. 3).
The VOs were imported into LIME, a software for VO interpretation (Buckley et al. 2019) to visualize and interpret the sections. Sandbodies were each mapped with polylines and given a unique name. Their width and thickness were measured, as well as the outcrop orientation, a qualifier for their completeness (complete when the full sandbody is exposed, and incomplete when the sandbody is partially exposed, i.e. the sandbody dimension is greater than the outcrop extent) and a description of the lithology and architectural elements present. The measured width refers to a minimum observed value and includes jump correlation across small (<50 m) gaps in the exposures (Rittersbacher et al. 2014). The length of the direct line between the two outermost points of the sandbody is considered as its apparent width (Gouw and Autin 2008) which is then corrected for obliquity following the approach of Fabuel-Perez et al. (2009). In this method, the angle of the exposed sandbody was combined with the palaeoflow direction to calculate the true width of the sandbody (Appendix C). Due to the inaccessibility of most of the cliff sections, regional palaeoflow directions for each formation, collected from sedimentary structures in the field and published studies, were used rather than measurements from individual sandbodies. The mean palaeocurrent direction of the Saltwick channel bodies is towards 182°(n = 56) and the Scalby channel bodies towards 170°(n = 57). No palaeocurrent data were measured for the Cloughton Formation and the average palaeocurrent data of this formation was adopted from Livera (1981) as towards 194°. The thicknesses of the sandbodies and the formations were measured vertically keeping the VO horizontal to overcome the measurement error due to slight tectonic dip of the sections. Several measurements were taken at an equal distance for each sandbody to record their maximum and average thicknesses.
The dimensions of crevasse splay deposits were also measured from VOs and where possible in the field. An associated study of modern systems (Rahman 2019, ch. 2) suggests most of the splays are generated at around 90°to the channel flow direction. Therefore, regional palaeocurrents were also used to correct apparent widths measured from the outcrops (Appendix C). Thickness data for the splays were based upon a minimum of 10 equidistant measurements along the splay and maximum, mean and minimum thicknesses were recorded. For both channel and splay deposits, the height of the top of the unit above the base of the respective formation was also recorded as a stratigraphic datum in a similar manner to the method used by Rittersbacher et al. (2014).
Pseudo-logs also known as virtual logs were generated from the VOs to study the lateral facies variation away from the outcrop measured sections. A total of 142 pseudo-logs were generated with a typical spacing of 200 m. These included 96 from the Saltwick Formation, 26 from Cloughton Formation and 20 from the Scalby Formation (Fig. 4). The thicknesses of different sandbodies and separating mudstones were recorded so that the sand/mud (equivalent to the net/gross ratio) ratio could be quantified.

Facies analysis
Facies characteristics of the fluvial successions were studied from sedimentary logs (Fig. 5) collected from the outcrop and borehole sections. A total of 24 lithofacies were identified in the fluvio-deltaic successions of the Ravenscar Group, which are summarized in Table 1 and Figure 6. The lithofacies were grouped into seven facies associations which were further grouped into those that form within the channels and those that form in the overbank environment (Table 2 and Figs 7 and 8). The associations are described below.

Channel deposits
Large-scale erosive-based bodies are generally comprised of coarse sediments deposited within the channels. Channel deposits Fig. 4. Schematic representation of the distribution of lithofacies based on virtual logs between Kettleness and Scarborough South Bay, a 40 km long section. The outcrop and core-based sedimentary logs were also considered in preparing the panel. Formation thickness adopted from Hemingway (1974) where formation top/base is not exposed and in the case of a data gap. are categorized into three facies associations based on their lithological characteristic. The channel-storey and stacking patterns are described in the subsequent sandbody architecture section.

Facies association 1 (FA1): sandstone-dominated channel deposits
Facies association 1 (FA1) is mostly comprised of the accretionary bar, and aggradational channel sandstone. The deposits occur in accretionary barforms that are up to 8 m thick and 120 m long. Stratal dips within the bar forms are between 7-25°( Fig. 7a, b). Strata with the accretionary sets typically have a uniform dip and are separated from similar bar forms within the same sandbody by minor erosion surfaces and a change in dip direction and angle. Palaeocurrents within the bar forms are oblique (>60°angle) to perpendicular to the dip of the strata. Aggradational channel deposits are generally 1.5-3 m thick. In multi-storey sandbodies, they are generally observed at the basal part and are eroded by the accretionary bars; however, they are also observed besides the accretionary bars (Fig. 7b). This facies association consists of grey to yellowish grey, fine-to medium-grained sandstone with rare locally persistent intraformational conglomerate observed at the base of the sandbodies. Sandstones display tabular, trough, lowangle cross-stratification and ripple cross-stratification with some massive and rarely sigmoidal cross-stratification. The cross-beds contain mud pebbles, carbonaceous and mudstone drapes, and wood fragments converted to coal (up to 30% in a few places). Cross-bed set thickness decreases upward from more than 1 to 0.1 m. Locally 10 cm to meter-scale syn-depositional minor faults were observed in both the core and outcrop sections. Slumped and water-escaped structures were also occasionally present.    Table 1.  Bioturbation was not observed in the studied section except for the presence of rootlets in the channel top. These bodies are interpreted as the deposits of large, dominantly meandering, fluvial channels. The horizontally stratified deposits, which make up a minor proportion of the succession, were deposited by the aggradation process in the base of vertically aggrading channels, while the bulk of the sedimentation took place within accretionary bars. The oblique (>60°angle) to perpendicular relationship between the accretion direction and the bedform palaeocurrent direction along with the lithofacies assemblage of the accretionary bars deposits indicate that these are lateral-accretion units deposited in the point bars (Miall 1996). Conglomerates at the base of channels represent bypass lags deposited during the cut of the channel while thick, intra-channel mud-clast conglomerates are the result of intraformational erosion and bank collapse. Facies association 2 (FA2): heterolithic channel deposits Facies association 2 (FA2) is composed of heterolithic lateral accretion elements, equivalent to inclined heterolithic stratification (IHS) of Thomas et al. (1987) and epsilon cross-stratification units of Allen (1963) and aggradational heterolithic fills (Fig. 7). Two types of heterolithic deposits are observed: fluvial heterolithic deposits (facies association 2A) and tidally influenced heterolithic channel deposits (facies association 2B).

Facies association 2A (FA2A): fluvial heterolithic deposits
Facies association 2A (FA2A) is composed of fine-to mediumgrained trough, tabular and ripple cross-stratified sandstone, alternating with thin mudstone layers (Fig. 7c). Bioturbation was rarely observed in the studied sections. Soft sediment deformation consisting of slump structures, convolute lamination and waterescaped features was observed locally. Abundant rip-up clasts and lignite/coal fragments were also present. A lateral fining trend both toward and away from the depositing palaeochannel was present in several sections, interpreted as point bars.
The IHS successions are sharp or erosive-based, tabular shaped bodies which are 2-3.5 m thick and up to 120 m wide. The crossstrata are centimetre-to decimetre-scale thick and show a low dip, in the range 4-15°. The IHS sequences were generally observed at the top of FA1 and associated with FA3 and FA6. They were studied in the Saltwick Formation, exposed at the Kettleness, Whitby East and Beacon Hill sections.
IHS are typically deposited in point bars of sinuous meandering streams (Thomas et al. 1987). The alternation of mudstone and sandstone and the hierarchical arrangement reflect a superimposition of several distinct types of depositional events. The absence of abundant bioturbation and bidirectional sedimentary structures and the association with proximal delta plain deposits indicate no marine (tidal) influence. The observed characteristics suggest that the alternating sand and mud layers were deposited as a result of fluctuating discharge at the late stage of sinuous palaeochannels (McGowen and Garner 1970; Thomas et al. 1987).

