- Copyright © 2015, SEPM (Society for Sedimentary Geology)
Density-driven submarine flows, including turbidity currents, play an important role in the transfer of sediment into deep water. These bottom-hugging flows often produce flow-transverse bedforms along their path. A sedimentological and geophysical survey of the Stehekin River delta in Lake Chelan, Washington, reveals a downslope-elongate field of bedforms on the delta foreset associated with hyperpycnal discharges of the Stehekin River. An analysis of the bedform morphologies, delta geometry, and density contrast between lake and river water suggests that these hyperpycnal flows are Froude-supercritical. The bedforms are likely cyclic steps, flow-transverse bedforms that are bounded by stable hydraulic jumps between alternating subcritical and supercritical flow regimes. The ability to examine the three-dimensional bed configuration produced by natural density-driven flows adds valuable information to the body of experimental work focused on the behavior of cyclic steps in flumes.
Bedforms record valuable information about both the environments in which they were deposited and the nature of the flows under which they were formed. The dynamics of many bedforms, especially ripples and dunes, are well understood (see Ashley 1990 for review). Less well understood are bedforms produced under Froude-supercritical flows. A flow is Froude-supercritical when the wave speed at the flow interface exceeds the velocity of the flow. This balance is quantified in the case of subaqueous flows by the densimetric Froude number (Frd):
where U is flow velocity, Δρ is the density contrast between the two water masses, g is gravitational acceleration, ρ is the density of the flow, and D is flow thickness (after Hand 1974). The critical densimetric Froude number is most often assumed to be unity. In open-channel flows, Froude-supercritical flows (Fr> 1) occur within a narrow band of conditions in which the flow is very thin and/or fast moving. By contrast, there is a wider range of flow conditions under which two-layer flows are likely to be supercritical. This is in large part due to the reduced density contrast at the flow interface which results in a slower interfacial wave speed (the denominator in Equation 1), yielding a higher densimetric Froude number.
Antidunes are a class of flow-transverse bedforms that occur under Froude-supercritical flow conditions. The surfaces of antidunes are in phase with the free surface of the flow (Kennedy 1963). Whereas antidunes are able to migrate both upstream and downstream, the former is more commonly observed. Upstream-migrating antidunes are generally symmetric, with stoss and lee faces being of a similar angle. Antidunes are also described as ephemeral features that tend to grow, migrate, and collapse, only to regenerate in a different position, leaving little in the way of a clear sedimentary signature (Kostic 2011).
Cyclic steps are upstream-migrating bedforms that result from an alternating series of subcritical and supercritical flow regimes divided by steady, upstream-migrating hydraulic jumps. Fast-moving supercritical flow on the lee face of each step promotes erosion, while low-velocity flow following each hydraulic jump promotes sediment deposition on the stoss face, causing the bedforms to migrate in an upstream direction (Parker 1996; Kostic et al. 2010; Kostic 2011).
Classic experimental work by Hand (1974), as well as more recent studies by Winterwerp et al. (1992), Kostic et al. (2010), and Cartigny (2012) have explored bedforms produced by supercritical density currents in laboratory environments. Fewer studies have explicitly described supercritical bedforms in nature (e.g., Fildani et al. 2006; Lamb et al. 2008; Covault et al. 2014), but many studies suggest the predominance of cyclic-step-related bedforms in submarine environments, including the sediment waves common in many continental-slope environments (e.g., Kostic 2011; Cartigny 2012). These field examples come from marine environments, where energetic forces (waves, tides, currents) are likely to affect both flows and the resultant bedforms, thus making their formative mechanisms difficult to isolate.
Here we present data from a comparatively quiescent environment, the subaqueous delta of the Stehekin River in Lake Chelan, Washington, USA. Satellite and aerial images of the mouth of the Stehekin River during high-flow conditions suggest that the discharge of the river plunges, as evidenced by buoyant debris marking a plunge point and the absence of any surface plume spreading lakeward (Fig. 1). Geophysical and sedimentological data sets show a kilometer-scale train of bedforms that appear to be cyclic steps on the foreset of the delta. Our objectives are to: (1) describe the physical setting of these bedforms, (2) determine whether hyperpycnal discharges from the Stehekin River are Froude-supercritical, (3) determine the bedform type based on geometry and stability in estimated flow conditions, (4) link bedform formation to the dynamics of the Stehekin River, and (5) relate flow dynamics to the planform characteristics of the bedform field.
