Quantifying the three‐dimensional stratigraphic expression of cyclic steps by integrating seafloor and deep‐water outcrop observations

Deep‐water deposits are important archives of Earth’s history including the occurrence of powerful flow events and the transfer of large volumes of terrestrial detritus into the world’s oceans. However the interpretation of depositional processes and palaeoflow conditions from the deep‐water sedimentary record has been limited due to a lack of direct observations from modern depositional systems. Recent seafloor studies have resulted in novel findings, including the presence of upslope‐migrating bedforms such as cyclic steps formed by supercritical turbidity currents that produce distinct depositional signatures. This study builds on process to product relationships for cyclic steps using modern and ancient datasets by providing sedimentological and quantitative, three‐dimensional architectural analyses of their deposits, which are required for recognition and palaeoflow interpretations of sedimentary structures in the rock record. Repeat‐bathymetric surveys from two modern environments (Squamish prodelta, Canada, and Monterey Canyon, USA) were used to examine the stratigraphic evolution connected with relatively small‐scale (average 40 to 55 m wavelengths and 1.5 to 3.0 m wave heights) upslope‐migrating bedforms interpreted to be cyclic steps within submarine channels and lobes. These results are integrated to interpret a succession of Late Cretaceous Nanaimo Group deep‐water slope deposits exposed on Gabriola Island, Canada. Similar deposit dimensions, facies and architecture are observed in all datasets, which span different turbidite‐dominated settings (prodelta, upper submarine canyon and deep‐water slope) and timescales (days, years or thousands of years). Bedform deposits are typically tens of metres long/wide, <1 m thick and make up successions of low‐angle, backstepping trough‐shaped lenses composed of massive sands/sandstones. These results support process‐based relationships for these deposits, associated with similar cyclic step bedforms formed by turbidity currents with dense basal layers under low‐aggradation conditions. Modern to ancient comparisons reveal the stratigraphic expression of globally prevalent, small‐scale, sandy upslope‐migrating bedforms on the seafloor, which can be applied to enhance palaeoenvironmental interpretations and understand long‐term preservation from ancient deep‐water deposits.


INTRODUCTION
Submarine canyon-channel-fan systems are important for transferring sediment and other materials (for example, organic carbon, microplastics and other pollutants) from terrestrial to deep marine environments (Normark et al., 1993;Hessler & Fildani, 2019;Kane et al., 2020;Pohl et al., 2020). Their sedimentary deposits store a wealth of information about the Earth's history (for example, climate and tectonic events, Clift, 2006;Castelltort et al., 2017;Daniels et al., 2018;Fildani et al., 2018;Mason et al., 2019), provide a long-term record of turbidity current events required for seafloor hazard assessments (e.g. Goldfinger, 2011;Clare et al., 2015Clare et al., , 2016Jobe et al., 2018) and host significant natural resources (for example, hydrocarbons, Pettingill & Weimer, 2002). Observations from modern environments are typically a primary source of information about sedimentary processes, which commonly inform interpretations in the rock record. However, modern deep-water environments are notoriously difficult to study due to their location at substantial water depths and distances from the shore, as well as the unpredictable and destructive nature of turbidity currents (Mutti & Normark, 1987;Paull et al., 2003;Sumner & Paull, 2014;Clare et al., 2020). As a result, the understanding of fundamental deep-water sedimentary processes has been limited compared to other depositional environments that are more easily accessed and readily observed on the Earth's surface.
Small-scale sediment waves (<300 m wavelength, <8 m wave heights) interpreted to be cyclic steps are evident in many modern sandy submarine channels and lobes (Hughes Clarke, 2016;Symons et al., 2016;Slootman & Cartigny, 2020). Recent investigations have used repeat bathymetric surveys to link turbidity current morphodynamics, such as bedform migration, to distinct depositional products in modern environments (for example, Squamish prodelta;   Symons et al. (2016). (B) Sketches highlighting additional characteristics associated with each bedform group including planform geometry, composition and common location within turbidite-dominated environments. (C) to (E) Linkages between seafloor features, sedimentary processes and stratigraphic products established from observations in modern environments. (C) Morphology of features interpreted to be cyclic steps on the seafloor, often referred to as sediment waves and scours. (D) Flow-seafloor interactions and migration for cyclic step bedforms in (C). (E) Stratigraphic products of these processes based on repeat bathymetry data (Hage et al., 2018) and seismic profiles (Migeon et al., 2001). Parts (A) to (C) adapted from Symons et al., 2016. Hage et al., 2018Vendettuoli et al., 2019). Despite recent work, detailed study of deposit geometry and its relation to formative bedform morphology and migration history has not been undertaken. A more comprehensive understanding of three-dimensional architecture is required in order to effectively identify these sedimentary structures in outcrops, which are often oriented oblique to the palaeoflow direction and variably-exposed. In addition, establishing distributions of bedform deposit morphometric properties and how they reflect formative bedforms is needed to support more refined interpretations of palaeoflow information from the rock record.
This study builds on and extends observations from modern environments (Fig. 1) to the interpretation of ancient deep-water deposits. The three-dimensional architecture and nature of sedimentary deposits are examined from upslope-migrating bedforms interpreted as cyclic steps in channelized slopes of the Squamish prodelta (British Columbia, Canada) and Upper Monterey Canyon (California, USA), and findings are compared to a succession of Late Cretaceous Nanaimo Group deep-water slope deposits that crop out on Gabriola Island (British Columbia, Canada). In doing so, ample statistics are produced that quantify bedform deposit scale and geometry as well as describe their 3D shape, distribution and sedimentary features. Collectively these observations directly link deposit characteristics to bedforms formed by naturallyoccurring turbidity currents, which is hypothesized to facilitate more robust interpretation of coarse-grained deep-water strata in outcrop. Although differences in spatial and temporal scales make the comparison of modern and ancient datasets difficult, it also means that each holds unique insight into the nature, occurrence and preservation of sedimentary processes on the seafloor. This analysis demonstrates the usefulness of integrating seafloor and stratigraphic datasets, which has proven particularly challenging for deep-water settings in the past (Mutti & Normark, 1987.