Facies association 2B (FA2B): tidally influenced heterolithic channel deposits
Facies association 2B (FA2B) is also composed of alternating sandstone and mudstone layers with additional tidal imprints. The grain size ranges from fine-to medium-grained sandstone with abundant mud drapes and alternating bioturbated mudstone layers ( Fig. 7d-f). FA2B exhibits a high bioturbation index (Taylor and Goldring 1993), up to BI 4 in mud layers and typically low in crossstratified sandstones (BI 1). Inertinite-dominated palynofacies and rare Acanthomorph Acritarchs microfossils are reported by Fisher and Hancock (1985) in an exhumed meander belt of the Scalby Formation. Abundant rip-up clasts and lignite/coal fragments are also present.
Sandstone layers are trough, tabular and ripple cross-stratified; alternating mudstone layers are thinly laminated and bioturbated (Fig. 7e). Cross-strata dip from 9 to 25°and strike at 200-286°. Measured paleocurrent ranges from 008 to 354°(average 170°). Soft sediment deformation structures are also observed in both sandstone and mudstone strata. The relative percentage of sandstone and mudstone varies vertically and laterally within the accretionary bars. Generally, the sand grain size and proportion decrease in the downdip portion. Occasionally, the aggradational heterolithic fills occur beside the IHS. Locally in the Long Nab foreshore and Scarborough South Bay cliff section, there are packages of IHS which occur adjacent to concave cut-banks. These are more shallowly dipping than the typical IHS.
Most of the studied heterolithic deposits are concave-upward, erosionally based and locally overlain by a thin, coarse sandstone layer, and exhibit an overall fining upward trend. The individual IHS bars are up to 4 m thick and more than 100 m wide. They are sigmoidal in shape, and vertically and laterally stacked with other IHS accretion bars. In the studied sections, the sandstone strata are 10-40 cm thick alternated by 5-10 cm thick mudstone strata.
These IHS were also deposited in point bars in sinuous meandering streams. Alternation of mudstone and sandstone reflect a rhythmic variation in the hydrodynamic conditions during deposition. The presence of mud drapes, alternating sandstone and mudstone layers, abundant bioturbation and saltwater microfossils suggest proximity to a palaeoshoreline and a degree of tidal influence in sedimentation (Fisher and Hancock 1985;Wood 1985;Wood et al. 1988). The lower angled bar on the concave side of the channel may represent counter-point bars which form at the downstream tails of conventional bar forms in slightly confined and translating point bars (Alexander 1992b;Smith et al. 2009).

Facies association 3 (FA3): mudstone-dominated channel deposits
Facies association (FA3) is composed of planar laminated mudstones which are superficially similar to the overbank associated FA6. FA3 occur within channel-shaped scours and are commonly associated with FA1 or FA2 (Fig. 8a). They exhibit a weakly developed fining-upward trend from a siltstone-dominated lower part to an organic-rich, claystone-dominated upper part. Bioturbation is not observed, but abundant millimetre-scale lignite /coal fragments are present. They are relatively small in size: 2-4 m thick and 30-40 m wide.
The geometry and composition of FA3 suggest deposition from suspension within channel-shaped depressions. These are interpreted as the fill of channels which are abandoned during avulsion (Smith et al. 1989;Alexander 1992a;Bridge 2006;Nichols 2009). Where present, the fining-upward sequence suggests a gradual rather than a sudden avulsion (Alexander 1992a). In the absence of a lateral context (i.e. a cliff section), these mud plugs are difficult to distinguish from regular overbank deposits although they are commonly more organic-rich, suggesting standing bodies of stagnant water. Where the geometry can be observed they provide dimensions of the parent channels (Li and Bhattacharya 2014).

Overbank deposits
Overbank sediments constitute the dominant proportion of most of the fluvial successions exposed in Yorkshire coastal sections. The overbank successions mostly consist of mudstone, fine-grained sandstone and mud-dominated heterolithic deposits. The sandstones typically occur in crevasse splays and crevasse channels (FA4), and levee deposits (FA5). Overbank mudstones are composed of grey to dark grey mudstones with abundant organic matter and lighter grey mudstones occasionally with root traces (FA6). The pedogenically modified mudstones and sandstones are discussed within FA7.
Facies association 4 (FA4): splay sheet and crevasse channel sandstone deposits Sharp-based, thin sand sheets and associated small ribbon-shaped sandstone deposits are common elements in the overbank of the fluvio-deltaic successions of the Ravenscar Group. Facies association 4 (FA4) is comprised of splay sheet deposits (facies association 4A) and crevasse channel deposits (facies association 4B).
Facies association 4A (FA4A): splay sheet deposits Sheets are typically up to 2 m thick but may be up to 6 m in extreme cases. They are composed of very fine to fine-grained, currentrippled cross-and planar-parallel-stratified (occasionally structureless), sharp-based, rooted sandstone (Fig. 8b, c and e). The larger sheets are comprised of a series of beds which represent individual events. They are sharp-based and show an upward gradation from massive to planar bedding to current ripples. These sandstone beds are commonly interbedded with thin finer grained facies and palaeosols. The top surface of the sheets is commonly rooted, suggesting subaerial exposure.
Splay sheets commonly occur as isolated bodies within FA6/FA7 or occasionally they are associated with major channel deposits (FA1 and FA2). The thickness, grain size and flow energy reflected from the sedimentary structures within the sheets decrease away from the channel, i.e. from the proximal to distal part. In the proximal part, several 30-60 cm thick sandstone beds are amalgamated to form a splay body. From the medial to distal part, the grain size decreases progressively, and individual sandstone beds are separated by 5-12 cm thick, silty, very fine sandstones or mudstones (Appendix D).
The sandstone sheets are interpreted as crevasse splay deposits. Each sandstone bed within the sheet represents a single flooding event. The tractional sedimentary structures dominated laterally extensive sand sheets in the overbank are interpreted as rapid deposition from the unconfined flow during overbank flooding (Burns et al. 2017;Gulliford et al. 2017). Upward gradation of large sheets from massive to planar bedding to current ripples suggest waning of flow strength (Steel and Aasheim 1977), followed by the deposition of the suspended particles from the standing flood water. Rooting of the beds suggests that the individual flows are intermittent, and more extensive pedogenesis at the top of the sheet represents plant colonization after abandonment. Sheets are arranged into a variety of genetically related architectural elements that are discussed in the architecture section below.

Facies association 4B (FA4B): crevasse channel deposits
The small ribbon deposits have a concave shape and are up to 3 m thick but typically less than 2 m. They are erosive-based and flanked on either side by splay sheets (FA4A). They are comprised of fineto medium-grained tabular, trough and ripple cross-stratified sandstones. Similar to FA4A, they are commonly associated with FA6/FA7 or occasionally with FA1/FA2.
The erosive-based, ribbon-shaped sandstone deposits flanked with splay sheets are interpreted as crevasse channels that fed the crevasse splays. The erosive base and concave-up shape of these overbank ribbon-shaped sandbodies suggest deposition from the turbulent and confined flows by scour and fill processes after flood breach of the parent channel (Fielding 1986;Bristow et al. 1999;Nichols and Fisher 2007;Burns et al. 2017;Lepre 2017).

