Large Bedforms Research Articles

Seabed geology of the Canadian eastern continental shelf

The physiography of the continental shelf off eastern Canada is irregular, developed by glacial erosion of a previously fluvially-dominated landscape. Northern shelves are deeper than southern shelves. Most surficial sediments on the shelf are relict or palimpsest. The principal modern source of sediment to the northern shelves is ice rafting and iceberg scour reworking of Quaternary sediments. Southern shelves receive sediment through erosion of Quaternary sediments; only small amounts of fine-grained sediment derived from coastal erosion and rivers escape from the coastal zone. Regional maps of sediment texture, carbonate content and heavy mineralogy consequently show differences between the northern and southern shelves. Large areas of the shelf show little net deposition. On the northern shelves, there is a surface veneer up to 0.5 m thick derived from ice rafting and iceberg turbation of underlying Quaternary sediment, modified by south-flowing currents [ Woodworth-Lynas et al. (this issue) Continental Shelf Research, 11, 939–961]. The overall effects of former iceberg turbation may extend to a depth of 10 m sub-bottom. On the southern shelves, bioturbation and perhaps storm-related currents rework exposed Quaternary sediments more slowly. Muds accumulate in deep basins on the shelves at rates of about 0.5 m per 1000 years; this accumulation is probably episodic and related to major storms reworking sediment from the surface sediment veneer in shallower areas of little net deposition. In water depths less than 110 m sand and gravel have formed as a result of reworking in the coastal zone during the post-glacial transgression. Over large areas of Georges Bank, the eastern Scotian Shelf and the Grand Banks of Newfoundland, such sands are mobilized during storms to form a wide suite of bedforms [ Amos and Judge (this issue) Continental Shelf Research, 11, 1037–1068]. Elsewhere, particularly in deeper water, sandy surfaces appear moribund or inactive and large bedforms are relict. Armoured gravel surfaces are common on the banks of the western Scotian Shelf. In areas of coastal transgression, barrier beach and foreshore sediments are superimposed on partially eroded estuarine and lagoonal sediments. In areas of coastal regression, deep water muds are being reworked by waves as relative sea level falls, to produce a surface lag of coarser sediment.

Wave transformation

Wave transformation is studied for a new data set collected on a mobile bed beach in a hydraulic model. Irregular waves are simulated on a beach with parallel contours. It was found that wave refraction is well described by Snell's Law, that wave shoaling is slightly overpredicted by linear theory, that bottom friction effect is small, even in hydraulic models which exhibit large bed forms, that percolation effect is small, even though the model grain size is quite large compared to the wave dimensions and that wave attenuation by wave saturation must definitely be considered.

Evolution of Pliocene to Recent abyssal sediment waves on Bounty Channel levees, New Zealand

Levees bordering Bounty Channel 900 km east of New Zealand accommodate a 400 m-thick sequence (maximum) of sediment waves that have formed since Pliocene times. These bedforms, with amplitudes of 2–17 m and wavelengths of 0.6–6 km occur in 4100–4900 m of water and were formed by turbidity currents, as indicated by their restriction to levee backslopes, the frequent occurrence of turbidites in cores and the preferential but not exclusive development of waves on the left-bank levee in accord with the Southern Hemisphere coriolis deflection. The wave field was instigated in the Late Pliocene when glacially lowered sea level allowed rivers draining the Southern Alps of South Island to discharge directly into Bounty Channel and its attendant canyons. The field grew vertically through the coalescence of small waves into larger bedforms that continually migrated across and up levee backslopes at an average rate of 5.6m/100 yrs. Wave growth decreased into the Late Pleistocene probably in response to progressive containment of turbidity currents as the relief of Bounty Channel increased to 200 m or more. The glacial periods of wave growth were interrupted by interglacial interludes of quiescence when the field was draped mainly by pelagic calcareous ooze.