Modern: Squamish Prodelta, British Columbia, Canada
The Squamish delta is located within a marine fjord in Howe Sound, British Columbia, and receives more than one million cubic metres of sediment per year from the glacially-influenced Squamish River ( Fig. 2A and B; Hickin, 1989). The delta is most active in the summer months due to an influx of seasonal meltwater, during which time >100 flow events may occur with ca 20% reaching the base of the prodelta slope (Hughes Clarke et al., 2012b;Clare et al., 2016;Hizzett et al., 2017;Stacey et al., 2019). Events consist dominantly of turbidity currents generated by river plume suspended sediment settling, and occasional subaqueous landslides produced by delta lip failures (Hughes Clark et al., 2014;Clare et al., 2016;Hage et al., 2019). The frequent occurrence of turbidity currents and position of the system at relatively shallow water depths (spanning 0 to 170 m) makes Squamish an ideal setting for studying turbidity current dynamics and triggers (Hughes Clark et al., 2012a, 2012bClare et al., 2016;Hughes Clarke, 2016;Hizzett et al., 2017;Hage et al., 2019;Stacey et al., 2019).
The Squamish prodelta slope contains three submarine channels (northern, central and southern) that transition into lobes 1 to 2 km from the delta lip ( Fig. 2A; Hughes Clark et al., 2014). Active crescentic or sinuous upslopemigrating bedforms are observed within the prodelta channels and lobes, and are interpreted to be supercritical-flow bedforms based on observations of flow-seafloor interactions (Hughes Clarke, 2016), morphodynamic modelling (Covault et al., 2017) and geomorphological characteristics (Hughes Clarke et al., 2012a;Hage et al., 2018;Vendettuoli et al., 2019;Slootman & Cartigny, 2020). Direct monitoring of flows in the northern channel discovered thickening in the high-density basal layer of turbidity currents (i.e. hydraulic jump) at lee-stoss breaks during upslope bedform migration, which is consistent with flow over cyclic steps (Hughes Clarke, 2016). Analysis of time-lapse surveys suggests that sediment is primarily reworked during bedform migration, and thus produces very incomplete successions with a delta-wide average of only 10% stratigraphic preservation after one year (Vendettuoli et al., 2019). Sediment cores and acoustic backscatter images indicate that bedforms in the prodelta channels and lobes dominantly consist of coarse-grained sand (Hughes Clarke et al., 2012a;Hage et al., 2018;Stacey et al., 2019).
In 2011, multibeam bathymetric surveys of the entire Squamish prodelta (93 surveys in total) were collected every weekday over four months in the summer (Fig. 2E;Hughes Clark et al., 2012a, 2012b. The spatial (0.1 m vertical, 2 9 2 m grid pixels) and temporal (every one to three days) resolution of the surveys allows seafloor changes resulting from almost daily turbidity currents to be captured in an extremely high level of detail (c.f. Hizzett et al., 2017;Hage et al., 2018). A subset of this dataset (12 surveys in total) covering the lower southern channel, channel-lobe transition zone and lobe was used in this study, including surveys that preceded and followed the seven flow events in the southern channel that extended to the lobe at the base of the prodelta slope ( Fig. 2E; Clare et al., 2016)  . The upper map shows seafloor changes associated with a turbidity current event. The lower map was created over a time period in which no turbidity current events occurred, where minor seafloor changes reflect errors associated with data acquisition and positioning. Mean and standard deviation seafloor change (*statistics for absolute magnitude of change) was calculated for each map over the area delineated by the yellow polygons, which encompasses the southern channel and lobe. (D) Seafloor profiles along bedforms in the southern channel thalweg from bathymetric surveys used to produce the difference maps shown in (B) and (C) (profile location indicated by blue-dashed line). (E) Timing of repeat bathymetry survey acquisition and turbidity current events/seafloor changes recorded in the southern channel and lobe over summer, 2011. Red arrows mark the timing of bathymetry surveys used in this study, which precede and follow turbidity current events that reached the base of the prodelta (adapted from Clare et al., 2016). See Appendix S2 for all seafloor difference maps and profiles between turbidity current events in the investigated area. created between surveys where flow events occurred document a distinct alternating pattern of erosion (up to 5.5 m) and deposition (up to 6 m) within the channel and lobe associated with bedform migration (Fig. 2C and D;c.f. Hughes Clark et al., 2012aHage et al., 2018;Vendettuoli et al., 2019). On the other hand, overall minimal seafloor change (average magnitude of 0.14 m) over the channel and lobe is recorded in difference maps created between surveys not affected by flow events (Fig. 2C and D; see Appendix S1 for table comparing the magnitude of seafloor changes between surveys). Error ranges established for a larger portion of the 2011 Squamish prodelta bathymetry dataset also revealed small seafloor changes, 4 cm AE 23 cm, in the absence of turbidity current events (Hizzett et al., 2017;Vendettuoli et al., 2019). Local outlier differences (up to 5.2 m) between these surveys appear to relate to minor positioning errors, which result in the slight offset of channel margins or bedform crests, or refraction residuals at the edge of swaths along the perimeter of the datasets ( Fig. 2C; Hughes Clarke et al., 2012a). In addition to bathymetric surveys, sediment cores obtained from the central channel in 2016 (Hage et al., 2018) were used to capture the sedimentological characteristics of bedform deposits on the channelized slope ( Fig. 2A).

Modern: Upper Monterey Canyon, California, USA
The Monterey Canyon is a large submarine canyon-channel system incised into the shelf in Monterey Bay, which extends for >400 km onto the Monterey fan off the west coast of California ( Fig. 3A; Normark, 1970;Greene, 1977). An estimated 200 000 to 500 000 m 3 of sediment per year is delivered to the canyon head by littoral transport sourced from nearby rivers and coastal erosion (Best & Griggs, 1991;Eittreim et al., 2002). Several processes are responsible for sediment transport in the Monterey Canyon including: sediment gravity flows and other forms of mass movement, which are capable of delivering and transporting mixed sediment loads of up to gravel-sized particles (Xu et al., 2004;Paull et al., 2018); and internal tides, which can resuspend and re-distribute fine-grained material (Xu & Noble, 2009;Maier et al., 2019b). Turbidity currents with varying runout distances have been documented to occur sub-annually (Xu et al., 2004;Paull et al., 2010Paull et al., , 2018Symons et al., 2017;Heerema et al., 2020) but rarely reach the Monterey fan (Stevens et al., 2014). Like the Squamish delta, the Monterey Canyon has been an important site for studying deepwater sediment transport processes and for developing sophisticated seafloor mapping and active flow monitoring technologies (e.g. Paull et al., 2003Paull et al., , 2010Paull et al., , 2018Xu et al., 2004Xu et al., , 2008Smith et al., 2005Smith et al., , 2007Symons et al., 2017;Maier et al., 2019a). The upper Monterey Canyon, located between 20 m and 200 m water depths, consists of four tributary heads that converge to a sinuous axial channel (Fig. 3B). The active tributaries and main channel contain upslope-migrating bedforms, which have been interpreted as cyclic steps based on their geometry and migration history (Cartigny et al., 2011;Kostic, 2011;Symons et al., 2016). Cores and samples from bedforms on the axial channel floor consist of gravel and sand (Paull et al., 2005(Paull et al., ,2010Maier et al., 2019).
Between September 2002 and March 2014 multibeam bathymetric surveys of the Upper Monterey Canyon were conducted almost annuallybiannually by the Seafloor Mapping Lab at California State University, Monterey Bay, with similar spatial resolution to Squamish datasets (ca 0.2 m vertical resolution, 3 9 3 m grid pixels). These data can be freely downloaded from the California Seafloor Mapping Project (http:// seafloor.otterlabs.org). Select bathymetric surveys within the Monterey dataset have been previously used to investigate geomorphic changes and sediment transport to March 2003surveys in Smith et al., 2005to February 2005surveys in Smith et al., 2007), and bedform migration (2002to February 2006surveys in Xu et al., 2008to March 2005surveys in Cartigny et al., 2011 within the Upper Monterey Canyon. Fourteen of the surveys with sufficient coverage and resolution, spaced 76 to 918 days apart were used in this study (Fig. 3D). Flow events in the Monterey Canyon are less frequent and not as well-constrained between acquisition periods as on the Squamish prodelta. Seafloor changes between each survey indicate the occurrence of at least one flow event; however, they likely reflect the combined effect of multiple events as monitoring programs in the upper Monterey canyon typically record between three and eight flows per year (Xu et al., 2004;Paull et al., 2010Paull et al., , 2018. On difference maps, considerable seafloor changes are observed at the canyon head and within the axial channel related to bedform migration (average magnitude of 1 m, up to 11 m), while minimal changes are observed on the canyon flanks (average magnitude of 0.14 m, up to 1.5 m) where presumably no or minor turbidity currents occur ( Fig. 3C and E; see Appendix S1 for table comparing the magnitude and distribution of seafloor changes between surveys).