Facies association 5 (FA5): mudstone-dominated heterolithic levee deposits
Facies association 5 (FA5) is composed of alternating very fine silty sandstones and mudstones (Fig. 8d). The thin silty sandstone layers show current ripple and planar-parallel stratification while the alternating mudstone layers contain parallel laminations with occasional deformation structures. Occasionally, wave-ripple lamination, planar lamination, and bioturbation are also observed. Deposits are up to a few metres thick and occur in laterally restricted bodies. They border the channel deposits and are gently inclined, thinning away from the channel body. The grain size also decreases away from the channel body.
The wedge-shaped geometry, mud-dominated heterolithic composition, unidirectional small-scale sedimentary structure and occurrence adjacent to the channel deposits suggest that these are levee deposits. They form as a result of successive small-scale overbank spilling of the channel without the development of crevasse splays (Elliott 1974;Bridge 1984). Local occurrence of symmetrical ripples and alternating dark grey mudstone suggest that they may have been deposited in a subaqueous setting.
Facies association 6 (FA6): overbank mudstone deposits FA6 is comprised of grey and dark grey to black coloured mudstone and sporadic very fine-grained sandstone lamina (Fig. 8e, f). The mudstones are poorly planar-parallel laminated or structureless and bioturbated (BI = 2-3). Subordinate small calcareous nodules and minor desiccation cracks are occasionally present. Individual mudstone layers are 0.5-3 m thick and laterally persistent. Often they are dark in colour and contain abundant organic matter with occasional very thin laterally restricted coal layers (Fig. 8e). The light grey coloured mudstones contain less organic matter (Fig. 8f ).
The freshwater bivalve Unio kendalli was observed within the lower part of the Saltwick Formation, 3 m above the formation base, 4 km south of Whitby (Hemingway 1974). At Hayburn Wyke, several marine microplankton were observed by Fisher and Hancock (1985) in the mudstones of the Saltwick Formation. Abundant Unio hamatus and Unio distortus occur with the Scalby Formation floodplain near Scarborough, Gristhorpe and Burniston (Hemingway 1974).
Laterally persistent, massive to planar-parallel laminated, thinly bedded mudstone suggests an overbank origin. Deposition in subaqueous conditions such as shallow-water lakes or floodplain ponds is supported by the dark grey to black colour, plane-parallel lamination, organic-rich mudstone with sporadic very fine-grained sandstone lamina, bioturbation and fossil assemblages. The light grey massive to poorly laminated mudstone with very little organic matter content and rootlets at the bed top is indicative of subaerial overbank deposition. The calcareous nodule, bioturbation and rootlets indicate pedogenic modification after deposition. Freshwater to brackish fauna suggest a minor marine influence in the southern part of the section.

Facies association 7 (FA7): palaeosol
Facies association 7 (FA7) is comprised of grey to dark grey and black mudstone to fine sandstone. This facies association is allied with FA1, FA2 or FA4 which it typically overprints. Deposits are characterized by abundant rootlets (Fig. 8c), siderite nodules and siderite spheruliths. There are also locally organic-rich horizons, nodules, bioturbation and colour mottling (Fig. 8g). The degree of modification from the original facies is varied. Examples such as those at the base of the Saltwick Formation are very well developed and 'mature' with the vertical grain-size trend, while those in the upper part of Saltwick and Scalby formations are less well developed (Fig. 8h).
Based on the presence of pedogenic features such as abundant roots, colour mottling, blocky texture and spherosiderites, this facies association is interpreted as a palaeosol. The presence of palaeosols was also described by Kantorowicz (1990) in the Long Nab Member at Burniston Bay, Alexander and Gawthorpe (1993) in the Salwick Formation at Whitby West Cliff and Rawson and Wright (1992) in both the Scalby and Saltwick formations. Two types of palaeosol are defined based upon the degree of paedogenesis: (i) comparatively mature palaeosols and (ii) poorly developed palaeosols. The former are characterized by the presence of abundant rootlets and extensive pedogenic modification. The latter are a darker grey and have sparse rootlets, and abundant carbonaceous material indicating their waterlogged origin. Livera (1981) suggested the immature palaeosols formed in waterlogged swamps. The observed palaeosols are limited and sporadic in nature, which suggests the exposed palaeosols may be the erosional remnants of more extensive floodplain deposits (Alexander and Gawthorpe 1993).

Sandstone body architecture
A total of 83 channel bodies, 191 splay bodies and eight crevassechannel bodies were identified and studied. Sandbody widths range from 15 to 2038 m and the thickness measured both in the field and in virtual outcrop ranges from 0.5 to 28 m with W/T ratio 15->541 (Fig. 9). Sandbody geometry, dimension and proportions change within and between formations. Sandbodies are broadly divided into channel sandbodies consist of FA1 and FA2, and overbank sandbodies consist of FA4 and FA5.

