A new sensor that measures the vertical profile of nearbed sediment concentration is described. The conductivity-based sensor is composed of eight electrode pairs separated in the vertical by 2.5 × 10−3 m. Electrode pairs are sampled at 16 Hz, with higher rates achievable. Each electrode pair response is linear over the range of conductivity tested from 0.2 to 0.65 mS cm−1 that exceeds the range of conductivity values corresponding to sediment–water mixtures from clear water to the packed bed limit of 0.65 m3 m−3. A laboratory test over a planar sloping beach indicates the capability of the sensor to simultaneously quantify sediment concentration profiles from roughly 0.01 m below to 0.1 m above the at-rest bed. The data indicate that the upper few millimeters of the bed are highly mobile and that bed dilation and sediment mobility vary considerably over a swash cycle.
The nearshore region of the coastal environment is highly dynamic because of breaking waves, strong currents, and high-sediment loads. Hydrodynamic models are now quite accurate in their prediction of wave shoaling and wave-forced currents (Newberger and Allen 2007). Unfortunately, the same is not true regarding the prediction of instantaneous sediment transport where researchers often report concentration predictions within a factor of 2–3 (Zedler and Street 2002). Some of this lack of predictive capability stems from the fact that sensors for measuring sediment transport are not as advanced as their hydrodynamic counterparts.
Sediment transport in the nearshore is often loosely described in terms of bedload and suspended-load components (e.g., Komar 1978). Suspended load is supported by turbulent fluid fluctuations. Bedload consists of particles that slide, roll, or hop along, and are nearly in continuous contact with, the bed. There is no exact definition for the transition between bedload and suspended load. However, one definition of bedload is sediment in motion within about 10 grain diameters of the at-rest bed (Wilson 1987). Thus, it is inherently difficult to locate the instantaneous bedload layer, owing to changing bed levels within the dynamic surf zone or even within a single swash event. Throughout this paper we will often refer to sediment transport as “nearbed,” meaning within about 0.01 m below or above the at-rest bed without specifically defining it as bedload or suspended load.
Over the last decade there have been numerous studies investigating the behavior of instantaneous suspension and suspended-sediment transport in the surf and swash zones (Conley and Beach 2003; Puleo et al. 2000). Several studies have shown the importance of turbulence and kinetic energy to the suspension signal (Butt et al. 2004; Scott et al. 2009) and demonstrated that beach type (reflective versus dissipative) has a pronounced effect on suspension within individual swashes (Miles et al. 2006). In many prior studies sediment suspension was quantified using optical backscatter sensors (Beach et al. 1992). The principle behind these sensors is that a light source is emitted into the water column, is backscattered off of suspended particles, and is received by a detector in the sensor. The amount of light backscattered is proportional to sediment concentration obtained via laboratory calibration curves. The optical approach is not appropriate in direct vicinity of the bed because optical sensors are usually linear in concentration out to only ∼0.08 m3 m−3, and the light signal can be backscattered off the bed itself. Thus, measurements are typically no closer than 0.01 m above the at-rest bed (Puleo et al. 2000) and often are only obtained within 0.03–0.04 m of the at-rest bed, omitting the bedload and nearbed suspended-load regime. For bedload with sediment concentrations up to the volumetric limit (∼0.65 m3 m−3), a different approach is needed.
In contrast to the numerous studies addressing sediment suspension, only several studies could be located that investigated instantaneous bedload concentrations under waves in a flume (Dohmen-Janssen and Hanes 2002) or oscillatory flow in a U-tube (O’Donoghue and Wright 2004a,b). To our knowledge, no study has addressed instantaneous bedload concentrations under breaking surf-zone waves. One published study was located that investigated instantaneous bedload concentrations in the swash zone (Yu et al. 1990). It contains just a few short time series on a low-energy beach but did indicate large variability in the quantity of sediment mobilized in the bedload layer. Interestingly, time-integrated sediment trap measurements suggest that bedload can be the dominant sediment transport signal in the swash zone depending on hydrodynamic conditions (Horn and Mason 1994). Thus, any test of predictive models for swash-zone sediment transport must include ground truth data of the bedload component.