Ancient: Twin Beaches Peninsula, Gabriola Island, British Columbia, Canada
The Nanaimo Group is an up to 4 km thick succession of Late Cretaceous forearc basin fill deposits, presently exposed in the Cowichen fold and thrust belt in south-western British Columbia ( Fig. 2A; Mustard, 1994  units consist of deep-water strata and make up the bedrock of the Gulf Islands in the Strait of Georgia (Figs 2B and 4B;England & Hiscott, 1992;Katnick & Mustard, 2003). Coarse-grained sandstone and conglomerate-prone deposits form the resistant cores of individual islands, which alternate with and laterally transition to fine-grained lithologies (Bain & Hubbard, 2016;Englert et al., 2018). The area of interest for this study is a succession of the Gabriola Formation exposed in the intertidal zone around the Twin Beaches Peninsula at the north-west end of Gabriola Island (Figs 2B and 4B). On Gabriola Island, upper Nanaimo Group units are exposed in a north-west plunging syncline (Fig. 4B). The Gabriola Formation consists of thick-bedded, coarse-grained sandstone with packages of interbedded very fine-grained sandstone and siltstone, interpreted to be largely deposited within or in association with submarine channels (England & Hiscott, 1992). Foraminifera from finegrained intervals within the formation on Gabriola Island indicate palaeo-water depths between 600 m and 1200 m (England & Hiscott, 1992). Palaeoflow measurements around the island range from south-west to north-east but are dominantly west/north-west, generally parallel to the orientation of the island (Fig. 4B).

METHODS
To characterize the stratigraphic products of upslope-migrating bedforms, this study integrates and compares observations of deposit geometry, architecture and facies from different marine settings (delta slope, proximal canyon and deep-water slope) over different temporal periods (daily, yearly, geological timescales; Fig. 5A).

Seafloor observations
Water depth values were extracted from each survey in the Squamish prodelta and Upper Monterey Canyon datasets along dip, oblique (45°) and strike-oriented profiles (Fig. 8A). Subsequent surveys were stacked to determine areas of erosion and deposition, and ultimately reconstruct the depositional products associated with upslope bedform migration (Fig. 6, Appendix S4). Bedform morphology (wavelength, wave height, lee and stoss slope angles), migration history (magnitude and direction) and stratigraphic evolution (amount of deposition and erosion/reworking) were determined for each time step in dip-oriented profiles in order to capture the nature of seafloor features and sediment transport (Fig. 6). The geometry (length and maximum thickness) and architecture of resulting deposits was then qualitatively and quantitatively evaluated in a variety of orientations (for example, dip, oblique and strike) relative to the flow direction (Figs 6 and 8A). Localized erosion and deposition observed in areas or time periods unaffected by turbidity currents likely reflects the presence of some artificial seafloor changes in stratigraphic reconstructions due to systematic surveying and/or positioning errors (Figs 2C and 3C;Hughes Clarke et al., 2012a). However, these changes are small compared to those associated with actual flow events (Appendix S1). Upslope-migrating bedforms have been previously described in the Squamish (Hughes Clark et al., 2012a, 2012bClare et al., 2016;Hizzett et al., 2017;Vendettuoli et al., 2019) and Monterey datasets (Smith et al., 2005(Smith et al., , 2007Cartigny et al., 2011) used here, and stratigraphic reconstructions have been made along bedforms in the Squamish prodelta southern channel (Vendettuoli et al., 2019). This study supports the observations from these initial investigations and provides new results including more detailed analysis of bedform morphology, migration and associated deposit geometries in three-dimensions, and the generation of large quantitative datasets along variably oriented cross-sectional profiles.
The sedimentological characteristics of bedform deposits were captured in seven piston cores from the Squamish prodelta, previously photographed and described by Hage et al.  (2018). For this study the cores were X-rayed and re-examined to ensure consistent analysis with ancient deposits, which included visual inspection and logging to document grain-size distributions, bed thicknesses and sedimentary features.

Outcrop characterization
Field work was conducted at Twin Beaches Peninsula on Gabriola Island to capture the nature of the outcrop and characterize deposits within the succession. Lithology was mapped around the peninsula using a differential GPS (global positioning system) and measurements of bedding orientation and palaeoflow indicators were collected. Drone images and structurefrom-motion photogrammetry software were used to create high resolution 3D outcrop models for delineating and measuring the geometries of sedimentary units (c.f. Nesbit et al., 2018). The architecture of deposits was further established through scaled field sketches and photomosaics. Detailed measured section, field observations and photographs were used to document the sedimentological characteristics of deposits.

RESULTS: MODERN
Bedform morphology, migration history and stratigraphic evolution Observations: Squamish prodelta Bedforms are crescentic on the Squamish prodelta and develop a sinuous planform as flows lose confinement on the terminal lobe ( Fig. 2A  The methodology for generating stratigraphic reconstructions and analyzing repeat-bathymetric datasets using an example from a segment along the Squamish prodelta southern channel (part of profiles indicated by the blue-dashed line in Fig. 2C). Bedform morphological attributes (wavelength, wave height and slope angles) were obtained from individual surveys. Subsequent surveys were stacked in order to determine areas of erosion and deposition resulting from bedform migration, and the ultimate preservation of deposits in the stratigraphic record. The length and thickness of initial deposits associated with each time step were measured, as well as the dimensions of reworked deposits in the final reconstructed stratigraphy. *Slope values are listed as magnitudes. Lee slope dip directions are consistent with the background slope while stoss slopes generally dip in the opposite direction, but in some cases have low-angle inclination in the same direction as the background slope. (C) Total erosion, total deposition and net sediment flux between each survey time step along dip-oriented profiles from the two datasets. Average crest migration distance and arrows representing crest migration direction are also shown where it could be determined between Squamish prodelta surveys. (D) Deposit preservation expressed by the reworked deposit area in the final stratigraphy as a percent of the initial deposit area for each time step. These graphs indicate a significant amount of deposit reworking, which results in low preservation of older stratigraphy.
time, flow events that interact with bedforms are interpreted as turbidity currents . Variable proportions of lee-side erosion and stoss-side deposition on bedforms produced upslope migration between daily surveys over distances of metres to tens of metres with slight angles of climb or fall (average 0.2°to 29°of climb across profiles, between 14°of fall and 90°of climb for individual bedforms; Fig. 7C). Along a dip-oriented profile of the channel thalweg, a combination of erosion and deposition occurs between all survey time steps; however, the system fluctuates between net erosional or net depositional conditions. Erosion on the leeside of bedforms led to reworking of previous stratigraphy where up to 75% of deposits (average 39%) associated with a particular time step are removed over the entire surveyed time interval ( Fig. 7D).