Channel-belt facies architecture
Channel bodies are typically subdivided based on their internal stacking patterns and/or their aspect ratio. Stacking patterns include single-storey, multi-storey and multilateral (Gibling 2006;Owen et al. 2017). Single-storey channel bodies are deposits with no internal storey surfaces in which the thickness of the deposit equates to the depth of the channel at the time of deposition (minus compactional effects) (Friend et al. 1979;Hirst 1991). Multi-storey bodies are vertically stacked channel bodies where multiple storeys are separated by erosional storey surfaces (Potter 1967). Multilateral bodies are formed by the lateral coalescence of several channel bodies representing the lateral migration of the fluvial system (Rittersbacher et al. 2014). The aspect ratio typically separates ribbons (W/T: <15) from sheets (W/T: >15). We adopted and modified terminologies from the classification scheme of Hirst (1991) and Owen et al. (2017), and defined five different types of channel body: (i) single-storey small ribbon (W/T: 5-10); (ii) single-storey broad ribbon (W/T: 10-<15); (iii) tabular multilateral sheet (W/T: >15-30); (iv) internally stacked multistorey sheet (W/T: 20 ->100); (v) partially amalgamated multistorey sheet (W/T: >15->102) (Fig. 10).
(i) Single-storey small ribbon channel bodies are isolated channel deposits with no internal storey surfaces. The channel bodies are comprised of FA1, FA2 and FA3. They represent the simplest classic channel geometry and the smallest channel dimensions (Fig. 10a-i and b-i). Thicknesses vary from 4-5.5 m and width from 25-40 m with W/T 5-10.
(ii) Single-storey broad ribbon channel bodies are larger than the single-storey small ribbon. They are 30-70 m wide and 3-5.5 m thick with W/T 10-<15. They are comprised of either single bar (FA1 and FA2; Fig. 10a-iia and b-iia) with abandoned mud fill (FA3) or multiple accretion bars formed by laterally migrated single-storey channel (Fig. 10a-iib and b-iib) followed by either aggradational fills of FA1 or FA3. (iii) Laterally stacked channel deposits consist of tabular multilateral sheet sandbodies and are bounded by flat-lying erosional basal surfaces and flat tops (Fig. 10a-iii). They are 180-> 300 m wide and 5-10 m thick with W/T up to 30. They may be filled with either inclined heterolithic strata (IHS, FA2) or accretion bars and aggradational fill of FA1. This channel type is observed in cliff sections of the Saltwick and Scalby formations. In the Saltwick Formation, several multilateral channel bodies were observed just above the Dogger Formation. The well-exposed 9 m thick and >135 m wide channel body c. 2.5 km SE of High Whitby provides a good example of this type of lateral stacking pattern in cross-section ( Fig. 10b-iii). The exhumed meander belt deposits of the Scalby Formation exposed between Long Nab and Scalby are excellent examples of this type of channel architecture in plan-view (Alexander 1992b). Ielpi and Ghinassi (2014) interpreted these foreshore outcrops to represent two vertically stacked multilateral channel bodies.
(iv) Internally stacked multi-storey sheets have a simple external geometry like multilateral channel bodies, but the internal stacking pattern is more complex. Both the multi-storey and multilaterally stacked channel deposits overlie a single major basal surface (Fig. 10a-iv).
Channel bodies are 600->1000 m wide and 9-20 m thick with a W/T ratio up to 100. They consist of both laterally and vertically stacked deposits of FA1 with occasional FA2 and FA3. Channel bodies typically comprise 3-4 stacked storeys in the vertical section and the storeys are 130-415 m (observed width) wide and 2.5-7.3 m thick. The original storey widths and thicknesses may have been greater but are eroded by younger channels that truncate the older channel storey. The channel deposits of the Saltwick Formation exposed at the High Whitby section are an example of the internally amalgamated multi-storey sheet (Fig. 10b-iv).  interpreted this channel as a multi-storey channel body with complex and long-lived history. The Moor Grit channel bodies at Hundle Point and the Scarborough South Bay sections are also a good example. These channels were interpreted by Alexander and Gawthorpe (1993) as complex multilateral sheets but are reclassified here based on the mapped storey surface pattern and major incisional basal surfaces.
(v) Partially amalgamated multi-storey sheet complexes are channel complexes where vertically and laterally stacked channel bodies are partially connected (Fig. 10a-v) with pockets of floodplain mudstones in between the sheets. This channel body geometry is generally comprised of FA1, FA2 and FA3. The channel bodies are very thick and wide; thickness ranges from 16 to 28 m and width from >312 to >2038 m with W/T up to >102. A combination of multi-storey and multilateral storey patterns are observed. Within individual channel bodies, stories are both amalgamated and may cross-cut one another. Individual storeys are 100->190 m wide (observed width) and 2.5-8.2 m thick.
Channel bodies at Beacon Hill Cliff (Fig. 10b-va), Whitby West Cliff ( Fig. 10b-vb) and Kettleness in the Saltwick Formation, and at Ravenscar in the Sycarham Member of the Cloughton Formation are examples of this type. The multi-storey offset channel deposits (type-3bi) and stacked sandbody with no obvious offset migration (type-3bii) of Alexander and Gawthorpe (1993), and the amalgamated complex of Hirst (1991) are similar to the partially amalgamated multi-storey sheet complex geometry. In partially amalgamated channel bodies, it is difficult to discern the individual channel/channel body. This feature along with the existence of pocket floodplain deposits suggests that the aggradation rate was comparatively low and the younger channels or channel bodies incised into the older channel bodies.

Overbank facies architecture
Crevasse channels and splays (FA4) are the major overbank constituents in all of the formations. A total of five types of sandbody architectures were observed in the overbank settings: (i) ribbon crevasse channel, (ii) single-splay sheet, (iii) amalgamated splay sheet, (iv) splay sheet complex and (v) wedge levee deposits (Fig. 11).
(i) The crevasse channel bodies are composed of ribbon sandstones of FA4. They are commonly 1.5-2.5 m (average 2 m) thick and 10-25 m (average 20 m) wide with W/T ratio less than 11 (Fig. 11a-i and b-i). They have an erosive base and cut into overbank mudstone, especially in the proximal or proximal medial part of the splay. They are interpreted to represent the channels that fed the splay deposits during flood events.
(ii) Single crevasse splays consist of FA4. They are small-scale, 0.3-2.2 m (average 1 m) thick and >20-300 m (80 m) wide with W/ T ratio >20-540 (average 95). They are lenticular or sheet-like sandbodies encased in overbank mudstone or bounded by palaeosols (Fig. 11a-ii and b-ii). They are interpreted to form as a result of overbank flooding and sediment deposition during excess discharge of single or multiple events through a single breach point in the channel. The lack of internal palaeosols or rooting suggests these features are relatively short-lived.
(iii) Coalesced moderate-to large-scale elongated ( parallel to the feeder channel) crevasse splay sandbodies are termed amalgamated splays. These can be subdivided into three different types based upon the stacking of the constituent splays. They may be vertically amalgamated, laterally amalgamated, and both laterally and vertically amalgamated (Fig. 11a-iii and b-iii). Vertically amalgamated splays are formed by an amalgamation of two generations of splay where >50% of the younger splay base is in contact with an older splay. Vertically stacked splays are >35-780 m (average 320 m) wide, 140 m long and 0.6-3.8 m (average 2 m) thick with W/T ratio >15-265 (average 85). They are recognized by sharp/erosional contact or rooted horizons between the splay bodies along with more than one coarsening or finingupward trend (Fig. 11a-iiia and b-iiia). They are generally thicker than the single and laterally amalgamated splay body. Laterally amalgamated splays are elongated; connected splay sheets form parallel to the feeder channel where splay-to-splay contacts are <50% (Fig. 11a-iiib and b-iiib). They are >70-930 m (average 270 m) wide, 0.53-2.2 m thick with W/T ratio 85-600 (average 200). They are generally larger than the single crevasse splays, and in cross-section, they have a lensoid geometry.
Laterally and vertically amalgamated splay bodies are stacked splay bodies where splay bodies are connected laterally and vertically separated by pockets of floodplain deposits. They are >50 to 1170 m wide, 2.5-5.8 m thick with W/T ratio >30-275 (average 150) (Fig. 11a-iiic and b-iiic). This type of splay body is interpreted to be deposited by aggrading sinuous feeder channels from multiple input points.
(iv) Splay complexes are large-scale >1285 m wide (length and width up to several km; 20 km described by ), up to 6.5 m thick with W/T ratio >240. They comprise several upwardcoarsening packages and crevasse channel deposits (FA4) (Fig. 11a-iv and b-iv). Sandstone bedsets are separated by thin, bioturbated to plane-parallel laminated mudstone. Splay complexes were observed in the Gristhorpe Member at Cloughton Wyke and Ravenscar. At Cloughton Wyke, clear tidal indicators such as planeparallel stratified sandstone with double mud drapes, flaser, wavy and lenticular beddings were observed. Sedimentary structures along with palynofacies analysis of Hancock and Fisher (1981) indicate the marine influence and deposion in a tidal influenced delta plain.
(v) Natural levees are wedge-shaped, laterally restricted, gently inclined, channel-associated bodies (Fig. 11a-v and b-v). They are rarely observed, 30-50 m wide and 2-3 m thick with W/T ratio c. 20. Levees are entirely composed of FA5. They are thicker in the channel proximal part and gradually taper away from the parent channel.
The studied facies associations and sandbody architectural patterns indicate that the non-marine successions of the Ravenscar Group were deposited in a fluvio-deltaic depositional condition (Fig. 12) dominated by highly sinuous channel deposits.