Lake Chelan is located in the northeastern Cascade Range of Washington State. It occupies a long, narrow valley of glacial origin (Whetten 1967). The lake is approximately 80 km long, with an average width of roughly 2 km. Lake Chelan has a maximum depth of 450 m (50 m below sea level), though much of the lake is shallower, with an average depth of ∼ 150 m. Lake level fluctuates by approximately 3 m annually due to outflow regulation for hydroelectric-power generation.
Lake Chelan is a cold, oligotrophic lake (Pelletier et al. 1989) (Fig. 2). The temperature of lake water below 100 m depth remains ∼ 6° C throughout the year. Surface warming during the summer months results in strong stratification from May through December, with a thermocline at a depth of ∼ 70 m (Pelletier et al. 1989).
The Stehekin River enters the northern end of Lake Chelan, near the village of Stehekin, Washington. It drains 830 km2 of mountainous, seasonally snow-covered terrain. Nearly all of the Stekein River watershed lies within the boundaries of North Cascades National Park or the Lake Chelan National Recreation Area. As a result, the watershed has been preserved in a relatively natural condition. The typical Stehekin River hydrograph is characterized by an extended spring freshet (May–July) and intense, short-lived, late-fall discharge events (Fig. 2). Average discharge of the Stehekin River is 40 m3 s−1, with peak flood discharges up to an order of magnitude greater. While numerous small streams enter Lake Chelan, the Stehekin River is the dominant source of water and sediment to the northern part of the lake.
Sedimentological and geophysical data were collected during June 2010 in the northern five kilometers of Lake Chelan near the Stehekin River delta. Kasten-style gravity cores (Kuehl et al. 1985) and Shipek surface grab samples were collected in water depths greater than 50 m (Fig. 3) using a modular coring barge. Shallow-water grab samples were collected from a small aluminum skiff. All geophysical data were collected using the R/V Lake Itasca (University of Texas Institute for Geophysics).
Multibeam bathymetric data were acquired using a 512-beam Reson SeaBat 7101 system mounted to the R/V Lake Itasca. These data were processed using CARIS software to produce a bathymetric surface (Fig. 4). Lake Chelan sits in a narrow and steep-sided basin in which GPS satellite reception is poor, and some associated artificial noise remains in the final bathymetric surface, such as NW–SE lineation in deeper water, and NE–SW lineation on the delta top (Fig. 4). Fifty kilometers of subbottom CHIRP (Compressed High Intensity Radar Pulse) profiles were collected using an EdgeTech SB216-S towfish operating with a pulse-frequency range of 2–16 kHz. Penetration and resolution varied throughout the survey area due to differences in sediment types and particularly the amount of biogenic gas in the sediment, which severely impeded imaging in the deepest parts of the lake.
Kasten cores were opened and processed in the field. A 2-cm-thick slab was removed along the length of each core, and the slabs were X-rayed in the laboratory using a digital X-ray system. The remainder of each core was cut into 2-cm intervals and bagged for radiochemical and grain-size analyses. Grain-size distributions of the mud fraction from each sample (<64 μm) were determined using Micromeritics Sedigraph 5100 and 5120 particle size analyzers. Grain-size distributions of the sand fraction from each sample were determined using a two-meter-tall automated settling column.
Sediment accumulation rates were determined using the procedure outlined by Nittrouer et al. (1979) in which 210Pb activities are determined by measuring the activity of its granddaughter, 210Po, relative to a calibrated spike of 209Po. 210Pb has a half-life of 22.3 years, and can be used to constrain sediment accumulation over the past century. Sediment accumulation rates were calculated by solving:
where S is the sediment accumulation rate λ, is the decay constant for 210Pb, A0 and Az are excess 210Pb activities at two points within the region of log-linear decay, and z is the difference in depth between Ao and Az.
Delta and Basin Morphology
The bathymetric survey reveals a narrow steep-sided basin that is being filled symmetrically by the prograding Stehekin River delta. Two channels cross the delta topset, each of which is connected to the river at its landward extent (Fig. 3). The primary (western) channel is wider and deeper than the secondary channel, and appears to carry much of the discharge during flood events (Fig. 1). The break in slope between the delta topset and foreset (i.e., the rollover) was found at a depth of ∼ 12 m at the time of this study. The upper 200 m of the foreset (measured along dip) is relatively steep (∼ 7°), and the lower foreset averages 3° (Fig. 5A). The glacially scoured sidewalls of the lake are universally steep, and in the southern end of the study area exceed 45°. Multiple small streams enter the lake from its east and west sides, but the contribution of sediment from these sources to the study area is negligible. Although the sidewalls and delta foreset are locally steep, there is evidence of only one slump feature, on the west margin of the lake, which has a surface area ∼ 15,000 m2.