Observations: Upper Monterey Canyon
In the Upper Monterey Canyon, a ca 400 m wide, steep (4.5°to 5.5°) area covered by sinuous bedforms leads into a narrower (ca 140 m), lower gradient (3.0°to 3.5°) channel containing crescentic bedforms (Fig. 3B). Bedforms have 13 to 139 m (average 42 m) wavelengths, 0.8 to 9.8 m (average 3 m) wave heights, 3.4°to 18.1°( average 10.6°) lee slopes, 0 to 5.6°(average 1.3°) stoss slopes, and are predominantly downslope asymmetric (Figs 3E, 7A and 7B). The temporal resolution of surveys is often not sufficient to resolve bedform migration in the Monterey Canyon. Bedforms are completely reworked between surveys making it difficult to correlate individual crests, although in some instances upslope movement is evident ( Fig. 3C and E; see Appendix S3 for comparison of bedform positions and profiles between surveys). Upslope migration, facilitated by erosion on the lee-side and deposition on the stoss-side of bedforms, is better documented over a shorter, 32 day time period in a study of the Upper Monterey Canyon by Smith et al. (2007). Seafloor changes are likely a result of sediment gravity flows or mass movement events, which have been previously attributed to coarsegrained sediment transport in the proximal canyon and directly observed further downslope (Smith et al., 2007;Paull et al., 2010;Paull et al., 2018). It is unclear whether all of the observed deposition occurs in association with bedforms; however, subsequent reworking by bedforms appears to dominate local patterns of erosion/deposition and the stratigraphic evolution between surveys regardless of the initial mode of deposition (Fig. 3C). Cumulative sediment flux along the channel thalweg varies from net erosional to net depositional conditions between survey time steps (Fig. 7C). Significant erosion causes complete removal or poor preservation (average 26%) of stratigraphy associated with a specific time period (Fig. 7D).

Integration and interpretations
Bedforms of similar scale and morphology (planform and cross-sectional) are present in the Squamish prodelta and Upper Monterey Canyon, and are additionally consistent with numerical modelling and other seafloor observations of sandy cyclic steps (e.g. Cartigny et al., 2011;Kostic, 2011;Normandeau et al., 2016;Symons et al., 2016;Covault et al., 2017;Slootman & Cartigny, 2020). The bedforms observed in the Upper Monterey Canyon are more symmetrical and have smaller aspect ratios (length/ height) compared to those measured on the Squamish prodelta ( Fig. 7A and B). These morphological differences could reflect the steeper slope gradient and/or specific properties of flows

Deposit geometry and architecture
Observations: Squamish prodelta Bedform deposits reconstructed by stacking repeat bathymetric surveys from the Squamish prodelta consist of discontinuous sedimentary bodies that are 4 to 289 m long downslope (median: 22 m), 2 to 191 m wide across slope (median: 28 m) and up to 2.3 m thick (median: 0.6 m; Figs 8 and 9). Initial deposits are generally similar in downslope length to the stoss side of bedforms (representing ca 60% of the bedform's wavelength) with thicknesses approximately half that of bedform heights (Fig. 8D).
However, comparison of deposit geometry between surveys indicates that deposit length and thickness decrease over time, with median values decreasing by 13 to 26% and 0 to 8%, respectively ( Fig. 8B to D; see Appendix S4 for videos showing the stratigraphic evolution along profiles). Strike-oriented cross-sections contain slightly longer and thinner deposits than dip and oblique-oriented sections (Fig. 9). Individual sedimentary bodies in final stratigraphic reconstructions are trough-shaped with crescentic or oval planform geometries, and lenticular, sigmoidal/wedge-shaped or tabular in cross-section (Fig. 9). Erosional bounding surfaces at the base of deposits dip in all directions but are most commonly oriented upslope at angles <10°. In dip and oblique-oriented sections, bedform deposits are more asymmetrical lenticular and sigmoidal with steeper upslope basal surfaces (dipping up to 30°) that erosionally truncate underlying strata. Subsequent units are offset upslope forming low-angle backset stacking patterns. Variably nested symmetrical lenticular, wedge-shaped and occasional tabular bedform deposits are more commonly observed in strike-oriented sections.

Observations: Upper Monterey Canyon
Stratigraphic reconstructions from the Upper Monterey Canyon are comprised of nested bedform deposits with downslope lengths between 6 m and 260 m (median: 18 m), across slope widths between 3 m and 90 m (median: 15) and maximum thickness up to 3.1 m (median: 0.5 m; Figs 8 and 10; see Appendix S4 for videos showing the stratigraphic evolution along profiles). In general, initial deposits are similar in downslope length to the stoss side of bedforms (representing ca 60% of the bedform's wavelength) with thicknesses approximately a third of bedform heights (Fig. 8D). Deposit dimensions in the final stratigraphy are typically smaller than measurements of initially deposited units (reduction in median length by 13 to 28% and median thickness by 4 to 45%) and exhibit comparable sizes regardless of perspective.
Deposit architecture is quite irregular, but can be broadly described as lenticular in cross-section (Fig. 10). Basal bounding surfaces are often erosional and dip in all directions, however, they are steepest at the upslope head and edges of deposits. Reconstructions from dip and obliquely-oriented profiles contain slightly asymmetrical lenticular to sigmoidal deposits with locally apparent backstep and forestep stacking.
Strike-oriented profiles contain vertically stacked and laterally offset symmetrical lenticular deposits as well as wedge-shaped deposits at channel or scour margins.

Integration and interpretations
Sedimentary bodies of comparable size and architecture are observed in reconstructed stratigraphy from bedforms in the Squamish prodelta and Upper Monterey Canyon, despite encompassing different time periods (days versus years) and potentially different numbers of flows (Fig. 5A). Reconstructed units reflect deposition in the trough and on the stoss side of bedforms and can preserve part of the bedforms' stoss and, in some cases, lee slopes (Figs 6 and 8D). Minor differences between datasets are related to the morphology of formative bedforms and their stratigraphic evolution. For example, units initially deposited in the Upper Monterey Canyon are slightly shorter, thicker and have broader distributions than those deposited in the Squamish prodelta (Fig. 8). This could reflect the shorter wavelength and greater height of Upper Monterey Canyon bedforms, as well as the likely more reworked and composite nature of deposits due to more significant erosion and longer time durations between surveys (Fig. 7). However, reworking of initial strata in both settings results in smaller remnant units being preserved in the final reconstructions with comparable dimensions (Fig. 8). Deposit geometry and stacking in dip-oriented, oblique-oriented and strike-oriented sections, although similar to the Squamish prodelta, is more irregular in the Upper Monterey Canyon possibly due to greater deposit amalgamation and/or more erosive and fluctuating flow conditions in the Monterey system. Within datasets, differences in the architecture and, to a minor extent, dimensions of deposits in 2D cross-sections are related to the cross-section orientation through stratigraphy (Figs 8 to 10).