Temporal variations in facies and sandbody architectures
Facies and sandbody architectures vary widely within and between the formations. VO data provide an opportunity to map facies variations and trends in the fluvial stacking patterns and architecture. In this study, we adopted the methodology of Rittersbacher et al. (2014) for looking at lateral and vertical trends in facies associations and channel body dimensions from VO data. Individual channel bodies were identified, mapped and quantified. Their vertical position relative to a stratigraphic datum was recorded along with the proportion of channels in a section. Their lateral position in a depositional dip profile and with respect to the major structures was also noted. Lateral and vertical trends were studied over a 40 km long, oblique (7°-45°except Whitby West Cliff where it is 90°) depositional dip section from Kettleness in the NW to Scarborough South Bay in the SE (Fig. 13). The detailed vertical variations of channel body width, thickness and cross-sectional area  were studied by plotting these parameters against their stratigraphic height measured from the top of the Dogger Formation (Fig. 14). These data were binned into 10 m thick packages to remove local lateral variations intrinsic to channel bodies. The majority of the channel bodies are between 50-500 m wide and 3-12 m thick (Fig. 14a, b). In the following sections the facies, facies associations and architecture along with their lateral and vertical variations in the fluvio-deltaic formations that make up the Ravenscar Group are systematically described.

Facies architecture of the Saltwick Formation
The Saltwick Formation was studied at six outcrops along 14.5 km of very steep coastal cliffs (Appendix E). The base of the formation is marked by an erosional unconformity on top of the marine sandstone deposits of the Dogger Formation or locally the Whitby Mudstone Formation (at Kettleness; Appendix E). The top of the formation is a marine flooding surface at the base of the shallow marine Eller Beck Formation. The Saltwick Formation varies in thickness from 28 to 45 m.
Ten field-based sedimentary logs and 96 virtual logs were used to study the facies proportion. All the facies associations (FA1-FA7) were observed in this formation. The proportion of coarse-grained facies (FA1, FA2, and FA4) in the formation ranges from 38-95% (avg. 65.7%) and the mudstone facies proportions range from 5-62% (average 34.3%) (Fig. 15). At the Kettleness, Whitby West, and in most parts of the Beacon Hill and Ravenscar sections, the Saltwick Formation is dominated by FA1 (up to 95%; Fig. 15). At the Whitby East and locally at the Beacon Hill sections it is dominated by FA6 (up to 75%) and FA4 (up to 28%) (Fig. 15).
A total of 41 channel bodies and 134 crevasse splay bodies were identified and studied. The corrected dimensional properties of the sandbodies are given in Table 3 and Figure 16. Most of the channel bodies are multi-storey and multilateral (n = 34) and show partially amalgamated, internally stacked and tabular multilateral geometries. They are >30-2038 m wide and 5-20 m (average 8 m) thick (Fig. 16). There are fewer single-storey channel bodies (n = 7) which are composed of FA1-FA3. They show both the small and broad ribbon geometries (Fig. 16). The width of the channel bodies ranges from 42-62 m (average 50 m) and thickness from 3-3.5 m (average 3.5 m) (Fig. 16). Crevasse splay bodies consist of the single and amalgamated splay bodies (overbank architectural type-ii and type-iii). Single crevasse splay bodies are less extensive and show lower W/T ratio than the amalgamated splay bodies (Table 3; Fig. 17).
In the Saltwick Formation, the channel body maximum thickness decreases from 28 m in the Kettleness outcrop in the NW of the study area to 11 m in the Hayburn Wyke outcrop in the SE of the area. The resultant coarse-grained facies proportion decreases from 95 to 38% towards the SE (Fig. 13). Within this regional trend, there are some local variations. The Kettleness, Whitby West and Ravenscar (east of Peak Fault) outcrops are dominated by multistorey channel-fill deposits whereas the Whitby East and Ravenscar Quarry Cliff sections (west of the Peak Fault) are dominated by overbank mud and crevasse splay deposits with a few small-channel bodies. Both channel and overbank dominated parts are present in the Beacon Hill Cliff section (Appendix E). Figure 18 is a 22 km long cross-section which includes 46 virtual logs to illustrate the lateral architectural and facies variations. These new data indicate that some thick multi-storey channel bodies are fault-bounded (at the Whitby West and Ravenscar sections) and others are not (at the Kettleness and Beacon Hill (Black Nab to Widdy Head) sections). Average channel body dimensions (width and thickness) in the Saltwick Formation increase until 20-30 m stratigraphic height and then decrease in the upper part of the formation (Fig. 14c-f ).
The overall thickness of the Saltwick Formation increases towards the SE from 28 m at Kettleness and Whitby East to 43 m at Ravenscar. The formation thickness varies locally around the Peak Fault in Ravenscar (Fig. 19) where the average thickness measured on the hangingwall side of the Fault is 43 m (max. 45 m due to channel scouring). In the footwall side, Pliensbachian Staithes Sandstone and Redscar Mudstone formation are juxtaposed against the Dogger and Saltwick formations on the hangingwall side (Fig. 19b, c). Over 50 m of Toarcian (Lower Jurassic) deposits of the Blea Wyke Sandstone Formation and the upper part of the Whitby Mudstone Formation are absent in the footwall section. The Eller Beck Formation is not exposed in the footwall part of the section, so the full Saltwick Formation thickness cannot be determined (Fig. 19d, e). A 1 m thick Dogger Formation is exposed at Quarry Cliff section (1 km west of Coastal Cliff exposure) in the footwall block. The formation thickness increases from 1 m in footwall to 19 m in hangingwall.

Facies architecture of the Cloughton Formation
The Cloughton Formation was studied in five outcrop sections which include 12.8 km of the cliff line. Only the lower part of the formation, i.e. part of the Sycarham Member, is exposed in the Whitby East and Beacon Hill sections. The complete formation is exposed at the Ravenscar, Hayburn Wyke and Cloughton Wyke Cliff sections although the exposures are partially covered in vegetation. The formation thickness measured in the Ravenscar section was c. 80 m.
Four field-based sedimentary logs and 26 virtual logs were collected to determine the facies proportion. All of the facies associations were observed in this formation. The average coarsegrained facies (FA1, FA2 and FA4) proportion ranges from 58-70% and the fine-grained facies (FA3 and FA6-7) from 30-42% (Fig. 15). In the Sycarham Member, the predominant facies constituent is FA1 (up to 59%), and in the Gristhorpe Member, the primary constituent is FA4 (up to 53%).
A total of 20 channel bodies and 21 crevasse splay bodies were identified and studied. The dimensional properties of the sandbodies are given in Table 3 and Figures 16 and 17. The majority of the channel bodies (n = 15) are multi-storey and represent partially amalgamated or internally stacked channel body architectures. They are composed of FA1-FA3, and are 25->765 m wide and 5-24 m (average 8 m) thick (Fig. 16). All of the observed multi-storey channel bodies occurred in the Sycarham Member, and only two single-storey channel bodies were studied in the Gristhorpe Member. The single-storey channel bodies are composed of FA1 and FA2. They are 30-70 m (average 42 m) wide and 3-4.5 m (average 3.9 m) thick ( Fig. 16) which show both the small and broad ribbon geometries. Three different types of splay body architectures were observed in this formation. The single-splay bodies are 46-295 m (avg. 120 m) wide, 68 m long and 0.5-2 m (average 1.1 m) thick. Laterally amalgamated splay is incomplete and >68 m wide and 1.28 m thick. The crevasse splay complexes are 175->1285 m (avg. 645 m) wide and 3-6.5 m thick (Fig. 17).
In the Sycarham Member of the Cloughton Formation, channel body thickness decreases from c. 20 m in the Beacon Hill outcrop (NW) to c. 10 m in the Cloughton Wyke outcrop (SE; Fig. 13). The coarse-grained facies proportion also decreases from 70 to 60% in the respective sections. Average channel body dimensions (width and thickness) in the Sycarham Member of the Cloughton Formation show an upward increase (Fig. 14c-f ). Channel body cross-sectional area (width × thickness) also shows a similar trend.