Based on previous studies it is clear there is a large gap in the knowledge of mobilization and transport of nearbed sediments in the surf and swash zones. One explanation for this difficulty is inadequate technology for quantifying sediment concentrations in the direct vicinity of the bed. To overcome this inability, a new conductivity concentration profiler (CCP) that can simultaneously profile the nearbed concentration has been developed (sections 2 and 3). This paper focuses on the use of the sensor for swash-zone applications, but it is not necessarily restricted to these. Discussed in sections 4–6 are the linearity of the CCP signal, the spatial resolution, and calibration procedures, respectively. Section 7 shows the sensor response for a solitary-wave-forced swash event on a planar-sloping laboratory beach. Section 8 discusses sensor considerations, and section 9 concludes.
2. Conductivity sensors
Using conductivity or resistivity (they are inverses of each other) of a sediment–water slurry to measure sediment concentration has been performed by several researchers (Dohmen-Janssen and Hanes 2002; O’Donoghue and Wright 2004a,b; Ribberink and Al-Salem 1995). Theoretically, the conductivity of the sediment–water slurry is a function of temperature, chemical composition of the sediment, and the sediment concentration. Thus, for a particular sediment type and relatively constant temperature the conductivity should be a function of sediment concentration only. Conceptually, the water is mildly conductive; saltwater is highly conductive and freshwater much less so. The addition of sediment grains as poor conductors near the conductivity (resistivity) sensor causes a conductivity drop (resistivity increase) that can be calibrated to yield a sediment concentration.
Early sensors used a profile of single electrodes inserted through the sidewall of a U-tube to measure the sediment concentration profile in an oscillatory boundary layer (Dick and Sleath 1991). Horikawa et al. (1982) employed a two-electrode pair for measuring sediment concentration at a single point. Their sensor differed from that of Dick and Sleath (1991) not only in that it used two electrodes, but that it was not fixed to a laboratory apparatus and thus could be redeployed at arbitrary horizontal and vertical locations within a flume. Presently, the most widely used conductivity sensor for sediment transport is known as the conductivity concentration meter (CCM; see, e.g., Ribberink and Al-Salem 1995). CCMs measure the sediment concentration at a single elevation using four electrodes oriented in a common plane. Electric current is supplied to the outer two electrodes, and the voltage across the inner two electrodes is measured. From calibration procedures, the measured voltage is returned as a sediment concentration. Since the CCM yields a single-point measurement, either the elevation relative to the at-rest bed must be adjusted under repeatable conditions or multiple CCMs must be deployed at different elevations at the same cross-shore location to determine a sediment-concentration profile. Standard deployment involves the introduction of the CCM from below a laboratory flume, where it can be raised or lowered using a linear actuator.
CCMs have been used to measure bedload concentrations and transport in the surf zone under nonbreaking waves (Dohmen-Janssen and Hanes 2002) and under oscillatory flows in a U-tube (O’Donoghue and Wright 2004a,b). Under nonbreaking waves it was found that the sheet flow layer varied between 10 and 60 grain diameters thick and that sediment fluxes in the sheet layer are much larger than those in the suspended-load regime. Sediment velocities for estimating flux were determined by cross correlation of the signals from two CCM sensors separated a short distance in the along-flow direction. The swash-zone study by Yu et al. (1990) also used CCMs but in a field application. Three CCM sensors were deployed from under the bed using a buried horizontal arm. A separation distance of between 0.002 and 0.003 m allowed for an estimation of the sheet flow thickness (∼0.008 m) and, what we believe, are the first measurements of instantaneous swash-zone bedload concentrations.