Deposit sedimentological characteristics
Observations: Squamish prodelta Sediment cores through bedform deposits on the Squamish prodelta are primarily composed of thick-bedded amalgamated sand (L1), with silty sand (L2.5) typically preserved at the bottom of cores or occasionally interbedded with thicker sand beds of L1 (Table 1; Fig. 11A and B). Coarse-grained sand beds are 10 to 100 cm thick, slightly normally graded, amalgamated, and largely structureless with some subtle internal laminations and grain-size breaks observed in cores and X-ray images. Equivalent facies are found in all sediment cores; however, individual beds cannot be correlated between cores despite their close spacing (cf. Hage et al., 2018). Similar characteristics have been reported

Observations: Upper Monterey Canyon
Sediment cores from the Monterey Canyon within and downslope of the study area have been previously described, although not examined in this study (Paull et al., 2005;Paull et al., 2010). Cores from the axial Monterey channel are quite variable, containing 40 to 200 cm thick massive, normally graded sand beds that are more poorly sorted at their bases and often contain floating gravel and mud clasts.  Bedding is planar or lenticular with sharp or loaded contacts L2.5: Silty Sand Traction and suspension deposition from highdensity turbidity currents (some bypassing) and/or low-density turbidity currents (Bouma, 1962;Mutti & Normark, 1987;Stevenson et al., 2015). L2.5 may also include deposition from debris or hybrid flows (Haughton et al., 2009;Talling et al., 2012) Lithology: Silty fine sand  (Fig. 8). These units resemble the deposits of high-density turbidity currents (c.f. Lowe, 1982) with high near-bed sediment concentrations, which suppress turbulence and lead to the formation of structureless deposits (Sumner et al., 2008;Talling et al., 2012;Hage et al., 2018;Paull et al., 2018). The origin of silty sands (L2.5) observed only in the Squamish prodelta cores is uncertain and interpreted to result from sediment gravity flow processes (Talling et al., 2012) and/or river plume suspension settling , possibly during periods of lower sediment delivery to the prodelta (such as in winter months) or when flows are bypassing or constricted to one side of the channel.

RESULTS: ANCIENT
Twin Beaches Peninsula stratigraphy

Observations
The Gabriola Formation at Twin Beaches Peninsula is exposed in NNW-NNE shallowly dipping (ca 7°to 12°) strata in quasi-horizontal beach and vertical wall outcrops around the peninsula that provide variably-oriented exposures (Fig. 12E). The peninsula is made up of a ca 65 m thick, at least 1.2 km wide succession of dominantly thick-bedded, medium-coarse grained sandstones (L1), in some areas interbedded with more heterolithic (L2) and mudstone-clast-rich deposits (L4; Figs 11C, 12 and 13; Table 1). At the base of the succession, these units sharply overlie and are lateral to siltstone-dominated deposits (L3) and chaotically bedded mudstone with sandstone beds (L5), giving the peninsula an overall channelform shape ( Fig. 12 (Tripsanas et al., 2008) Interpretations Large-scale channel-form architecture dominated by amalgamated coarse-grained turbidites (L1) is consistent with deposition in a deep-water slope conduit (cf. Gardner et al., 2003;Hodgson et al., 2011;Macauley & Hubbard, 2013), as well as previous interpretations of the Gabriola Formation and older coarse-grained formations on Gabriola Island (England & Hiscott, 1992;Englert et al., 2018). Underlying siltstone-dominated (L3) and discordant/disturbed (L5) strata are interpreted as low-energy slope deposits resulting largely from suspension settling (Talling et al., 2012) and slump deposits related to mass-wasting processes (Tripsanas et al., 2008), respectively. A refined regional context, which was not the focus here, would be needed to define a more precise depositional environment or slope position for the succession exposed at Twin Beaches Peninsula. However, the Gabriola Formation extends along the outcrop trend for ca 70 km to the southeast up palaeo-depositional dip, suggesting a minimum distance from the shoreline for the studied deposits at water depths of hundreds of metres (England & Hiscott, 1992).

Deposit geometry and architecture
Observations A considerable portion of the succession contains relatively small-scale discontinuous sedimentary bodies (Fig. 13). These sedimentary bodies are 2 to 84 m (median: 9.5 m) long and 0.1 to 4.8 m (median: 0.64 m) thick (Fig. 14A). In some areas highly amalgamated sandstones form larger packages that can be >100 m long, several metres to tens of metres thick, and contain variably mappable internal surfaces; however, definitive units cannot be resolved due to a lack of grain-size variability along portions of key sedimentary body boundaries. There is no clear trend in deposit dimensions around the peninsula despite differences in outcrop orientation (Fig. 14B).    Stratigraphy is generally characterized by a variety of nested lenticular and wedge-shaped units with some more tabular geometries (Fig. 13). Deposit basal bounding surfaces are often erosional, undulatory and may include superimposed, small-scale stepped scours (Fig. 13B); basal surfaces dip in all directions, typically between 3°and 32°(average ca 15°) although shallowly inclined surfaces were generally difficult to measure (Fig. 14B). In some oblique and dip-oriented outcrops, flute-shaped deposits with steep upslope faces are observed to truncate underlying strata, forming a series of upslope-stepping units (Figs 13A and 14C). Other similarly oriented outcrops contain upslope-inclined lens-shaped beds that thicken upslope, (Figs 13C and 15) while strike-oriented outcrops are more dominated by lenticular and tabular deposits (Figs 13D and 14D).

Interpretations
The prevalence, size and discontinuous nature of the units suggest that they are the deposits of relatively large, three-dimensional bedforms. Nested units with erosive basal surfaces indicate subsequent reworking and are interpreted to make up successions of bedform deposit remnants. These remnant deposits are characterized by backset stacking as well as asymmetrical scour fill geometry, which are consistent with the products attributed to supercritical flow bedforms (Alexander et al., 2001;Cartigny et al., 2014;Fedele et al., 2016;Slootman & Cartigny, 2020) and comparable to the cyclic step deposits reconstructed from the Squamish prodelta and upper Monterey Canyon seafloor datasets (Figs 9 and 10). Differences in architecture around the peninsula appear to relate to unit orientation relative to palaeoflow direction as well as the presence of potentially different styles of deposits (for example, degradational versus aggradational bedforms; Figs 1, 14 and 15). However, observations of architecture are also complicated by the structural dip, palaeoflow variability (commonly spanning 75°t o 110°) and the nature of exposure (flat beach or vertical wall).
Bedforms interpreted as cyclic steps are observed in a variety of modern slope conduit environments (Figs 1A, 2A and 3A) including channels with variable levels of confinement (Piper et al., 1985;Paull et al., 2010;Maier et al., 2011;Stacey et al., 2019), channel-lobe transition zones , sandy overbanks/splays (Fildani et al., 2006;Hill, 2012;Carvajal et al., 2017) and intraslope lobes . Thus, the coarse-grained succession of bedform-dominated deposits encapsulated in a composite channel-form body at Twin Beaches Peninsula can be attributed to a broad slope conduit characterized by a bedform field or laterally migrating channels containing trains of bedforms.