Changes in paleocurrents during the development of an obliquely convergent plate boundary (Sulaiman fold-belt, southwestern Himalayas, west-central Pakistan)

Pre-orogenic Cretaceous shelf sandstones at the northern and southern ends of the Sulaiman Range in west-central Pakistan show basically westward dispersal, which is off the craton. However, Paleocene marine sandstones in the south spread east-southeast, and early Eocene deltaic sandstones in the north spread southeast and north. These are mostly onto the craton, and indicate slope reversal of the Cretaceous shelf. Post-Oligocene fluvial molasse shows mostly southward drainage. A 3852 m section of molasse in the northern Sulaiman Range comprises a lower 2378 m of claystones and upwardly fining sandstones that represent meandering rivers and an upper 1474 m that are dominated by conglomerates that represent gravelly braided rivers. Paleocurrents in the lower, meandering, part of the section were generally southward, but very variable, whereas those in the upper, braided, part were more uniform and show eastward flow. A 3146 m section of molasse in the southern Sulaiman Range comprises (1) a basal 1160 m sequence of claystones and sandstones (meandering streams with very diverse flow directions, overall to the southwest), (2) a middle 1603 m sequence dominated by pebbly sandstone with large bedforms (big sandy braided rivers with less diverse flow to the southwest), and (3) an upper 383 m conglomeratic sequence (gravelly braided streams that still flow uniformly southwest). Rivers in the early molasse paralleled the modern Sulaiman Range in the north and flowed obliquely into it in the south, which suggests no uplift of the modern front of the range until the onset of modern transverse (eastward) drainage and the associated influx of conglomerates. In present and admittedly uncertain lithostratigraphic terms, uplift of the modern front of the range apparently began about 700 m lower in the north than in the south. The northern and southern molasse sections otherwise show broadly similar histories of river development.

Stability of giant sand waves in eastern Long Island Sound, U.S.A.

A combination of a highly accurate bathymetric surveying technique and in-situ submersible observations and measurements were used to assess the migrational trends and morphological changes of large sand waves ( Ht ≤ 17 m) in eastern Long Island Sound. Although residing in a high-energy tidal environment characterized by a net westward sediment flux, the large bedforms are relatively stable over the short term. Over a 7 month period, 55.1% of a total 2942 m of sand wave crestline lengths migrated less than the horizontal accuracy limits of navigation (2 m). Approximately 35% of the remaining sand wave crests migrated less than 4 m. Net migration of the sand wave crests in the study area was 0.2 m. In addition, the bulk form (center of area in profile view) or the base of the sand waves showed little, if any, movement. These data, in conjunction with flow data within the sand wave field, suggest that net migration rates are greater than the time span of this study and/or the sand waves move in response to large residual flows created by high-energy, aperiodic storm events. The latter scenerio suggests that day to day processes only serve to rework and modify the sand waves.

Experiments on Bed Configurations in Fine Sands Under Bidirectional Purely Oscillatory Flow, and the Origin of Hummocky Cross-Stratification