Our interest is a sensor to profile nearbed sediment concentrations in the swash zone. None of the previously described sensors were specifically designed for swash-zone research, even though the CCM was used by Yu et al. (1990) in the swash zone. So a new profiler has been developed. The conductivity concentration profiler (CCP) described here is based on a conductivity approach in the spirit of the Horikawa et al. (1982) sensor but for conductivity rather than resistivity. The prototype CCP is constructed of acrylic glass as a stable, nonconductive unit (Fig. 1). A small stainless steel tube is connected to the acrylic, enabling the sensor to be mounted into a deployment arm. The prototype CCP consists of eight electrode pairs composed of 2.5 × 10−4 m stainless steel wire. Each electrode pair is a conductivity sensor with the measuring portion of the electrode emerging slightly from one side of the sensor but being essentially flush with the acrylic. The use of stainless steel mitigates corrosion and is chosen as a cost-efficient alternative to platinum wires or platinization of the stainless steel. Electrode wire is shielded along its length except for the end that emerges from the acrylic. Thus, the unshielded portion is the circular cross-section of the stainless steel wire. The backside is made as slim as possible by gluing the wires in a single layer and covering them with silicone. The wires run up through the stainless steel tube and into a long section of Tygon tubing such that only the unshielded portion of the electrode is in contact with the sensing environment. The overall dimensions of the prototype CCP are 0.052 m long × 0.096 m wide × 0.066 m thick (see section 8 for discussion regarding size). Each electrode pair is separated by 2.0 × 10−3 m at the same elevation, and adjacent electrode pairs are presently separated by 2.5 × 10−3 m in the vertical.
Figure 2 shows a wiring diagram for a single electrode pair. All other electrode pairs have the same configuration. A Wavetek master oscillator provides an alternating current (ac) of Vout = 5 V at 2.7 kHz. The ac mitigates electrolysis on one of the electrodes from each sensor pair that would occur with direct current (dc). Given due diligence in calibration and usage we have found that dc is adequate, but only for several uses of a given sensor before electrolysis renders one of the electrodes from a sensor pair useless. The signal range and noise vary with the ac carrier frequencies that have ranged up to 30 kHz in previous studies of conductivity probes (Chua et al. 1986; Ncube et al. 1991; Voloudakis et al. 1999). We used a carrier frequency of 2.7 kHz, as a trade-off between dynamic range and stable signals. After generation, the ac passes through a series of 120V/16V EM57580 Zenith transformers to step up the voltage and isolate the current from each of the other electrode pairs. The voltage on the opposite side of the transformer is approximately V = 18 V and supplies the input voltage to each respective bridge circuit. Each electrode pair is connected via cable to one leg of the Wheatstone bridge consisting of four 110-kΩ resistors. The output of the bridge is connected to a full-wave rectifier consisting of four 1N4148 switching diodes and a 0.22-μF filter capacitor with a 150-kΩ resistor connected in parallel to produce a full-wave-rectified signal (essentially recovering a dc signal) before logging on a DATAQ logger in differential mode at 16 Hz. On the opposite side of the active (electrode) leg of the bridge, a 470 pF capacitor is added to approximately balance the bridge with the electrode in air. The 470 pF capacitor is used to compensate for the bridge internal capacitive imbalance due to the use of alternating current.
The conductivity is influenced by a volume around the electrode pair that is proportional to the distance between electrodes and their diameter (see section 5). When poorly conductive (highly resistive) sediment is located within the volume, the conductivity decreases, bringing the bridge closer to its balanced state and resulting in a voltage decrease in the logging system.
4. Linearity of the CCP signal with varying conductivity
The premise for the CCP is that its voltage response to variations in sediment concentration (conductivity) is linear. Linearity of the CCP was tested on two different occasions by direct comparison to a commercial conductivity meter over a range of conductivities. The CCP and conductivity meter were placed in a cylindrical stirring vessel (0.37 m high, 0.11-m diameter) with voltage readings taken at 16 Hz over 5-s durations for various fluid conductivities (Fig. 3). Mean values from the readings indicate that over the range of conductivities for sediment–water mixtures from 0.2 to 0.65 mS cm−1 the CCP response is linear. In all cases, the voltage variability over each 5-s duration was less than 2.5 × 10−3 V, and the r 2 correlation coefficient for each electrode pair ranges between 0.91and 0.99 significant at the 99% level. The linear-fit slopes vary from 2.65 to 4.66 V/(mS cm−1), and the y intercepts range from 0.26 to 0.88 V. Slope and intercept variability is likely due to slight differences in electrode pair spacing, the amount of unshielded electrode that is in contact with the mixture, and imperfections in the circuitry (e.g., the resistors are ±5% accurate). However, the fact that each electrode is calibrated independently and the fits are consistent from two different tests alleviates any difficulty with variation between the linear fits.