Bedforms originate on the steep part of the upper foreset near the slope break and extend 1300 m downlake into water depths > 115 m (Figs. 5A, 6). The bedforms have sinuous crests and are composed of well-sorted fine sand (D50 = 2.8 φ; 150 μm). The most proximal bedforms have an amplitude of ∼ 2 m and a wavelength of ∼ 20 m, with bedform wavelength increasing toward the distal limit of the bedform field, as determined from the CHIRP line in Figures 5A and 6. These bedforms occur as a bathymetric high relative to the surrounding lake bottom. This is in contrast to bedforms found in many other natural settings (e.g., rivers, continental slopes), where bedforms occur within obvious channel bathymetry. There is no evidence of local channelization or levee topography associated with the bedform field in Lake Chelan (Fig. 5B).
Sediment on the delta foreset is divided into four grain-size populations on the basis of position on the foreset (Fig. 7). Samples collected along a 900-m transect down the axis of the bedforms are well sorted, with a mean diameter (D50) of 2.8 φ (150 μm). This stands in contrast to sediment collected ∼ 200 m off axis of the proximal limit of the bedform field, which is much finer (D50 = 6.1 φ; 14 μm). The off-axis sediment has a distribution of grain sizes similar to sediments collected beyond the distal limit of the bedforms. The bedforms are not apparent in either the multibeam or CHIRP datasets beyond ∼ 1300 m down the foreset. The three samples collected downlake of the bedforms become finer toward the distal end of the system from D50 = 5.0 φ (31 μm) to D50 = 6.6 φ (10 μm). Cores from intermediate positions on the foreset (KC2, KC3, and KC4) (Fig. 4) show little variability within the finer-grained sediment below 20 cm. The most-proximal and most-distal cores (KC6 and KC1, respectively) show considerably more variability in grain size below 10–20 cm (Fig. 8).
Sediment Accumulation Patterns
Sediment accumulation rates determined via 210Pb analysis of the kasten cores are generally greatest in the proximal part of the basin (KC2: 0.50 cm y–1) and diminish with increasing distance from the Stehekin River mouth (KC1: 0.34 cm y–1) (Fig. 9). Sediment accumulation rates from KC2 and KC4, both of which were collected at a similar distance from the delta rollover, show a decrease in the rate of sediment accumulation toward the lateral margin of the lake. The most proximal core, KC6, does not show the highest accumulation rate, but similar to KC4, it was collected more toward the lateral margin of the lake. Plots of excess 210Pb from all cores exhibit steady-state log-linear decay (R2 = 0.76–0.97) from the bottom of each core to within 10–20 cm of the surface. Excess 210Pb activities in the upper 10–20 cm of each core are uniform and low relative to activities directly below this interval.
The most noticeable feature on the foreset of the Stehekin River delta is the elongate field of bedforms that runs from the uppermost foreset ∼1300 m along the lake floor to a depth exceeding 100 m. We hypothesize that these bedforms are the product of hyperpycnal flows originating from the Stehekin River. In the following discussion, we characterize the type of density-driven flows acting in this system, classify the bedforms, and relate flow conditions to both the upstream fluvial system and the sedimentary signals preserved in lake-bed sediment.