ABSTRACT We made experiments on bed configurations in bidirectional purely oscillatory flows at a wide range of oscillation periods T and maximum orbital velocities Um in two closed flow ducts, 0.1 m and 0.4 m wide. The motivation was that understanding of bed configurations in oscillatory flows with long periods and high velocities during storm-induced deposition in the shallow ocean is needed for interpreting the stratification in such deposits. The ripples produced were classified geometrically as two-dimensional (2D) or three-dimensional (3D) in plan-view geometry and dynamically as oscillatory-current ripples (dependent upon the patterns of flow and sediment transport specific to the existence of the oscillation itself) and reversing-current ripples (dependent on he existence of the current, in one direction or the other, during the oscillation). Runs in the small duct with effectively 0.19 mm and 0.30 mm sand showed the familiar small 2D oscillatory-current ripples at short T and low Um passing into much larger ripples, with spacing up to 2 m, at periods of 6-20 s and velocities up to 1 m/s, with no apparent limit on size with increasing T and Um. Even in such a narrow duct there was some evidence of 3D geometry of the large ripples. Runs in the large duct with effectively 0.11 mm sand at a single period of about 9.5 s showed a progression from small 2D oscillatory-current ripples through small 3D reversing-current ripples to large round-crested 3D oscillatory-current ripples with prominent superimposed 3D to 2D reversing-current ripples. The large 3D ripples, which increased in spacing to 1-3 m at the h ghest attainable Um of 1 m/s, shifted in position and changed in size with time, seemingly at random, causing substantial local and temporary erosion and deposition at a given point on the bed. There was no net aggradation during the runs, but a deposit was synthesized by taking a time series of sidewall bed profiles during a run in which large 3D ripples developed from an initially plane bed and changed irregularly in size and position as they grew toward equilibrium, and then plotting the profiles one above the previous with a constant upward increment. The resulting stratification is strikingly similar to much hummocky cross-stratification in ancient sandstones. This suggests that some hummocky cross-stratification is generated during sediment fallout from strong purely oscillatory flows at moderate to long oscillation periods as large 3D oscillatory-current bed forms develop from a planar bed during strong but waning flow.

Flume Experiments on the Transport of Heavy Minerals in Gravel-Bed Streams

ABSTRACT Heavy-mineral transport and deposition by shallow unidirectional flows in a gravel-bed channel were studied in a water-recirculating sediment-feed flume 6 m long and 0.15 m wide. The sediment was poorly sorted gravel with mean size 3 mm and 3% by weight of finer magnetite, lead, and tungsten. The heavy minerals became concentrated into a thin layer (the heavy infralayer) composed of nearly 100% heavy minerals lying beneath a thin surficial layer of low-density sediment (the light supralayer). Heavy minerals were not transported past a given location on the bed until the heavy infralayer was fully developed there. Heavy minerals were transported on the upper surface of the heavy infralayer only when local and temporary erosion of the light supralayer, caused by large clasts, clast jams, or downstream-moving dunelike bedforms, exposed the heavy infralayer. In two runs, one with sediment feed and one without, overall degradation of the bed was forced by slow lowering of a tailgate at the downstream end of the channel. In the sediment-feed run a downstream-prograding heavy infralayer formed as in the nondegrading runs. In the run without feed, the surficial heavy-mineral concentration increased gradually at all points along the channel to the end of the run. In light of these results, we hypothesize that ultimately the surficial heavy-mineral concentration would reach a steady state and the bed slope would change until the flow can transport all of the heavy-mineral sediment supplied by local exhumation. Although no runs were designed for aggradation, a too-small initial slope caused overall bed aggradation in the early parts of two runs. Heavy minerals in the feed sediment were transported only a short distance downstream to form a zone of higher but downstream-decreasing heavy mineral concentration extending vertically through the deposit. We hypothesize that the downstream length of this zone would increase as the heavy-mineral feed rate increases relative to the bed aggradation rate.

Bedload-sediment transport through the Connecticut River estuary

The Connecticut River is the third-largest river on the east coast of the United States, and its estuary is microtidal and sand dominated. Discharge of the river annually varies from tens to thousands of m 3 /s, and circulation in the estuary spans the full range of estuarine mixing modes. During half of the year, the estuary is partially mixed with mutually evasive tidal currents established between the flood-dominated channel and ebbdominated shoal margins. The distribution, orientation, and tidal modification of large bedforms in the estuary demonstrate that there are major pathways of bedload-sediment transport through the estuary to Long Island Sound. Calculations of shear stress on the bed, based on current-velocity measurements at 47 stations through full tidal cycles during low-discharge conditions, indicate that most of the estuary is essentially ebb dominated with respect to bedload-sediment transport. Observations of bedform occurrence and orientation during high discharges indicate that bedload transport is ebb oriented throughout the entire estuary. Analysis of dredging records and historic charts of the estuary since 1915 show little net change in bathymetry and suggest that the bed of the estuary may be in equilibrium with its sediment supply. The combined evidence argues that the Connecticut River estuary is not a site of significant bedload storage, and that it is a conduit of sand into Long Island Sound.