Facies architecture of the Scalby Formation
The Scalby Formation was studied in 4 km of coastal cliff sections at Long Nab, Scarborough South Bay, Cromer Point and the northwestern part of Cloughton Wyke. The complete vertical section of the formation is not exposed in any of the studied    sections. The stratigraphic thickness of these outcrops ranges from 28-40 m. The beds dip towards the south, and in the northwestern sections, the lower, sand-dominated part of the formation (Moor Grit Member) is exposed at the top of the cliff while towards the south, the Moor Grit passes into the subsurface and the cliff sections are comprised of the finer-grained Long Nab Member (Appendix E). A total of 26 field-based sedimentary logs, 27 borehole core logs and 20 virtual logs were studied. The average coarse-grained facies (FA1, FA2 and FA4) of the formation ranges from 60-70%, while the fine-grained facies (FA3, FA6 and FA7) range from 30-40% (Fig. 15). FA1 is the predominant facies constituents (up to 46%) of the formation and FA2, and FA4 individually comprise 7-8%.
A total of 21 channel bodies and 36 crevasse splay bodies were identified and studied. Channel body widths were corrected in relation to the palaeocurrent direction of individual channel bodies (135-225°) when data were available; otherwise, the predominant palaeocurrent direction (170°) was used. The dimensional properties of sandbodies are given in Table 3 and Figures 16 and 17. The majority of the studied channel bodies (n = 15) are multi-storey and mainly represent internally stacked and tabular multilateral sheet architectures. They are composed of FA1 and FA2 and are 50->1000 m wide and 4.5-10 m thick with W/T ratio 15->100. The single-storey channel bodies are also composed of FA1 with occasionally FA2 and FA3. They were observed in the upper part of the formation, and show both the small and broad ribbon architectures (Appendix E).
No lateral trends in channel body dimensions were observed in the Scalby Formation, but an increase in the fine-grained facies proportion and the presence of IHS and other evidence for marine influence, towards the SE were seen (Fig. 13). There is a continuing upward decrease in average channel body dimensions in the Scalby Formation which is slightly interrupted between 170-180 m above datum by a comparatively thick channel body at Burniston Bay. The channel body W/T ratio and cross-sectional area (width ×thickness) also show a similar trend (Fig. 14e-h).

Discussion
The data collected during the course of this study allow a unique insight into a series of sections that have already been extensively studied (Hemingway and Knox 1973;Leeder and Nami 1979;Livera and Leeder 1981;Alexander 1992a, b;Eschard et al. 1992a;Alexander and Gawthorpe 1993;Ielpi and Ghinassi 2014). The addition of VO data and the integration of VOs with outcrop and behind-outcrop sedimentary logs allows quantification of the deposits that has not been previously possible. These data have been used to address four issues for these sections: (i) the sequence stratigraphy, including vertical and downdip trends in fluvial stacking and architecture, (ii) the role of syn-depositional faulting in locally controlling the depositional systems, (iii) architecture of overbank sandbodies and their role in paralic reservoirs, and (iv) comparison of sandbody dimension with analogues. These aspects are discussed below.

Sequence stratigraphy and channel-stacking patterns
In this study, we recognize two unconformity bounded sequences which consist of smaller parasequences bounded by marine flooding surfaces. The lower sequence, termed here Sequence A, is bounded at the base by the angular Mid-Cimmerian Unconformity (Hemingway 1974;Alexander 1989;Underhill and Partington 1993) and separated from the overlying Sequence B by the Late Bajocian unconformity (Powell 2010;Fig. 4). Both the sequences are type 1 sequences of Van Wagoner et al. (1988). Sequence A is Aalenian to Bajocian in age, comprises the majority of the Ravenscar Group and contains five parasequences (PS 1-5; Fig. 4 and Table 4). There is an increase in the proportion of marine deposits within successive parasequences, suggesting they are part of a long-term transgressive trend. The overall decreasing channel proportion and dimensions, and increasing overbank mud proportion, also reflect the retrogradational trend (Table 4).
Sequence B unconformably overlies Sequence A and represents a return to a fluvio-deltaic setting. This sequence is comprised of the Scalby Formation. Earlier workers have ascribed a significant time gap to the unconformity at the base of the Scalby Formation (Powell 2010), although a palynological study by Riding and Wright (1989) suggested only a minor stratigraphic break. Regardless of the duration, erosively based, multi-storey amalgamated channel deposits abruptly overlie marine deposits, suggesting a major reorganization of the depositional system. Vertical trends in channel body proportions and dimensions in Sequence B also show a retrogradational stacking pattern (Table 4). The following section discusses the channel stacking patterns within the two sequences.
Fluvial stacking patterns are primarily controlled by two key factors: the accommodation to sediment supply ratio (A/S), especially with respect to fluvial base-level (Wright and Marriott 1993;Shanley and McCabe 1994;Holbrook et al. 2006) and the position within the depositional system, especially in distributive fluvial systems (DFSs; Hartley et al. 2010;Weissmann et al. 2010Weissmann et al. , 2013Rittersbacher et al. 2014;Owen et al. 2015). Stacked multistorey fluvial sandbodies such as those that occur in the lower parts of the Saltwick Formation (PS 1) and the upper part of Sycarham Member (PS 2), typically represent a low A/S where the rate of accommodation creation is low and/or sediment supply is high. Conversely, systems with small, single-storey channels and a high proportion of preserved floodplain fines such as the Gristhorpe (PS 3) and Long Nab (PS 6) members are more typical of systems with very high A/S where rapid subsidence preserves more of the section from reworking by channels.  suggested that low accommodation as a result of low subsidence or sea-level fall associated with high sediment supply favours vertically incised channel and channel avulsion in a stable channel belt forming these stacked multi-storey channel bodies, whereas the high A/S lead to shallower channel incision and less stable channel formation which creates isolated, low width-to-thickness of the channel body. Alexander (1989) suggested that in a low-lying delta plain like the Cleveland Basin, fluctuation of relative sea level plays an important role in a fluvial stacking pattern. The vertical and lateral stacking of the channel bodies and the relationship to the flooding surfaces and syn-sedimentary faults within Sequence A are considered further below.
Previous authors (Eschard et al. 1992a) have suggested that the Moor Grit Member at the base of the Scalby Formation represents an 'incised valley fill'. This interpretation suggests that the A/S is negative for a period of time and, following a relative base-level fall, a valley was cut and sediment was bypassed further downdip. In such a model, the multi-storey fluvial deposits of the Moor Grit Member are laid down as base-level starts to rise, creating accommodation. A further increase in the rate of accommodation results in a transition to more heterolithic channel-fill deposits which may be tidally influenced at the base of the Long Nab Member. Further increase in the A/S resulted in the deposition of the overlying, low-net:gross upper part of the Long Nab Member. The depositional model of Sequence B is similar to the model proposed by Shanley and McCabe (1993) where the base-level change controls the fluvial architecture. For the interpretation of the incised valley to be confirmed, it would be necessary to map out valley margins and interfluves and to map a lowstand shoreline further basinward. While this is not possible with the data available, the valley model is supported by (i) the presence of a biostratigraphically confirmed unconformity at the base (Leeder and Nami 1979) and (ii) a demonstrable change in accretion bar style from clean sandstone-dominated thick channel bars at the base (basal part of Moor Grit Member) to the downstream-translated point bars at the middle to the heterolithic deposits of the less confined expanding point bar at the upper part (Eschard et al. 1992a;Ielpi and Ghinassi 2014).