The bedforms on the foreset of the Stehekin River delta occur between ∼ 60 m and ∼ 115 m water depth. At these depths no surface-wave energy reaches the lake bed, and the lacustrine setting precludes any powerful currents as would be common in the marine environment. Consequently, the bedforms must be formed by bottom-hugging flows moving down the foreset of the Stehekin River delta. Because of the steep slope of the foreset, any such flow will be Froude-supercritical. The dependence of the densimetric Froude number on slope is evident by substituting an equation for the velocity of a density current into the densimetric Froude number equation (Eq. 1), for example the modified version of the Chézy Equation described by Kneller and Buckee (2000):
where is density-current velocity, g is gravity, S is slope, D is flow thickness, and fb and fi are Darcy-Weisbach friction coefficients at the top and bottom interfaces of the flow, respectively. Substituting Equation 3 into Equation 1 and simplifying, yields the densimetric Froude number (Frd):
Middleton and Southard (1984) suggest using a value of 0.01 for the summed Darcy-Weisbach friction coefficients (fb + fi) for large turbidity currents. This modified formulation of the densimetric Froude number is also described by Hand (1974), who suggests that the Darcy-Weisbach friction coefficients usually sum to ∼ 0.057 based on experimental results. Using suggested Darcy-Weisbach friction coefficients from Middleton and Southard (1984) and Hand (1974) as endmembers, the critical slope needed to produce a Froude-supercritical flow is between 0.00125 (0.07°) and 0.007 (0.40°), respectively. Because the slope of the bedform field on the Stehekin River delta is generally > 10 times steeper than such critical slopes, bottom-hugging flows will be Froude-supercritical in this system.
Antidunes versus Cyclic Steps
Antidunes and cyclic steps are the bedforms produced under Froude-supercritical flow conditions (e.g., Kennedy 1963; Hand 1974; Taki and Parker 2005; Sun and Parker 2005; Kostic et al. 2010; Cartigny 2012). Antidunes and cyclic steps share a number of attributes, and distinguishing between the bedforms is challenging. The stratification geometry of the two types is distinct, but we are unable to resolve sufficient detail in our CHIRP data to distinguish between them (Fig. 5). However, it is possible to determine whether antidunes or cyclic steps could form in this setting by constraining the flow velocities associated with each type of bedform. The flow speed over antidunes is equal to the celerity of a wave of the same wavelength as the antidunes (Hand 1974). In this case, the interface between the flow and the overlying ambient water is in phase with the bed (i.e., the upper surface of the density current is in phase with the undulations of the antidunes). This wave at the upper interface of the flow, and the antidunes below it, are held nearly stationary because the interfacial wave propagates upstream at approximately the same speed as the flow moves downstream. Flow velocity is therefore related to antidune wavelength by
where is the mean current velocity, g is gravitational acceleration, L is the wavelength of bedform, Δρ is the density contrast across the interface, ρ is the density of the flow, and ρ′ is the density of the ambient fluid (Hand 1974). A range of solutions to Equation 5 are calculated using a bedform wavelength of 25 m (mean of the 40 bedforms measured from CHIRP data) (Fig. 10). Such flow velocities are too slow to be realistic, given the steep slope of the delta foreset, as shown by solving Equation 3 using the average slope (S) of the bedform field of 5°, a flow thickness (D) of 2.5 m, and a range of summed Darcy-Weisbach friction coefficients (fb + fi) between 0.010 and 0.057. Such solutions show that the flow velocities expected on the foreset exceed the velocities that are compatible with antidunes of the observed wavelength (Fig. 10). A flow thickness of 2.5 m was chosen based on estimates from Kostic (2011) that suggest the ratio of cyclic step wavelength to flow thickness is ∼10. The wavelength of antidunes would need to be ∼ 10 times the observed bedform wavelength in order to produce an upstream-propagating interfacial wave with a celerity sufficiently high to balance the downstream flow velocities expected to occur on the steep foreset. Based on these relationships, we posit that the bedforms on the foreset of the Stehekin Delta are cyclic steps.
Morphology of Cyclic Steps
The morphology of cyclic steps has been explored in flume experiments, where external forces common in natural environments (e.g., waves, currents, tides) are eliminated. Recent work by Spinewine et al. (2009), Kostic et al. (2010), and Cartigny (2012) has characterized the development and evolution of cyclic steps in laboratory experiments. Fortunately, in the relatively quiescent environment of Lake Chelan, such external forces are also negligible, allowing the attribution of sedimentary signatures to hyperpycnal flows, which permits straightforward comparison with these experimental studies.
Spinewine et al. (2009) report an upward-concave profile in the experimental slope on which cyclic steps are observed. This upward-concave profile is also a characteristic of sediment wave fields according to Lee et al. (2002), which in most cases are likely cyclic steps (Spinewine et al. 2009; Kostic et al. 2010). The longitudinal profile through the bedform field on the foreset of the Stehekin delta is similarly concave upward (Fig. 5A).