Large-Scale, Low-Amplitude Bedforms (Chevrons) in the Selima Sand Sheet, Egypt

Landsat images of the Selima sand sheet in southwestern Egypt display alternating light and dark chevron-shaped patterns that occur downwind from low scarps and major dune fields. Images acquired between 1972 and 1988 indicate that these features move as discrete bedforms at a rate of up to 500 meters per year. Extremely long-wavelength (130 to 1200 meters), low-amplitude (10 to 30 centimeters) bedforms were measured in the field; the light chevrons seen in the orbital data may be thin accumulations of active sand sheet deposits in the lee of these bedforms. Dark chevrons contain an admixture of coarse-granule lag deposits that are continually winnowed by aeolian erosion on the windward sides of the large bedforms. Sediment transport budgets derived from orbital and field analyses suggest net movement of up to 83,000 cubic meters per year for a single light chevron; such measurements can be used as a check on similar calculations from dunes and other smaller scale features to determine sand transport budgets for large areas of the eastern Sahara.

Large bedforms and associated hydraulic conditions within microtidal-inlet channels, southern Gulf of St. Lawrence, Canada

A variety of large bedforms, at two scales (sand waves, dunes), occur within the throat section of-small microtidal-inlet channels. The larger forms, which have an average length of 32 m and height of 1.0 m, remain ebb oriented throughout the tidal month but are only actively modified at large tides. The smaller forms, which may be superimposed on the larger, are partially reworked each tidal cycle but are also predominantly ebb oriented. Tidal-current measurements confirm the dominance of ebb-tidal flow in the inlet throats and indicate that despite the small tidal range (maximum 1.1 m), ebb currents may exceed 1.0 m s−1 with associated shear velocities of 0.14 m s−1.

Early Permian giant cross-beds near Nqutu, South Africa, interpreted as part of a shoreface ridge

Sandstones and siltstones of a giant cross-bed sedimentary association at the base of the coal-bearing Vryheid Formation cannot be satisfactorily explained in terms of the currently accepted fluviodeltaic model for these strata. In view of the sedimentary structures and associational characteristics of the facies, a shoreface ridge interpretation is adopted instead. The sandridge, probably detached from the shoreface, was situated on the slope towards the basin plain and associated with some elements of a minor submarine fan complex. Rapid shoreface by-pass in the course of a marine transgression seems to have provided favourable conditions for sandridge formation. The major bedform appears to have been in water of about 45 m depth, possibly some kilometres from the contemporary shoreline. There is evidence that it was mainly constructed by the action of strong, easterly flowing currents, probably of relaxant or geostrophic origin. These were aided by ebb-dominant tidal currents. Possible actualistic models include North Sea and Bering Sea sandridges. Currents of the strength needed to build such large bedforms could only have existed in an open-ended seaway and it is suggested that the Ecca sea must have communicated, at least intermittently, with open ocean to the northeast and possibly the southeast as well as towards the west.

Multitrack surveying of large bedforms

Data are presented from multitrack surveys of large, sandy bedforms in the Fraser River estuary, Canada. A detailed contour map and block diagram provide information on three-dimensional bedform geometry that could not be obtained using conventional survey methods. The technique has implications for estimation of bedload transport and for bedform classification.