Syn-tectonic sedimentation in the Ravenscar Group
Previous studies have interpreted the degree of channel amalgamation (i.e. reoccupation of channels of the same location through time) and lateral facies variation in the Ravenscar Group to be controlled by syn-depositional movement on the Whitby and Peak faults (Fig. 20a;Alexander 1986;Alexander and Gawthorpe 1993). In these interpretations, the multi-storey channel bodies were thought to have been formed as a result of preferential stacking of channels in the hangingwall, associated with a marked change in formation thickness across the faults.
In the Whitby East Cliff, the Saltwick Formation is 28 m thick. In the Whitby West Cliff section, neither the Dogger nor the Eller Beck formations are exposed, but there is a 20 m thick multi-storey amalgamated Saltwick Formation channel body present. This rapid facies change was used by Rawson and Wright (1992) to invoke a NNW-SSE-trending fault along the River Esk, and Alexander and Gawthorpe (1993) proposed that the fault was responsible for the changes in facies stacking patterns. There has been discussion concerning the throw of the fault, if present. Hemingway et al. Table 4. Marine v. fluvial sedimentation in the Ravenscar Group and changing fluvial sandbody proportion and dimension between parasequences (1968) and Rawson and Wright (1992) suggested that the Dogger Formation is not more than 12 m below the present foreshore on the west side and in a more recent study Kulpecz (2008) described no significant formation thickening across the fault along the course of the River Esk. In this study, we did not observe any evidence of significant thickening of the Saltwick Formation across the Whitby Faults on the virtual outcrops. Furthermore, the occurrence of large, stacked and amalgamated channel bodies, which have previously been described as evidence for fault movement, also occur in several other sections such as Beacon Hill, where the virtual outcrop shows no evidence for fault movement. Therefore, we suggest that there is no evidence that the Whitby Fault controlled or influenced sedimentation.
Further south, previous studies reported that the thickness of the Saltwick Formation increased from 30 m to 57 m across the Peak Fault at Ravenscar (Hemingway 1974). The virtual outcrop supports this increase in thickness but suggests that it is less, 13 m from 30 m in the footwall (considering 30 m of formation thickness developed in the west of the fault) to 43 m in the hangingwall.
Detail of the stratal relationships around the Peak Fault are obscured by poor outcrop. Determining thickness variations of the Saltwick Formation across the fault was not possible due to the absence of Eller Beck Formation exposure in the footwall side of the fault. Consequently, two models that explain the thickening of the Saltwick Formation towards the SE are considered.

Syn-depositional fault movement
There is clear evidence that Peak Fault was active during the deposition of the Dogger Formation and there is demonstratable thinning on to the footwall block (Hemingway 1974;Milsom and Rawson 1989). Evidence for continued movement during the deposition of Saltwick is less clear. There is an apparent thickening, although this is less than previously described and there is a change in channel-stacking patterns with a series of large, multi-storey channel bodies being deposited in the hangingwall block of the fault (Fig. 20a). These channel bodies have previously been cited as evidence that fluvial systems were trapped against the fault scarp (Alexander 1986;Alexander and Gawthorpe 1993) and unable to avulse laterally. However, as with the Whitby Fault, comparable stacked channel bodies are seen in numerous places along the coast where they are demonstrably not associated with faults or fault scarps.

Simple downdip thickening
At Whitby East cliff, the Saltwick Formation is 28 m thick; at the Beacon Hill section, about 4 km SE of the Whitby Fault the formation is up to c. 33 m thick and further 8 km SE at Ravenscar the thickness has increased to 43 m. These values illustrate a general, downdip thickening of 1.25 m km -1 (Fig. 20b). The complete sections 22 km long on either side of the Peak Fault suggest that the observed 13 m of thickening could simply be part of this trend.
Faults as a secondary control on channel belt location As stated above, it is clear that the Peak Fault was active during deposition of the Dogger Formation but less certain during Saltwick times. It is possible that locally, either through differential compaction of the underlying, thicker, Dogger Formation, or minor faulting, that the Peak Fault influenced the location of the major stacked channel bodies in its hangingwall. However, deposition is also influenced by the regional floodplain tilting and base level changes.
A similar scenario has also been seen in the age-equivalent Ness Formation in the Norwegian North Sea (Johnsen et al. 1995;Ryseth 2000). Both the preferential occurrence of channel bodies, in hangingwall blocks, parallel to minor faults, and the main channel bodies that either cut directly over faults and ignore them or are developed away from faults, were studied there.
Both cases (Saltwick Formation in Yorkshire and Ness Formation in the North Sea) suggest that faults had limited surface expression during deposition. The depositional systems will infill any subtle topographic lows associated with the faults and may locally impact channel location, but overall the coastal plain drainage systems were capable of ignoring the faults and were able to be more controlled by, e.g., the base level changes or overall tilt of the coastal plain from the land towards the sea (Ryseth 2000;Mack et al. 2003).
The fluvio-deltaic Sycarham and Gristhorpe members of the Cloughton Formation and the Scalby Formation sections do not have an extensive enough dataset to explore the influence of the syndepositional fault movement and preferential channel development.  (Alexander and Gawthorpe 1993). Between these two faults is the horst, but several thick multi-storey channel bodies (up to 20 m thick) are observed in the horst, which is unusual. (b) Scenario-2: simple downdip thickening. The channel bodies shifted through time due to normal accommodation-controlled channel avulsion. The formation thickness increased from 28 m at Whitby East Cliff through 33 m at Beacon Hill Cliff to 43 m at Ravenscar East Cliff due to downdip stratigraphic thickening of the Saltwick Formation.
The channel body morphology-changing pattern from broad, thick, multi-storeyed channel body to isolated, small channels stacked in floodplain fines or vice versa in vertical section and decreasing channel body dimension towards depositional dip indicate that base level changes or overall tilt of the coastal plain from the land towards the sea may play an important role, for example, in the Saltwick Formation.