5. Spatial resolution
Spatial resolution is a critical parameter of the CCP in quantifying sediment concentration. On the one hand, too fine a spatial resolution is not desired because the electrode pair would essentially measure the presence or absence of only a few sand grains and the signal would be inherently noisy due to interstitial gaps and inhomogeneities in the concentration field. On the other hand, too coarse a resolution is not desired because it will overly smooth the vertical profile and will blur any sharp gradients in concentration between the at-rest bed and above fluid or between the bedload and suspended-load layer.
For a single, spherical electrode sensor, Gibson and Schwarz (1963) defined an effective cell volume (CV) as the volume of a sphere with a conductivity equal to 0.9/C∞, where C∞ is the conductivity between the probe and a sphere with an infinitely large radius. They determined this to be roughly 10 times the spherical electrode radius. For a profile of single electrodes, Dick (1989) gives a detailed approach for determining the spatial resolution. She was able to show that the vertical electrode spacing of several millimeters was adequate to resolve the sediment profile without significant interference from adjacent sensors or difficulties due to the assumption of homogeneity over the measurement range. For the CCMs, Dohmen-Janssen and Hanes (2002) suggest a vertical resolution on the order of 1.0 × 10−3 to 1.5 × 10−3 m, but the method by which this was determined is not described.
a. Theoretical estimate
Early experimental work on the dynamic response of a two-electrode conductivity probe suggested that roughly 99% of the CV is contained within a sphere equal to the electrode spacing (Gregg et al. 1981). In that case the electrode diameter was 1 × 10−4 m. Hill and Woods (1988) presented a theoretical approach in an effort to determine the spatial response of a two-electrode conductivity probe. They defined the electric potential, Vp, for two electrodes located equal distance (+a, −a) from an origin (an electrode spacing of L = 2a) as
where q1 and q2 are the electric charges, r1 and r2 are the radii from the charge locations, and k is the electrostatic constant (see Fig. 4). The current flow between the electrodes must pass through a surface normal to the plane in which the electrodes are located. As an example, if the electrodes are located on the x axis in the horizontal plane (x, y), the current must flow across the plane defined by coordinates (x = 0, y, z; shown as the gray line in the two-dimensional image in Fig. 4). Integrating the current density over this planar surface and assuming a homogeneous medium, the total current, J, passing between electrodes is [see Hill and Woods (1988) for the intermediate details]
where the charges are assumed to be of equal and opposite magnitude, q, and σ is the conductivity of the medium. To determine the effective CV the integral in (2) is decomposed into two parts (Hill and Woods 1988) as
where b represents the radial distance from the midpoint between the electrodes in the plane of the electrode pair. The ratio, Jb, between the first integral and the sum of both integrals in (3) quantifies the effective CV as
Figure 5 indicates that 55% of the expected current flow occurs within a distance of b = 1, equivalent to the electrode spacing. Ninety percent of the current flow occurs within a distance equal to 5 times the electrode spacing, equivalent to 0.01 m for the prototype CCP based on this theory. It should be noted that Hill and Woods (1988) suggest that the relevant volume is the volume within the lines of current flow that intersect the plane. Since this volume is not spherical, they conclude that the cell is primarily sensitive to conductivity fluctuations in the volume between the electrodes.
b. Experimental approach
In the previous simplified approaches for estimating effective CV, necessary assumptions included a homogeneous conducting medium and a spherical electrode. Generally these assumptions are not valid. In the case of measuring sediment concentration, the conductivity of the medium changes considerably as a function of the relative concentration of water, sediment, and air. In addition, the electrodes are not spherical but instead a circular cap of small but finite thickness (refer to the description of the sensor, wherein only the end of the electrode wire is unshielded). Thus, while the simplified theory provides one estimate of the spatial resolution, procedures were developed to estimate the resolution experimentally. Using previous experimental efforts (e.g., Hill and Woods 1988) as a guide, the sensor response was tested across an interface.