Paleocurrents Beside an Obliquely Convergent Plate Boundary (Sulaiman Foldbelt, Southwestern Himalayas, West-Central Pakistan): ABSTRACT

Cretaceous to Holocene paleocurrents at the north and south ends of the Sulaiman Range show flow trending westward, then variably eastward, southward, and finally eastward. Pre-orogenic Cretaceous shelf sandstones show dispersal to the southwest, north-northwest, and west-southwest off the craton. However, Paleocene marine sandstones in the south spread east-southeast, and early Eocene deltaic sandstones in the north spread southeast and north. These sandstones indicate slope reversal of the Cretaceous shelf. Post-Oligocene fluvial molasse shows mostly southward drainage. The 3.852 in northern molasse section comprises a lower 2,378 m of claystones and upwardly fining sandstones that are thought to represent meandering rivers, and an upper 1,474 m dominated by conglomerates, representing gravelly braided rivers. Lower paleocurrents were to the southeast, but variable, whereas upper paleocurrents were more uniform and gradually shifted from southeast to east The 3,146-m southern section comprised (1) a basal 1,160 m claystone and sandstone sequence (meandering streams with very diverse flow directions, overall to the southwest), (2) a middle 1,703 m sequence dominated by pebbly sandstone with very large bedforms (big sandy braided rivers with less diverse flow to the southwest), and (3) an upper 283 m conglomeratic sequence (gravelly braided streams that still flow uniformly westsouthwest).more » Early flow obliquely toward the Sulaiman Range suggests no uplift of the range until the onset of modern eastward drainage and the influx of conglomerates, which occurred 700 m lower in the north. The northern and southern sections otherwise show similar histories of fluvial sedimentation.« less

The Hawkesbury Sandstone South of Sydney, Australia: Triassic Analogue for the Deposit of a Large, Braided River

ABSTRACT The Hawkesbury Sandstone is a Triassic sheet sandstone, extensively exposed in the Sydney Basin, New South Wales, particularly along the coast near Sydney. Unidirectional paleoflow in the sandstone, its freshwater biota, and abundant mudrock intraclasts indicate fluvial deposition. Sheet morphology, low paleocurrent variance, abundant erosion surfaces, and the paucity of in situ mudrocks point to a braided fluvial system. Three facies assemblages have been recognized: stratified sandstone, massive sandstone, and a minor mudrock assemblage. The stratified sandstone assemblage is dominated by stacked sets of planar cross-strata and minor trough cosets in sequences 6-23 m thick, bounded by erosion surfaces. Significant paleocurrent changes between channel sequences indicate that they were initiated by the avulsion of major channel systems. In some cases the channel sequences fine upward, with planar cross-stratal sets overlain by trough sets that decrease in magnitude upwards, fining up into mudrocks. The massive sandstone assemblage occurs principally as massive sandstone in elongate erosional features oriented transverse to paleoflow. The massive sandstone commonly contains large mudrock intraclasts and is attributed to failure of high channel banks and/or large bedforms during falling-water stages. The mudrock assemblage comprises rippled and horizontally laminated fine sandstone, siltstone, and shale, with minor mudstone. It is attributed primarily to floodplain deposition, but abandoned channel fills are also present. The Hawkesbury Sandstone does not conform to existing models for braided-fluvial deposition in that planar cross-strata accumulated in deeper parts of channels, whereas trough cross-strata formed in shallower water. The large scale of planar cross-strata (sets up to 7.5 m thick), mudrock intraclasts (up to 38 m long), bank-collapse scars (up to 11 m deep) and abandoned channel fills (up to 18 m deep) show that the Hawkesbury River was very large, with deep main channels and high but variable discharge.