Architecture of overbank sandbodies and their role in paralic reservoirs
Crevasse splays are the primary coarse-grained architectural element in overbank deposits. Three different splay geometries: single splays, amalgamated splays and splay complexes are categorized based on their stacking pattern. Single splays are small-scale, isolated, lenticular or sheet-like sandbodies. The length/width ratio of single-splay bodies suggests a lobate geometry. Crevasse splays studied in modern systems (Rahman 2019, ch.2), also show lobate single-splay geometries in similar climatic settings (humid to subtropical) to these in fluvio-deltaic deposits. Amalgamation of splays form moderate to large sandbodies which show three different splay stacking patterns. Vertically amalgamated splays are up to 3.8 m thick sandstone sheets where the individual splays are separated by erosional contact or rooted horizons which suggest an amalgamation of two separate splays. In laterally amalgamated splays, individual splays are connected to the contiguous splays and form lensoid, boudinage geometry in cross-section. Both the lateral and vertical amalgamation processes form comparatively thicker (up to 5.8 m), stacked sandbodies where the individual splays are connected laterally and vertically with pockets of floodplain mudstones. Amalgamated splays developed as a result of compensational stacking in the high accommodation parent channel systems (van Toorenenburg et al. 2016;Gulliford et al. 2017).
The observed crevasse splay complexes are the thickest (up to 6.5 m) and most extensive (>1200 m wide at the Cloughton Wyke section;  reported c. 20 km) overbank sandbodies. The coarsening upward trend and tidal indicator sedimentary structures (flaser, lenticular and wavy beddings) suggest the splay progradation in the interdistributary bay is similar to that of the crevasse splay complexes in the modern Mississippi Delta (Rahman 2019, ch.2).
Splays constitute a significant proportion of all of the studied delta plain successions (up to 28% in the Saltwick Formation at Whitby East Cliff and 53% in the Gristhorpe Member at Cloughton Wyke). The volume of crevasse splay deposits was estimated from the studied dimensional data. The single-crevasse splays have a larger aerial extent but comperatively low volumes (10 2 -10 5 m 3 ) because they are thin. The amalgamated crevasse splays which formed as a result of lateral and vertical stacking of single splays attain several metres thick sandbody with parent channel parallel large aerial extent, and their rock volumes range on the order of 10 2 -  extensive bodies with a thickness up to 6.5 m, and with their rock volume on the order of >10 7 m 3 .  suggested their rock volume to be up to 10 9 m 3 in the Gristhorpe Member. Coeval splay bodies are connected to the parent channel and occasionally truncated by the younger channel, providing channel body connectivity. In the delta plain/paralic settings splays may enhance the reservoir volume and occasionally reservoir connectivity.

Comparison of sandbody dimension with analogues
The dimensional data of studied sandbodies were analysed within a broader context by comparing with channel body dimensions of similar depositional environments from published literature. Channel body dimensions were compared with data from delta plain and meandering river systems, compiled by Gibling (2006), lower coastal plain deposits of the Lower Williams Fork Formation, studied by Pranter et al. (2009) and fluvial and coastal plain successions of the Blackhawk Formation, studied by Rittersbacher et al. (2014) (Fig. 21).
The channel body dimensions of the fluvio-deltaic succession of the Ravenscar Group show similarity with previously studied systems in similar depositional environments (Fig. 21). The compiled delta plain channel body W/T of Gibling (2006) are entirely placed within the W/T envelope of the Ravenscar Group. Gibling (2006) reported that delta plain distributaries are 3-20 m thick with W/T <50 which are characterized by predominant vertical accretion. However, the delta plain channel body dimensions in the Ravenscar Group are widely distributed, ranging from 15-2034 m wide and 3-28 m thick with W/T 5->102. They are characterized by predominant lateral accretion. The dominance of multi-storey channel body types along with point bar deposits imply that most of the channel bodies are laterally migrated and vertically or obliquely stacked.
As the studied channel bodies in the Ravenscar Group are dominated by lateral accretion deposits they were compared to the meandering channel body dimensions of Gibling (2006). The W/T values of some of the largest channel bodies in the Ravenscar Group (for example, channel bodies in the lower part of the Saltwick Formation and exhumed meander belt of the Scalby Formation) are comparable to the lower-end members of meandering channel bodies. The valley-fill deposits of the Moor Grit Member (>1000 m wide, 10 m thick with W/T >100) are similar to the 'Valley Fill within Alluvial and Marine Strata' type of Gibling (2006).
The channel bodies of the coastal plain Lower Williams Fork Formation, Coal Canyon, Colorado (Pranter et al. 2009 Data from the Ravenscar Group show a similar trend to the Blackhawk Formation in the Wasatch Plateau, USA (Rittersbacher et al. 2014). In the Blackhawk Formation, Rittersbacher et al. (2014) showed that channel bodies are 7-1020 m wide and 2-23 m thick with W/T up to 90, which cover most of the envelope of the Ravenscar Group. The channel bodies in the Blackhawk Formation show an increasing upward trend similar to that in the Sycarham Member of the Coughton Formation whereas the Saltwick and Scalby formations show a decreasing upward trend.
Crevasse splay body dimensions of different fluvio-deltaic successions in the Ravenscar Group show a similar trend to those in previous studies (Fig. 22). Previously studied dimensional data of crevasse splays suggest they may extend up to 25 km with thickness up to 6 m and W/T up to 4170 (Kerr 1990;Smith 1993;Jorgensen and Fielding 1996;Anderson 2005;Arco et al. 2006;Pranter et al. 2014;Burns et al. 2017;Gulliford et al. 2017). In the present study, observed crevasse splay bodies are up to >1285 m wide and 6.5 m thick with W/T ratio >541. The similarity of the sandbody dimensions of the Ravenscar Group with those of the equivalent depositional environments demonstrates that the studied fluvio-deltaic sections can act as an appropriate analogue for the North Sea reservoirs or any reservoirs deposited in similar environments.

Conclusions
The application of digital data acquisition techniques along with traditional field techniques and behind outcrop well data is a powerful approach to study large and inaccessible cliff exposures even in areas that have been previously well studied. Virtual outcrops provide a means to observe the large-scale geobody trends that are not otherwise obvious, and provide access to otherwise inaccessible sections such as the Jurassic successions of the Yorkshire coast. The studied succession is subdivided into two, unconformity-bounded sequences, and the lower sequence is comprised of five upward-shallowing successions, bounded by marine flooding surfaces. Both sequences show an overall retrogradational stacking pattern. Overall, the entire stratigraphic section shows a thickening down depositional-dip.
The fluvial sandbodies within the succession have been classified into five architectural types: single-storey small ribbon, singlestorey broad ribbon, tabular multilateral sheet, internally amalgamated multi-storey sheet and partially amalgamated multi-storey sheet. These show a systematic arrangement within the parasequences. Overbank sandbodies have also been mapped. These include ribbon crevasse channels, single-splay sheets, amalgamated splay sheets, crevasse splay sheet complexes and wedge-shaped levee deposits. Dimensional data collected from the sandbodies have utility as an analogue for similar fluvial reservoirs such as the Ness Formation in the Brent Group (North Sea) and the Middle Jurassic Åre Formation in the Norwegian Sea.
Finally, the more complete dataset available from the virtual outcrop studies suggests that previous models for structural controls on channel body distribution may not be correct, especially in the case of the Whitby Fault. While major sandbodies do occur adjacent to the fault, they are not unique to this position, and similar bodies occur in other sections where no faults are observed. There is clear evidence for pre-Saltwick movement on the Peak Fault and potential that faulting continued through Saltwick times; however, faulting is considered to be a local, rather than regional, control on sedimentation patterns.