Deposit sedimentological characteristics
Observations Sedimentary bodies at Twin Beaches Peninsula are mainly composed of medium to thick beds of medium to coarse-grained, amalgamated sandstone (L1) and less commonly include mudstoneclast conglomerate (L4) and thin-medium bedded sandstone and siltstone (L2; Figs 11C, 13, 14 and 15; see Table 1 for lithofacies characteristics). The bounding surfaces of deposits are commonly erosional, and delineated by fine-grained drapes (L2) or amalgamated sandstone-on-sandstone contacts that are highlighted by beach erosion (Fig. 14C and D). Sandstone beds are normally graded, often only in the coarsest grain-size fraction, and have sharp, slightly undulatory and/or loaded contacts (Fig. 11C). Beds are largely massive with some subtle, localized sedimentary features including internal amalgamation surfaces, decimetre-scale foresets or backsets and planar stratification (Tb3a, Tb3b or Tb2, Figs 11C and 12). Mudstone clasts are common in lags at bed bases or in lenses within beds (Fig. 11C).

SYNTHESIS: MODERN AND ANCIENT RESULTS
Active channels in the Squamish prodelta and Monterey Canyon contain crescentic and sinuous, downslope asymmetric bedforms with wavelengths between 10 to 140 m and 0.3 to 9.8 m wave heights (Fig. 7). Upslope migration facilitated by variable degrees of lee-side erosion and stoss-side deposition suggest that bedforms can be classified as partially erosional to partially depositional cyclic steps (cf. Slootman & Cartigny, 2020) and produce remnant deposits (with 0.72 to 0.87 length and 0.55 to 1.0 thickness preservation ratios) that preserve ca 45% of the bedform wavelength, preferentially the stoss slope (Figs 8 and 16).
Similarities are also documented between deposits in modern (Squamish prodelta and Upper Monterey Canyon stratigraphic reconstructions) and ancient (Nanaimo Group stratigraphy) datasets (Figs 5A, 16 and 17). Successions contain discontinuous, erosion-surface bound, trough-shaped sedimentary bodies on the order of tens of metres long, often <1 m (up to several metres) thick that are composed of slightly normally graded, internally structureless sands/ sandstones (Figs 16 and 17; see Appendix S4 for videos of deposit stratigraphic evolution). Deposit architecture in 2D planform and vertical crosssections through stratigraphy is variable due to deposit reworking (Fig. 17B). Dip and obliqueoriented cross-sections contain more asymmetrical lenticular and sigmoidal deposits characterized by basal bounding surfaces with overall shorter, steeper up-section (<30°) and longer, gently upslope-inclined down-section (<10°) sides (Figs 16A, 17B and 17C). Deposits are nested and offset from one another, often in an upslope direction to form backstepping stacking patterns. Strike-oriented cross-sections contain more symmetrical lenses as well as wedgeshaped and tabular deposits; basal surfaces have a range of dips from flat to <30°that are more equally oriented towards either side of the section ( Fig. 17B and C). Sedimentary units may be stacked vertically, laterally offset or nested into each other. The equivalent scale, architecture and sedimentological characteristics of deposits within the Nanaimo Group succession at Twin Beaches Peninsula to those associated with bedforms in the Squamish prodelta and Upper Monterey Canyon support the hypothesis that these deposits formed by similar upslope-migrating bedforms, in this case cyclic steps (Figs 16 and 17).
Comparable findings from each study area support process-based relationships for flow-seafloor interactions and resulting stratigraphy, which pertain to multiple sedimentary environments ( Fig. 5A; Symons et al., 2016;Slootman & Cartigny, 2020). Seafloor datasets demonstrate the occurrence of morphologically similar cyclic step bedforms between and within different subaqueous settings spanning a range of water depths, system scales and sub-environments (lobes, channels, channel lobe transition zones, etc.; Figs 1A, 2A, 3A and 5A). Further, comparable bedform deposits can result despite differences in temporal scale, which may influence bed-scale characteristics due to variations in flow frequency and the increasing impact of preservation and the potential occurrence of rare flow events over longer time periods. The identification of cyclic step deposits in the ancient succession suggests that the stratigraphic products of dynamic bedforms such as those observed on modern delta fronts and canyons can be preserved over long timescales in the rock record, even under dominantly bypass conditions and where stratigraphic preservation potential is low.

Diagnostic recognition criteria
The relatively large size of seafloor bedforms and a lack of established recognition criteria for cyclic step deposits has been used to explain their previously limited identification in ancient deep-water successions (e.g. Piper & Kontopoulos, 1994;Hage et al., 2018). Based on present knowledge and the results of this study, diagnostic criteria (facies, scale and architecture) for recognition of cyclic steps in 1D, 2D and 3D exposures of coarse-grained deep-water deposits are reviewed (Figs 16 and 17).

Facies
Cyclic step deposits composed of amalgamated, massive, poorly graded sandstone (L1) and high energy interbedded sandstones and siltstones (L2) are consistent with facies traditionally ascribed to partially bypassing and waning highdensity turbidity currents ( Fig. 17A; Lowe, 1982;Sumner et al., 2008;Talling et al., 2012;Stevenson et al., 2015). some cases a greater proportion of planarlowangle, upslope-inclined, often diffuse stratified facies are observed either within or at the top of beds (Postma et al., , 2021Ito et al., 2014;Lang et al., 2017;Yang et al., 2017;Ono & Plink-Bj€ orklund, 2017;Postma & Kleverlaan, 2018;Cornard & Pickering, 2019). Recent work has questioned some of the process-facies linkages in classic turbidite models and proposed new associations, including top-cut-out units containing structureless Bouma (1962) Ta divisions with soft sediment deformation and stratified Tb divisions (Tb4 to Tb2), related to deposition on cyclic steps . However, predictive facies models may be complicated by variability along cyclic steps, incomplete preservation, the presence and reworking of unrelated mass-wasting deposits, and turbidity current characteristics such as flow unsteadiness (for example, eventual waning) and high sediment fallout rates. For example, Hage et al. (2018) suggested that the lack of stratification observed in bedform deposits on the Squamish delta may be a result of preferential preservation of hydraulic jump zone deposits and high near-bed sediment concentrations. Thus, facies are variable and often non-unique; they alone are presently not conclusive for recognizing cyclic step deposits in a 1D section such as a core or laterally restricted outcrop. Further investigation is required to determine whether distinct cyclic step facies exist and can reliably differentiate deposition from highdensity turbidity currents under different flow conditions.