Formation of Scalloped Cross-Bedding Without Unsteady Flows

ABSTRACT Scalloped cross-bedding--compound cross-bedding with internal bounding surfaces that cyclically scoop into the previously deposited foresets and into the sediment below the set--is a common and distinctive structure in eolian, fluvial, tidal, and nearshore-marine sands. Scalloped cross-bedding in shallow-marine deposits previously has been interpreted to be produced by cyclic flows, such as neap-spring tidal flows, which are known to cause cyclic fluctuations in the depth of scour in the troughs of migrating bedforms, but scalloped cross-bedding also originates by a process that does not require fluctuating flow: migration of small bedforms across the lee slopes or along the troughs of larger bedforms. Intersections of the troughs of the two sets of bedforms form topographicall low scour pits, and cyclic passage of these scour pits through the outcrop plane--the plane that later becomes an outcrop surface--causes the lower-set boundary to rise and fall. Scalloped cross-bedding formed by fluctuating flow superficially resembles that formed by superimposed or intersecting bedforms, but, as illustrated in three-dimensional computer plots, the two kinds of structures commonly can be distinguished by directional properties of the bedding. Scallops deposited by alongslope-migrating, superimposed bedforms have cross-bed and bounding-surface dip patterns that lack bilateral symmetry and have cross-bed dips that are asymmetrically distributed relative to bounding-surface dips. Scallops with dip patterns that are bilaterally symmetrical and with cross-bed dips that are symmetrically distributed relative to the bounding-surface dips can be produced either by fluctuating flow or by downslope or upslope migration of superimposed bedforms. An example of nearshore-marine scalloped cross-bedding of Pleistocene age was examined in detail in a coastal terrace of Monterey Bay, California. The three-dimensional structure and directional properties of the bedding suggest that the deposit was produced by a series of small bedforms migrating offshore, down a rip channel that was bounded on one side by a migrating oblique bar.

Estimates of sand transport in the Oosterschelde tidal basin using current-velocity measurements

Estimates of sand transport have been made from mid-depth velocity measurements in the Oosterschelde tidal basin, southwestern Netherlands. This involves: (1) The use of a two-region composite boundary-layer model in deriving the local spatially averaged bed shear stress from velocity distributions over large bedforms (megaripples and sandwaves). The two regions are divided at 100 cm above the bed, with an outer-region roughness length of 1.5 cm and an inner-region roughness length of 0.15 cm. (2) The application of a modified Bagnold's transport equation with the transport coefficient K depending on dimensionless excess shear stress in the form of a power law. The calculations indicate very active sand movement in the present-day Oosterschelde tidal basin. Local sediment circulation may develop around shoals. Channel-floor erosion may occur in parting areas of sediment transport. The net sediment transport is mainly in the ebb direction. Estimates of sediment budget for the last 20 years show that large amounts of sand have been eroded from the Oosterschelde tidal basin and transported to the ebb-delta area. These results are comparable with other estimates from echo-sounding data, reflecting the rapid adjustment of the basin to the increased tidal prism. Such active sediment redistribution and major morphological evolution processes would have important influence on the structural organization and sequential development of estuary-ebb delta deposits. Information about the sedimentary processes in present-day estuary tidal basins is valuable for a better understanding of ancient examples.

Sediment slides and turbidity currents on the Laurentian Fan: Sidescan sonar investigations near the epicenter of the 1929 Grand Banks earthquake

The epicenter area of the 1929 Grand Banks earthquake on the continental slope south of Newfoundland has been investigated using Sea MARC I, a deeply towed, midrange sidescan sonar with a 4.5-kHz subbottom profiler. Shallow slides pass downslope into debris flows on the muddy continental slope east of the epicenter. At the head of the Eastern Valley of the Laurentian Fan, west of the epicenter, arcuate slide scars cut undisturbed upper-slope sediment and lead downslope to a lineated erosional seabed. At a water depth of about 1600 m, this erosional seabed passes into extensive fields of 100-m-wavelength gravel waves situated on the broad, irregular valley floor. The gravel waves become better developed downslope and extend at least to water depths of 3000 m. All these morphological features appear fresh on the sidescan sonograms, suggesting that they date from the 1929 earthquake event, and the distribution of slides corresponds to the area of instantaneous cable breaks in 1929. The upper limit of erosion on valley walls suggests that the 1929 turbidity current was less than 300 m thick. Timing of cable breaks downfan suggests that flow velocities were sufficient to rework gravel deposits into large bedforms during waning flow stages over elevated areas of the valley floor. Similar cross-bedded coarse sands and gravels are common in ancient channel deposits.