The optimal approach would be to test the spatial resolution across a sediment–water interface. It was determined that this was not practical after several attempts because either extruding the CCP from the bed or forcing the CCP into the bed at known increments caused the bed to obviously deform. Thus, the datum with which the spatial resolution was meant to be determined across did not remain fixed and rendered the results difficult to interpret. Instead, a sharp conductivity interface between fluid and air was created in the stirring vessel with the motor off using isopropyl alcohol (isopropanol) with a minimal amount of dissolved sodium chloride. Isopropyl alcohol was used to reduce surface tension effects that were encountered when water was used as the fluid since water would readily adhere to the electrode pair when it was slightly above the fluid–air interface. The CCP was lowered through the interface at 1 × 10−3 m increments using a manually controlled stepper. At each stop, the voltage was recorded for 5 s at 16 Hz (Fig. 6). The voltage scale has been normalized between 0 and 1 using the lowest and highest readings, representing air and fluid, respectively. Profiles are intentionally not offset from each other to indicate similarity in the signal for all electrode pairs. It can be seen that the electrodes maintain a normalized voltage close to zero while in air. There is an abrupt change in voltage as the electrodes move across the interface. Once in the fluid mixture, the normalized voltages are nearly constant.
How well each electrode pair resolves the interface (assumed to be of infinitesimal thickness) yields an indication of the CV in the vertical direction, the direction of interest for sediment-profiling capabilities. To determine the CV, the lowest elevation where the normalized voltage is below 0.05 (5% of the normalized range) and the highest elevation where the normalized voltage is above 0.95 (95% of the normalized range) are retained from each electrode pair. Taking the difference between these elevation values provides an estimate of the CV as 1.8 × 10−3 ± 2.5 × 10−4 m (mean ±1 standard deviation) from the eight electrode pairs. As an example, for 2.0 × 10−4 m diameter sediment typical of many beaches, this would represent 7.5–10.5 grain diameters. It is anticipated that the actual CV is less than this value because the effect of surface tension could not be completely mitigated. Even using the isopropyl alcohol, some fluid adheres to each electrode pair, causing a voltage reading higher than it should have been for a sensor that was deemed above the interface based on the experimental setup (visualize the meniscus in a cylinder for an adhering fluid). Nevertheless, this test indicates that the electrode pairs can resolve several measurements in the vicinity of the at-rest bed based on the expected CV. Also, since the voltage registered by the electrode pairs drops off with radius squared according to electromagnetic theory, sediment grains closest to the electrode pair will have a larger impact on the recorded voltage than sediment grains farther away. Finally, this test indicates that for the profiler described here, electrode pairs should not be separated by any less than approximately 1 × 10−3 m if minimal overlap (smoothing) of signals is desired. If some smoothing of the vertical profile is acceptable, electrode pairs could have a vertical separation of perhaps 5 × 10−4 m, but any closer would be physically challenging because of the diameter of the electrode itself.
6. Calibration with sediment
Calibration of conductivity or resistivity sensors for sediment concentration is not trivial. There is little reference to calibration in the peer-reviewed scientific literature. For their resistivity-based sensors, Dick (1989), Dick and Sleath (1991), and Horikawa et al. (1982) used a tank with a stirrer or forced upward jet to suspend high concentrations of sediment. At the instant a voltage measurement is recorded, a small amount of the fluid–sediment mixture is extracted to determine the sediment concentration. A calibration curve is developed after repeating this procedure for different sediment concentrations. The drawback to this approach is that the calibration for a given concentration must be obtained at precisely the same instant a voltage is recorded unless the mixture is perfectly homogeneous in space and time (not usually the case in a mixing vessel). For the CCM, two-point calibrations (sediment-free fluid and packed bed) with the assumption of linearity with concentration are normally performed (M. Dohmen-Janssen and T. O’Donoghue 2007, personal communication). We note that in all calibration attempts using conductivity-based sensors, the user should use the same sediment and liquid as will be used in the study. Differences in liquid conductivity and any variations in sediment coatings can alter the conductivity of the sediment–fluid mixture.