The cyclic steps on the foreset of the Stehekin delta were imaged in both the bathymetric and CHIRP datasets. Because artifacts in the multibeam bathymetry interrupt any continuous line along the length of the bedform field, we have chosen to measure bedform wavelengths from a single CHIRP line (Fig. 6). Measurements from the 40 bedforms along this CHIRP line show cyclic-step wavelength increases with distance from the rollover between delta topset and foreset (Fig. 11B). For this analysis, the wavelength of each bedform is defined as the distance between adjacent troughs (Fig. 6B). The P-value of the regression is very small (P = 0.0001), indicating a significant relationship between bedform wavelength and distance from the delta rollover. This relationship is consistent with prediction, assuming that the bedforms are cyclic steps.
In the proximal part of the bedform field, both the steep gradient of the bed and the excess density of the flow promote rapid acceleration and destabilization of the flow, yielding tightly spaced hydraulic jumps. In the more distal part of the bedform field, where the gradient is decreased and mixing reduces the excess density of the flow, the acceleration and destabilization of the flow requires greater distance, yielding cyclic steps spaced farther apart. The observed relationship between local foreset gradient and bedform wavelength shows that the bedforms with the shortest wavelength are found in areas with the steepest slopes, and the bedforms with the longest wavelength occur in areas where the gradient is least steep (Fig. 11C). The observed distal increase in bedform wavelength is inconsistent with an interpretation of the bedforms as antidunes. If the bedforms were antidunes, the balance described by Equation 5 would suggest that their wavelength would be expected to decrease, rather than increase, as a result of diminished velocity and excess density toward the distal part of the bedform field.
Cyclic steps produced by density currents in the laboratory are typically built from plastic particles with a settling velocity equivalent to fine sand (e.g., Spinewine et al. 2009; Kostic et al. 2010), or from fine sand itself (e.g., Cartigny 2012). These experimental sediment types have a narrow range of grain sizes, in which the sediment particles act as bedload, and a saline solution is used as a proxy for the fine particles that would serve to increase density in a natural flow. The sediment mixture used in the experiments of Spinewine et al. (2009) has a very narrow grain-size distribution; nonetheless the authors observe downstream fining of the deposited material. The range of grain sizes in the Stehekin delta system spans three orders of magnitude from coarse sand through clay. In this natural environment, distal deposition of sand occurs primarily within the bedform field, where hyperpycnal flow velocities are sufficiently elevated to transport sand. The silt and clay fraction carried during hyperpycnal flows fines laterally away from the high-velocity bedform field, as well as distally down the axis of the lake toward the south. Such planform descriptions of cyclic steps have no counterpart in the experimental literature because of the laterally confined flumes in which cyclic steps have so far been studied. In order to understand the signatures of cyclic steps in an unconfined natural system, we must connect the processes acting within the bedform field to the fluvial source of the hyperpycnal flows.
Relating Sedimentary Signatures to Fluvial Processes
Signals from the Stehekin River.—Data from the Stehekin River are limited to measurements of water discharge just upstream of Lake Chelan (USGS gauge #12451000), and 11 measurements of suspended-sediment concentration at the same site (Nelson 1973). A suspended-sediment discharge of ∼ 5700 m3 y–1 was estimated from these water samples collected in the lowermost Stehekin River (Nelson 1973). However, the suspended-sediment rating curve from which this value was determined, only includes data from flows below ∼ 200 m3 s–1, roughly four times less than the peak discharge. Consequently, sediment discharge during flood events, which likely dominates the delivery of sediment to Lake Chelan, remains poorly constrained. A second estimate of sediment deposited in the most proximal delta over the last 9000 years reveals an average sediment discharge of ∼ 19,000 m3 y–1, although this value likely underestimates the amount of fine sediment carried in suspension and deposited in deep water (Riedel 2008). However, this value also integrates over a time period that includes extensive alpine deglaciation following the last glacial maximum, and may therefore overestimate the volume of sediment discharged in modern times.