Scale and stratigraphic architecture
The scale and architecture of deposits in 2D and 3D sections are more diagnostic for recognition and facilitated the identification of cyclic step deposits in the Nanaimo Group ( Fig. 17B and C). Successions dominated by beds with overall spoon/flute-shaped geometry, low-angle upslope inclination and backset stacking patterns are associated with cyclic steps in the analyzed datasets (Figs 16A and 17C). These observations are generally consistent with lenticular units and/or backset stratification described in flume experiments (Ono et al., 2021;Yokokawa et al., 2009;Cartigny et al., 2014;Fedele et al., 2016) as well as numerical and conceptual models of cyclic steps (Kostic, 2011;Vellinga et al., 2018;Slootman & Cartigny, 2020). Comparable architecture (asymmetrical scours and backset beds that thicken upslope) and surface dip angles have also been observed and attributed to cyclic steps in other deepwater outcrop investigations (Postma et al., , 2021Ito et al., 2014;Gong et al., 2017;Lang et al., 2017;Yang et al., 2017;Ono & Plink-Bj€ orklund, 2017;Postma & Kleverlaan, 2018;Cornard & Pickering, 2019). However, some deposit characteristics vary between studies, such as the scale of structures in terms of their length/height (ranging from decimetres to tens of metres long/centimetres to metres tall) and thickness of stratification (laminationbed scale), the nature and continuity of bed boundaries, and the prevalence of the described features (i.e. isolated or dominant in a succession). Several factors can influence the geometry and appearance of cyclic step deposits. Bedform deposit dimensions generally relate to the morphology of associated bedforms. However, scaling relationships may be ambiguous because bedform deposit size also depends on the degree of subsequent erosion (Fig. 8;Yokokawa et al., 2009;Vellinga et al., 2018;Slootman & Cartigny, 2020). Similar observations have been made for sedimentary structures associated with ripples and dunes where cross-set thickness is related to both aggradation rate and bedwave height distribution (Bridge, 1997;Leclair et al., 1997). Several studies have identified aggradation rate as the main controlling factor on cyclic step deposit architecture ( Fig. 1F; Hage et al., 2018;Vellinga et al., 2018;Slootman & Cartigny, 2020). Moreover, recent flume experiments by Ono et al. (2021) demonstrate that hydraulic jump size, stability and migration rate as well as grain-size distribution are also important. Experiments reveal complex, variable successions produced by cyclic step hydraulic jump dynamics under overall constant flow conditions, which share characteristics with the deposits of other upper-flow-regime and lower-flow-regime bedforms (Ono et al., 2021).
The stratigraphic products of cyclic step bedforms (10 to 140 m wavelengths, <10 m wave heights) formed by the passage of multiple turbidity currents with mixed sediment loads (including mudgravel grain sizes), under low-aggradation conditions are captured in this study (Figs 5A,16 and 17). Repeat bathymetry surveys from the Squamish prodelta indicate stable, periodic hydraulic jump upslope-migration during flow events (for example, Figs 2D and 6), whereas the nature of hydraulic jumps in the upper Monterey Canyon is unclear and may be more irregular.
Stratigraphic reconstructions suggest that in these cases, stratification may occur at the bed scale (i.e. decimetre scale) where between-flowunsteadiness results in nested, sharply bounded beds, which are subtly offset upslope (Figs 9A, 10A, 16A, 17B and 17C). Although deposit variability exists, asymmetrical geometry and local backset stratification is evident and diagnostic in each succession. It may be more difficult to recognize cyclic step deposits in 2D strike-oriented outcrops where backset bedding is not apparent (Figs 9A, 10A, 17B and 17C). In addition, as exemplified in the upper Monterey canyon dataset, inconsistent and more diverse deposit geometry and stratification may result due to greater erosion and reduced preservation , or the potential occurrence of small and/or unstable hydraulic jumps (Fig. 10, Ono et al., 2021).
When considered as a whole, the cyclic step successions documented here are distinct from the products expected from lower-flow-regime bedforms, which are dominated by consistent foreset stratification and preservation of steeper lee slopes ( Fig. 17; Slootman & Cartigny, 2020). However, it is less clear whether other upperflow-regime bedforms (for example, antidunes), which may also be relatively long-lived (West et al., 2019), could have comparable depositional signatures in deep-water settings. In addition, the identified recognition criteria may not be valid for all styles of cyclic steps as variability in aggradation rate, hydraulic jump dynamics, flow characteristics, sediment calibre and bedform longevity may result in different stratigraphic expressions.

Palaeoflow conditions
Bedform deposits are important for palaeoenvironmental interpretations and can preserve morphometric information useful for reconstructing the nature and evolution of sediment transport from ancient successions (Allen, 1982;Best & Bridge, 1992;Bridge, 1997;Leclair et al., 1997;Alexander et al., 2001;Jobe et al., 2012;Cartigny et al., 2014). The presence and architecture of Nanaimo Group deposits exposed around the Twin Beaches Peninsula suggest that the succession formed under supercritical flow conditions in a fluctuating but overall low aggradation regime, such as that found in channels of the Squamish prodelta and Upper Monterey Canyon (Figs 7C and 16A).
The dimensions of bedforms in the modern study areas and bedform deposits in all three locations are also relatively consistent (Fig. 16). The scale of deposits is important for directly tying features in the rock record to bedforms observed on the seafloor and their associated flow characteristics. Thus, the slightly smaller bedform deposits observed at the Twin Beaches Peninsula compared to those in modern reconstructions (Fig. 16) could reflect: smaller formative bedforms due to slight differences in slope gradient or flow characteristics (Cartigny et al., 2011;Kostic, 2011;Fricke et al., 2015;Normandeau et al., 2016;Slootman & Cartigny, 2020); more extensive reworking (Hage et al., 2018;Vellinga et al., 2018;Vendettuoli et al., 2019); or a bias related to comparing outcrop and bathymetry datasets. The longer timescales involved in ancient deposits enhance the likelihood of encompassing large, extreme events which could affect the size and preservation of deposits (Miall, 2015). In addition, both datasets include uncertainties related to differentiating individual event beds, which could affect deposit size distribution. Deposits reconstructed from bathymetry datasets are likely composite and consist of unrecognized, smaller geometries due to the occurrence of multiple flows between two surveys, while bed amalgamation can make individual units difficult to delineate in outcrop.
The similar scale of seafloor bedforms and bedform deposits in the investigated areas is likely related to properties of flows that can occur in multiple turbidity-current-dominated environments and water depths. Observations of flows in Squamish prodelta channels and Monterey Canyon, although not from the exact location of the bathymetric surveys used in this study, are characterized by turbidity currents with relatively fast, thin and dense basal layers (1 to 3 m thick, reaching 1 to 5 m s À1 velocities) that interact with bedforms ( Fig. 5B; Postma et al., 1988;Sohn, 1997;Hughes Clarke, 2016;Jobe et al., 2017;Paull et al., 2018;Wang et al., 2020). Similar flow characteristics have been previously associated with other relatively small-scale, coarse-grained, upslope-migrating bedforms based on numerical and conceptual models of cyclic steps (Cartigny et al., 2011;Symons et al., 2016). Thus it is likely that energetic, stratified, highdensity flows are ubiquitously responsible for bedforms of this type and also formed the bedform deposits of the Cretaceous Gabriola Formation.

Palaeoflow direction
Bedform deposits are not only significant for providing insight into flow and sediment transport conditions, but their associated stratification can be used to reconstruct palaeoflow direction (Allen, 1982;Rubin, 1987). Typically this stratification occurs at the scale of laminations or thin beds, which form sets bound by larger bedding surfaces (Allen, 1982). In the cyclic step deposits described here, low-angle stratification occurs within medium to very thick beds where smallerscale internal stratification is absent, crude or variably oriented. In addition to potentially long length scales, these factors make the precise determination of palaeoflow direction from stratification alone more challenging.
Palaeoflow directions can be directly constrained by measuring the orientation of the basal surface of trough-shaped deposits, as shown in the modern systems (Figs 9, 10, 17B and 17C). However, in order to do this effectively, it is important to identify the component of the 3D surface (upslope/foreset, edge or downslope/backset portion) being measured. This can be accomplished by determining the general outcrop orientation relative to the palaeoflow direction either by: independent measurements of other nearby palaeoflow indicators (for example, ripple foresets and sole marks); or examination of overall deposit architecture where subtleties in individual deposit geometry, proportions and stacking can differentiate depositional dip and strike-oriented sections ( Fig. 17B and 17C). The measured strike, down-dip and up-dip direction of edge, foreset and backset surfaces, respectively, can then be used as the palaeoflow direction. Backsets tend to have overall low dips (<10°), which may be difficult to measure in the field and be particularly sensitive to structural tilting (Fig. 16A). However, stepped or undulatory segments can provide locally steeper surface measurements ( Fig. 13; Ono et al., 2021;Spinewine et al., 2009;Yokokawa et al., 2009;Cartigny et al., 2011). Interpretation of palaeoflow direction based on strike and dip measurements from the basal surfaces of the Nanaimo Group deposits were consistent with palaeoflow determined by measurement of other features (for example, ripple foresets; Fig. 14B).