For the profiler described in this work, our goal was to perform calibrations in the stirring vessel described previously using a small dc motor with an attached stirrer to homogenize the mixture. The setup would require keeping sediment mobilized with essentially a homogenous concentration for concentrations nearing the packed-bed limit of roughly 0.65 m3 m−3. We attempted calibrations for theoretical sediment concentrations of 0.06, 0.14, and 0.26 m3 m−3 (Fig. 7) but realized from visual observation that the concentrations were not homogeneous as a function of space and time. To illustrate this variability, the CCP was extracted upward from the bottom of the vessel through the water column at 0.1-m intervals stopping to collect data for 5 s at 16 Hz (Fig. 7; mean values shown). For illustrative purposes, data are cast into sediment concentration using a two-point calibration rather than showing the data as raw voltage. In all cases, the individual calibrations cluster together, indicating that the individual electrode pairs measure the same trends in concentration as a function of elevation. For the 0.06 m3 m−3 case, the calibrated sediment concentrations are nearly equivalent to the theoretical concentration (assuming a homogeneous mixture) in the vessel (gray lines in Fig. 7), with little variability as a function of elevation. There is a slight curve in the profile near an elevation of 0.04 m, the elevation of the stirrer. The low sediment concentration allows for the sediment to be essentially homogeneously mixed. When the sediment concentration increases to 0.14 m3 m−3, the estimated concentration displays more variability as a function of elevation. Again the values for individual electrode pairs cluster together and are near the theoretical sediment concentration. Near the location of the stirrer, the concentration drops as sediment is mixed upward and downward due to turbulence. Near the free surface, the concentration drops to almost zero as a strong transition to essentially sediment-free water was observed. Finally, for sediment concentrations about half the packed bed limit, there is an increase over the theoretical concentration by roughly 35% due to mixer turbulence damping from the high concentration. Above the location of the mixer for most of the vessel, the concentration is nearly uniform but disagrees with the theoretical concentration by 38%. Again near the free surface a well-defined layer of low sediment concentration is observed as the turbulence is completely damped from the sediment in the water column.
The above cases are presented to highlight the inherent difficulties in calibrating the CCP or other sensors with high sediment concentrations. Other designs for the mixing vessel may have produced different results. However, we do not feel that any other simple procedure [even the type described by Dick and Sleath (1991)] would be free from imhomogeneities in the concentration field as a function of space and time. Thus, for the CCP, as with the CCM, a two-point calibration (sediment-free fluid and packed bed) is deemed appropriate, given the linearity of the sensor response as a function of conductivity (Fig. 3).
7. Laboratory swash event
A simple laboratory experiment was set up to test the response of the CCP under a single swash event driven by a solitary wave. A 0.17-m solitary wave was forced over a planar sloping beach. The inner surf and swash zones consist of a 1:12 slope composed of 0.45-mm median grain diameter sand merged with a 1:34.2 impermeable offshore slope. The sand was washed over a 0.2-mm sieve to remove fine particles. The CCP was positioned on a horizontal arm and buried so that the third electrode pair from the top of the sensor was 5.0 × 10−4 m above the at-rest bed level. The sensor is deployed from below the bed to minimize flow disturbance. The deployment position was 0.98 m landward of the still water level, and the swash excursion measured a total of 3.6 m. A buried wave gauge was deployed at the same cross-shore position as the CCP to monitor the water depth.
In preliminary tests, we noticed some sensor drift that is inherent in many two-electrode conductivity probes (Ncube et al. 1991). This sensor drift is negated by submerging the CCP for a period of time (at least 20 min) prior to collecting data, allowing the sensor to equilibrate with the water.