Using the 210Pb sediment accumulation rates calculated from five cores collected on the subaqueous Stehekin delta (Fig. 9), it is possible to construct a rough sediment budget for the survey area. The sediment budget is based on a simplified pattern of sediment accumulation within the survey area. Sediment accumulation is defined to be 5 mm y–1 at the proximal edge of the delta, and to diminish linearly, reaching a value of 2 mm y–1 along the distal edge of the survey area. In most depositional systems, especially those as quiescent as Lake Chelan, sediment accumulation is greatest nearest the sediment source, and decreases distally, a pattern consistent with the bathymetry of the delta slope within the survey area (e.g., Desloges and Gilbert 1994). Because the basin is very narrow, and because CHIRP cross sections show sediment accumulation over the full basin width (excluding steep bedrock basin walls), the simplified geometry assumes equal rates of sediment accumulation across the lake. This simplified geometry suggests an average of ∼ 9000 m3 of sediment accumulates in the study area annually. The yearly accumulation calculated here falls between the two estimates of sediment discharge reported in Nelson (1973) and Riedel (2008) (5700 m3 and 19,000 m3, respectively).
Although a rating curve based on the 11 suspended-sediment-concentration values from Nelson (1973) is only a first-order approximation of the sediment-discharge behavior of the Stehekin River, it highlights the important contribution of intense, short-lived discharge events to the cumulative annual discharge of sediment (Fig. 12). This rating curve is calculated as a linear regression through log-transformed measurements of suspended-sediment concentration and water discharge:
where C is suspended-sediment concentration, Q is water discharge, and a and b are the intercept and slope of the linear regression, respectively (after Syvitski et al. 2000). This relationship can also be written in the more common power-law formulation:
Applying this rating curve to ten years of water-discharge data up to and including the study period (water years 2001–2010), illustrates the importance of short-lived flood events to the export of sediment from the Stehekin River to Lake Chelan on daily and annual timescales (Fig. 12). The contribution of these flood events can be seen both as peaks in daily sediment flux (Fig. 12B), as well as nearly stepwise increases in the cumulative sediment discharge curves (Fig. 12C). The peak in suspended-sediment export in May 2006 corresponds to the flood and hyperpycnal flow shown in Figure 1. While we are unable to say which peaks in sediment export produce hyperpycnal flows, 10 years of data (Fig. 12) include four peaks of a similar or larger magnitude than the May 2006 peak, suggesting hyperpycnal flows of this scale may occur on average every 2–3 years in this system. The duration of high-discharge events is similar between years, with each lasting 3–5 days. Even if a strongly plunging flow lasts for only a portion of each high-discharge period, such flows would be classed as sustained hyperpycnal flows as opposed to surge-type flows. The duration of surge-type flows, often the product of submarine slope failures, is usually measured in minutes (Mulder et al. 2003). This difference in duration between surge-type flows and longer-duration, or sustained flows, impacts the character of the resultant sedimentary deposits.
Signals Preserved in Sediments.—In addition to forming cyclic steps along their flow path, hyperpycnal flows originating from the Stehekin River leave a signature outside the bedform field. Profiles of excess 210Pb show a zone (∼ 10 cm thick) of constant activity at the top of each core (Fig. 9). While this closely resembles a surface mixed layer (e.g., Nittrouer et al. 1979), X-radiographs of cores show no evidence of bioturbation (Fig. 13). Such layers may record single events such as a large flood of the Stehekin River, like that which occurred in 2006. Other low-frequency events, such as landslides and temporarily increased sediment yield due to forest fires, may also produce sediment-laden discharges capable of plunging and creating the bedforms observed in the study area.
Most density-driven flows can be divided into a head, a body, and a tail. In a sustained density current, the passing of the turbulent head of the flow represents a small fraction of the total flow duration. In such a flow, the long-duration, relatively stable body of the flow dominates the sedimentary signature left by passing current. In a surge-type flow, the turbulent head of the flow represents a larger proportion of the flow duration and likely dictates the character of the resultant morphology (Lamb et al. 2004). The spatial distribution of grain sizes on the Stehekin delta indicates strong lateral shear between the flow and the ambient lake water. Approximately 1 km from the topset, hyperpycnal flows moving downslope remain sufficiently focused to deposit fine sand along the axis of the bedform field (> 80% sand, Fig. 7), and much finer sediment less than 150 m to the side (i.e., <50% sand in KC6, Fig. 8). In addition to changes in grain size, the focused nature of these flows is captured in the CHIRP data as a sharp transition between the coarser, acoustically structureless sediments within the bedform field, and finer flat-lying laminated sediments on either side (Fig. 14).