Seabed processes on the northeastern Grand Banks of Newfoundland; Modern reworking of relict sediments

On the northeastern edge of the Grand Banks of Newfoundland, between 140 and 70 m depth, a thin Holocene sediment veneer unconfortmably overlies Tertiary siltstones and sandstones. These surficial sediments show a depth-controlled facies distribution. Below 90 m the facies range from continuous fine sand, through lag gravel and sand ribbons with arcuate sand waves and megaripples normal to the axis of the ribbons. Above 90 m depth alternating sand ridges up to 3 m in height and lag gravels occur. Megaripples and arcuate sand waves (400 m wave length) with megaripples on their stoss side are developed on the terrace. The regional geology suggests that many of the surface sediments were inherited from Pleistocene deposits subsequently reworked in coastal environments by an early Holocene transgression. Side-scan sonar surveys, bottom photography and submersible observations demonstrate intermittent sediment mobilization of recent material. In particular, photographs show asymmetrical ripples in well-sorted sand after seasonal storms. Side-scan sonar coverage in the Hibernia area shows fields of two- and three-dimensional megaripples (5 m length) which appear to be migrating across trawl marks left by recent fishing operations. These sedimentary bedforms are attributed to unidirectional storm driven currents, ocean currents and extreme waves. The larger bedforms may reflect paleoceanographic conditions while the small bedforms appear to be modern.

Sand Waves in Lower Cook Inlet, Alaska

Acoustic (geophysical) methods were applied to investigate the sea-floor geological conditions, particularly sand waves, which might place constraints on construction in an area of potential offshore development in lower Cook Inlet, Alaska. Multisystem, high resolution, marine acoustic survey data were collected using a sparker, tuned transducer, acoustipulse, fathometer, and side-scan sonar. Based on 65 sea-floor samples obtained in the northern portion of the study area, the sea-floor soils consisted of over 90 percent sand by volume. This portion of lower Cook Inlet contained the most numerous sand waves and contained sand waves of 5 ft (1.5 m) or greater height over an 80 percent length of the survey lines. The possibility of sand wave movement is examined by reviewing the two phase (water-sand) flow regime and by repeat geophysical profiling. The present flow regime appears to be conducive to the formation of ripples and dunes, but tidal flow durations may be insufficient for the formation of large bedforms.

Sand waves on an epicontinental shelf: Northern Bering Sea

Sand waves and current ripples occupy the crests and flanks of a series of large linear sand ridges (20 km × 5 km × 10 m high) lying in an open-marine setting in the northern Bering Sea. The sand wave area, which lies west of Seward Peninsula and southeast of Bering Strait, is exposed to the strong continuous flow of coastal water northward toward Bering Strait. A hierarchy of three sizes of superimposed bedforms, all facing northward, was observed in successive cruises in 1976 and 1977. Large sand waves (height 2 m; spacing 200 m) have smaller sand waves (height 1 m; spacing 20 m) lying at a small oblique angle on their stoss slopes. The smaller sand waves in turn have linguoid ripples on their stoss slopes. Repeated studies of the sand wave fields were made both years with high-resolution seismic-reflection profiles, side-scan sonographs, underwater photographs, current-meter stations, vibracores, and suspended-sediment samplers. Comparison of seismic and side-scan data collected along profile lines run both years showed changes in sand wave shape that indicate significant bedload transport within the year. Gouge marks made in sediment by keels of floating ice also showed significantly different patterns each year, further documenting modification to the bottom by sediment transport. During calm sea conditions in 1977, underwater video and camera observations showed formation and active migration of linguoid and straight-crested current ripples. Current speeds 1 m above the bottom were between 20 and 30 cm/s. Maximum current velocities and sand wave migration apparently occur when strong southwesterly winds enhance the steady northerly flow of coastal water. Many cross-stratified sand bodies in the geologic record are interpreted as having formed in a tidal- or storm-dominated setting. This study provides an example of formation and migration of large bedforms by the interaction of storms with strong uniform coastal currents in an open-marine setting.