For the solitary-wave-driven swash event, the water depth at the sensor location rapidly increases to 0.06 m and then slowly decreases (Fig. 8a and solid black line in Fig. 8b). The slow decrease in depth at the tail end of the swash event is due to the thin sheet of water moving seaward as well as residual infiltration of fluid at this location on the beach face. Nearbed sediment concentrations were collected at 16 Hz during this event and determined from the voltages using the two-point calibration method described previously (Fig. 8b). The vertical scale in Fig. 8b is relative to the initial at-rest bed level, and the color scale indicates sediment concentration. At the beginning of the event (t = 2–2.75 s), there is a rapid response of the sediment concentration below the at-rest bed level. Unlike suspended-sediment concentration measurements under swash forcing, the response is a decrease in concentration as the packed bed dilates and sediment is carried upslope and into the lower water column. The effect of this shearing extends to approximately 6.0 × 10−3 m below the at-rest bed level. This is consistent with previous findings that the first 0.01 m of bed material is highly mobile under individual swash events (Puleo et al. 2000), but it is believed that some of this observed depth of mobilization layer is due to unavoidable scour (see section 8). During the uprush, sediment concentrations exceed 0.1 m3 m−3 at 6.0 × 10−3 m above the at-rest bed. During flow reversal (t = 3–5 s), the amount of sediment in the water column decreases, and the concentration at and below the at-rest bed level increase as the sediment settles. During backwash (t = 5–9 s), the concentrations just above the at-rest bed increase again, while those below the at-rest bed decrease as the grains are sheared. Again, sediment motion extends to roughly 6.0 × 10−3 m, with the depth of penetration increasing slightly with duration (see section 8). As the backwash water depth approaches only 2.0 × 10−3 m (t > ∼9 s), sediment rapidly settles out of the water column. In this case, the at-rest bed level accreted approximately 1.0 × 10−3 m over this single swash event.
While we believe the concept behind the CCP is robust and will provide useful nearbed sediment concentration data, there are some sensor characteristics that need to be considered. It is widely known that bubbles can affect conductivity sensors. In fact, many studies interested in quantifying void fraction use conductivity sensors, implying their sensitivity to bubbles (e.g., Vagle and Farmer 1998). In freshwater settings, such as laboratory flumes, bubbles are much less a concern due to their larger size (increasing buoyancy and reducing residence time) and the general decrease in production (Haines and Johnson 1995; Puleo et al. 2004, 2006; Slauenwhite and Johnson 1999). For the laboratory study described here, once the swash tongue reached the region of data collection, bubbles were nonexistent. For full-scale laboratory studies using freshwater it is still anticipated that bubbles will not be a concern in the swash zone but may be problematic near the break point. In saltwater environments, including the swash zone, bubbles are more readily generated and known to persist for longer periods of time (Haines and Johnson 1995; Puleo et al. 2004, 2006; Slauenwhite and Johnson 1999). Residence time and size could cause the CCP to be adversely affected. However, given that the CCP would be deployed spanning a region roughly 0.01 m below and above the at-rest bed, bubble effects would likely be minimal. We note that bubble effects on reported sediment concentrations in saltwater environments have yet to be tested with the present CCP configuration.
A second consideration of the CCP is sediment transport rate estimation. The sensor described in this work is capable of measuring the sediment concentration only. To be truly beneficial for understanding and estimating morphological variations, these values must somehow be coupled with a velocity to yield a sediment transport rate. At present there are no widely-used sensors that are specifically designed to profile sediment concentrations into the nearbed layer. Some success has been obtained using a high-frequency acoustic sensor to determine velocities in the nearbed layer (e.g., O’Donoghue and Wright 2004a), but may not be able to extend much beyond the at-rest bed level. Many sediment transport applications instead utilize single-point current meter measurements from elevations either at or near the location where concentration is measured to determine a sediment transport rate. This is not practical for the CCP because single-point current meters cannot be placed in the nearbed layer. Clearly this poses a major limitation for sediment transport studies. There are, however, some approaches that show potential. In past studies using conductivity-based sensors, the cross-correlation of sediment concentration time series between two closely-spaced sensors was used to infer the sediment velocity (Dohmen-Janssen and Hanes, 2002). The same approach may be able to be applied with the CCP, but having two profilers in close proximity may increase the likelihood for adverse scour effects (see below). Having one sensor located in the wake of another sensor may also yield incorrect concentration measurements due to flow alteration. Particle image velocimetry has been used to determine sediment velocities in the nearbed layer (Cowen et al. 2006). This technique also uses cross-correlation to estimate velocity but does so based on images of individual sediment grains rather than on times series of bulk concentration. Coupling this technique with the CCP could potentially enable sediment velocities at elevations corresponding to measured sediment concentrations to be determined.