The focused character of the hyperpycnal flows on the Stehekin delta may result from their long duration, like the sustained turbidity currents described in experiments by Alexander et al. (2008). While these experimental flows were subcritical, the authors argue that sustained turbidity currents exhibit limited lateral expansion, due to the dominance of vertical turbulent momentum exchange (Alexander et al. 2008). In contrast, the deposits associated with surge-type turbidity currents tend to be more fan-like, the result of predominantly lateral momentum exchange. The hydraulic jump associated with a cyclic step, in which kinetic energy is exchanged for potential energy in a turbulent vertical jump, is an efficient mechanism for vertical turbulent momentum exchange (Long et al. 1991; Kostic and Parker 2006). On the Stehekin delta, the combination of sustained flows and cyclic hydraulic jumps appears to promote strong focusing of the flow, resulting in a narrow band of bedforms, sharp lateral gradients in grain size and bedding character, and locally elevated bathymetry within the bedform field. In essence, the development of stable hydraulic jumps appears to promote vertical mixing at the expense of lateral mixing, resulting in self-confinement of these hyperpycnal flows. Perhaps this aspect of cyclic-step morphology has so far gone unnoticed because depositional cyclic steps have primarily been studied in laterally confined flumes where this behavior would not be observable (e.g., Winterwerp et al. 1992; Taki and Parker 2005; Sun and Parker 2005; Kostic and Parker 2006; Cartigny et al. 2011, 2014).
Evidence of hyperpycnal flows from the Stehekin River is preserved in a kilometer-scale train of bedforms on the foreset of the Stehekin River delta. An analysis of the characteristics of these bedforms, and the hyperpycnal fluvial discharges that created them, supports the following conclusions:
Hyperpycnal discharges of the Stehekin River are supercritical (Frd> 1). This is due to the steep slope of the foreset (∼ 3–7°), and the reduced gravity condition of two-layer flows relative to subaerial flows.
The bedforms on the foreset of the Stehekin Delta are cyclic steps. An interpretation of the bedforms as antidunes is rejected because the wave speed of the upstream-propagating interfacial wave, with which antidunes would necessarily be in phase, would be too slow to balance the high downstream velocity (due to steep bed slope) of the hyperpycnal flow. The trend of increasing bedform wavelength toward the distal end of the system is consistent with the interpretation of the bedforms as cyclic steps, and inconsistent with antidunes.
The linear nature of the bedform field, and associated sharp lateral gradients in grain size and bedding character, are promoted by the presence of hydraulic jumps and the sustained nature of the hyperpycnal flows. Hydraulic jumps serve to limit lateral momentum exchange by promoting vertical momentum exchange. Sustained flows are dominated by the stable body of the flow and minimize the signature of the turbulent head of the flow.
The cyclic steps on the Stehekin delta foreset are formed by bottom-hugging hyperpycnal discharges of the Stehekin River. These flows likely occur during short-lived high-discharge events during which sediment export from the Stehekin River is highest. During these events, the excess density of the Stehekin River discharge is sufficient to cause the river outflow to plunge and sculpt the bed into a series of cyclic steps.
Previous studies have described cyclic steps that are approximately two orders of magnitude larger than those in Lake Chelan and are associated with submarine canyons (Fildani et al. 2006; Lamb et al. 2008). The depositional cyclic steps described here bridge the gap between those large field examples and the sub-meter-scale cyclic steps characterized by laboratory experiments (Winterwerp et al. 1992; Kostic et al. 2010; Cartigny 2012). Experimental cyclic steps produced in narrow flumes are unable to the resolve the planform geometry of the sedimentary bodies produced by unconfined Froude-supercritical flows, like those in Lake Chelan. Further observations are needed to fully describe the hydrodynamic mechanisms that produce the elongate, nonchannelized bedform field observed in Lake Chelan. The flow dynamics and depositional patterns described here are likely applicable to other lacustrine systems, and to marine environments where sustained Froude-supercritical density currents are likely to occur, including the continental slope during sea-level lowstands.
This study was made possible through cooperation with the National Park Service. The Royalty Research Fund of the University of Washington provided significant funding for this project. Rip Hale, Kyle Womack, Kerrie Sampelayo, Jim Shobe, Kieran Dunne, Dan Nowacki, Katie Boldt, Kristen Webster, and Miles Logsdon provided excellent field and lab assistance. Comments from John Swenson, Paul Myrow, and two anonymous reviewers improved an earlier version of this paper. We thank the University of Texas Institute for Geophysics, who provided help with R/V Itasca logistics.
- Revision received July 18, 2014.
- Accepted March 23, 2015.