Modern to Ancient comparisons in deepwater settings
The inaccessibility of modern deep-water environments has previously limited direct observations of deep-water sedimentary processes, and thus the ability to create and validate turbidite depositional models . This study demonstrates how integrated seafloor datasets, including repeat bathymetry, flow monitoring and sediment coring, can be leveraged to improve understanding of deep-water stratigraphy by providing new inspiration for evaluating and informing interpretations of the sedimentary record. These comparisons allow direct linkages to be established between bedforms produced by naturally-occurring turbidity currents and sedimentary structures observed in the rock record, in this case providing robust support for positive identification of cyclic steps in deep-water strata.
Differences in the spatial and temporal resolution of datasets make the comparison of modern and ancient settings difficult but also complementary (Mutti & Normark, 1987). Modern investigations can only detect features above the resolution limits of imaging techniques (typically >1.0 m horizontal, >0.1 m vertical) and monitor processes that operate over short timescales (on the order of days to years). Integrated seafloor datasets including repeat monitoring and surveying are presently biased towards sandy, shallow environments (<250 m water depth) where high resolution imaging is more accessible and flow events occur more frequently (sub-annually); however, it is uncertain how these observations extend to deeper-water, mud-rich systems and longer timescales Vendettuoli et al., 2019). For example, roughness calculations from the deep-water, muddy Congo Canyon suggest an absence of bedforms . On the other hand, outcrop investigations can capture smaller features (typically at centimetre to metre scale) that represent longer term perspectives (thousands to millions of years) but are biased by incomplete preservation and limited context.
Between seafloor and outcrop investigations, upper-flow-regime bedforms have been reported spanning a wide range of sizes, on the order of centimetres to kilometres (e.g. Symons et al., 2016;Lang et al., 2017;Cornard & Pickering, 2019). Although similar processes may operate at different spatial and temporal scales (Fig. 1A), it is important to consider compatibility in terms of size and duration before making direct modern to ancient comparisons. Study area and data acquisition limitations still inhibit the identification of 100 to 1000 m scale bedform features (for example large-scale sediment waves and large-scale scours; Fig. 1A) in most outcrops, while smaller features often observed in outcrop are not resolvable on the seafloor (Mutti & Normark, 1991). This study features cyclic steps, morphologically equivalent to small-scale sediment waves, and their associated sedimentary structures at an intermediate scale (on the order of metres to tens of metres; Fig. 1A). Modern study areas capture flow conditions and low net accumulation over surveyed time periods while identification in the Nanaimo Group succession demonstrates that preservation can occur over geological timescales and at greater water depths.
Although cyclic steps are not present in all modern deep-water systems and their deposits have been reported from some deep-water outcrops (e.g. Lang et al., 2017;Cornard & Pickering, 2019), bedforms are much more widely identified on the seafloor than their sedimentary structures documented in deepwater successions. Under-representation in the rock record could be a result of insufficient recognition or, in some cases, recognition may not be possible due to indistinct stratigraphic products (for example, large deposits, uniform grain-size distributions and non-unique facies). The apparent discrepancy in cyclic step deposit occurrence may also be related to bedform preservation, which may be poor (as observed in the Squamish prodelta and upper Monterey Canyon), and ultimately only favoured under specific conditions and/or in certain depositional environments. Additionally the role of bedforms in larger infrequent events, which deposit or flush large volumes of sediment in channels and have the largest impact on the stratigraphic record (cf. Mountjoy et al., 2018;Vendettuoli et al., 2019), is at present poorly constrained.

CONCLUSION
Repeat bathymetric surveys from two modern environments (Squamish prodelta, British Columbia and Monterey Canyon, California) establish process to product relationships for macro-scale upslope-migrating bedforms (<300 m wavelength, <8 m wave heights) interpreted as cyclic steps. Quantitative comparisons of reconstructed bedform deposits from these datasets are used to inform interpretations of cyclic steps in a Late Cretaceous (Nanaimo Group) deepwater succession exposed on Gabriola Island, Canada. Bedform deposits from the two modern systems and ancient outcrop strata are similar in scale, facies and depositional architecture, with each characterized by low-angle backstepping, trough-shaped lenses, which are typically tens of metres long/wide, <1 m thick and composed of massive sands/sandstone. The identification of comparable deposits in prodelta channel, upper canyon and deep-water slope channel environments supports process-based relationships associated with low-aggradation cyclic steps (on average 40 to 55 m wavelengths, 1.5 to 3.0 m wave heights, 39 to 74% deposit reworking). Bedforms of similar size and morphology are found abundantly in coarse-grained turbidite-dominated environments. Thus, this dataset may represent typical bedform deposit dimensions and geometries that form beneath turbidity currents with dense basal layers 1 to 4 m thick and 1 to 5 m s À1 flow velocities under low-aggradation conditions.
The scale and three-dimensional architecture of deposits are key criteria for identification of the bedform deposits in outcrop, while facies alone are non-unique in the examined datasets. For cyclic steps that persist over multiple flow events, backset stratification develops at the bed-scale where between-flow-unsteadiness and low-aggradation result in nested, erosion-surface-bound units. Deposit architecture and stacking is useful for determining the general palaeoflow direction while more precise directions can be obtained by appropriately measuring the orientation of foreset, backset and/or edge components of basal bounding surfaces. More work is clearly required to establish recognition criteria and understand process to product relationships for the wide variety of seafloor bedforms, as well as sedimentary structures documented in the rock record. Integration of novel seafloor datasets provides added insight into palaeoenvironmental interpretations of deep-water stratigraphy that in turn can fill gaps in current understanding of seafloor processes, such as the extent to which processes operate in deeper environments and over longer timescales. support from a NSERC Alexander Graham Bell Canada Graduate Scholarship. M. Cartigny is supported by a Royal Society Research Fellowship. M. Clare acknowledges funding from the CLASS program (NERC Grant no. NE/R015953/ 1). Paul Durkin and Sarah Southern provided assistance with fieldwork and contributed to early discussions. Additional discussions with Brain Romans, Jacob Covault and Svetlana Kostic positively influenced our interpretations of the outcrop belt. The authors would also like to thank J€ org Lang, an anonymous reviewer and Associate Editor Alexandre Normandeau for their thoughtful and helpful comments on this manuscript.

DATA AVAILABILITY STATEMENT
The data that supports the findings of this study are available in the supporting information of this article. The Monterey bathymetry data were derived from the following resource available in the public domain (Seafloor Mapping Lab): https://seafloor.otterlabs.org/.