A third consideration of the CCP is localized scour. Any sensor that is near or penetrates the bed in either a unidirectional or oscillatory flow has the potential to induce scour. The present size of the CCP was necessarily large to undergo simple testing to determine viability in measuring sediment concentrations. In addition, the cross-sectional area was nearly rectangular creating a hydrodynamically bluff body. When observing the bed only prior to and after the solitary swash cycle, we observed that there was no scour around the sensor. Indeed, smaller sensors such as the fiber optic backscatter sensor (FOBS; Beach et al. 1992) and a fiberscope Cowen et al. 2006; Dudley 2007) that have diameters on the order of 3.0 × 10−3–4.0 × 10−3 m have not been reported to cause scour. The data shown in Fig. 8 indicated sediment mobilization down to 6.0 × 10−3 m from the at-rest bed. After noting these depths from the CCP data, the same wave conditions were again run and the thickness of the mobilized layer was measured through the wave flume glass wall to be maximally 4.0 × 10−3 m throughout the swash cycle. There may be some sidewall boundary layer effects minimizing the thickness of the mobilized layer at the flume wall. These potential sidewall boundary layer effects were not measured or pursued. Because the measurements from the flume sidewall did not match those from the CCP data, repeated wave runs were undertaken where the CCP response during the swash cycle was visually observed from an overhead viewpoint. It was seen that for a portion of the uprush and a portion of the backwash a small scour depression formed around the CCP. In between these times, there was no visible scour, such that at the beginning and end of the cycle no scouring was apparent. Thus, we believe the true depths of mobilization are slightly less than those indicated in Fig. 8 and are more commensurate with the 4.0 × 10−3 m that was measured through the flume sidewall. Since the concept of a conductivity profiler has been established and verified to produce results consistent with expectations, future versions will be made much slimmer and more streamlined to reduce the adverse effects of scour. We believe a CCP with a more streamlined eye shape with a maximum diameter of 3.0 × 10−3 m is achievable if a material such as impact-resistant polycarbonate is used to minimize potential bending or breakage.
A conductivity concentration profiler (CCP) for estimating nearbed sediment concentrations has been developed. The CCP uses the same principle as past single-point sensors in that the conductivity of a sediment–water mixture is a function of the sediment concentration. Electrode pairs are shown to have a linear response over the range of concentrations from clear water to the packed sediment bed. Resolution tests indicate a cell volume (CV) on the order of 1.8 × 10−3 m. Vertical separation of adjacent electrode pairs for this CCP is 2.5 × 10−3 m but based on resolution tests could be reduced to roughly 1.0 × 10−3 m if some overlap in the signals is acceptable.
The CCP with eight electrode pairs was tested in the swash zone of a laboratory beach under solitary-wave forcing. Sediment concentrations were measured over a vertical range of roughly 0.016 m, with the lower 0.01 m being below the at-rest bed in this initial study. The ability to simultaneously measure the sediment concentration at multiple elevations quantifies mobilization depths during the swash event with mobilized sediment concentrations exceeding 0.5 m3 m−3.
Future versions of the CCP will be more streamlined and slimmer in profile, reducing potential scouring. The CCP will yield complimentary measurements to those obtained higher in the water column. For instance, when coupled with fiber-optic backscatter sensors (FOBS), swash-zone sediment concentrations from the depth of no motion through the water column (perhaps to the free surface) can be obtained. This new measurement capability should enhance our understanding of nearshore, specifically swash-zone, sediment transport by including what is the dominant transport mode under certain conditions (Horn and Mason 1994).
This work was supported by the University of Delaware and the National Science Foundation Award OCE-0845004.
Corresponding author address: Jack A. Puleo, Center for Applied Coastal Research, University of Delaware, Ocean Engineering Building, Newark, DE 19716. Email: email@example.com