Effects of Hydrodynamic Conditions on Sediment Oxygen Demand: Experimental Study Based on Three Methods

The first method conducts experiments in a continuous flow system with intact sediment core samples (Fig. 1). As a stirring device was used to agitate the overlying water, the flow velocity could be controlled by changing the rotation speed of the stirring device, but the flow pattern was artificial and dissimilar to that in the field. The flow velocity distribution in the core was previously measured in detail using a laser Doppler velocimeter to obtain the relationship between the rotation speed of the stirring device and the shear velocity, calculated by fitting measured velocity profiles to a logarithmic profile.

In this system, a downstream peristaltic pump supplies water from a feeding tank to the cores and removes it to sampling bottles at a constant discharge rate. The feeding tank, cores, peristaltic pump, DO meter (TOA, DO-25A), and sampling bottles are connected by Tygon (Courbevoie, France) tubing and polypropylene connectors. Input and output DO and nutrient concentrations are measured to evaluate the mass flux at the sediment-water interface. Most of the previous experiments with core samples have used a batch system, in which environmental conditions in the overlying water may change over time, providing a proportionally negative effect on the accuracy of the exchange rate (Miller-Way and Twilley 1996). To avoid this drawback, a continuous flow system was used to produce a steady state.

Because discharge rates are determined by the diameter of the pump tube in the continuous flow system, the hydraulic residence time can be easily controlled. As a result, DO concentrations in the cores can be regulated by changing the discharge rates. SOD is given by the following equation, assuming that the DO concentration is equal to that at the outlet under well-mixed conditions:where $V=\text{volume}$ of overlying water in the cores; $Q=\text{discharge}$ rate; $C_{\text{in}}$ and $C_{\text{out}}=\text{DO}$ concentrations at the inlet and outlet, respectively; $A=\text{surface}$ area of the sediment; and $r_w$ is the DO consumption rate per unit volume of the overlying water. A preliminary experiment using dyes confirmed that the overlying water in cores was sufficiently homogenized within a few minutes even under the lowest rotation speed; therefore, the assumption of a well-mixed environment was considered to be correct.

Integration of Eq. (1) shows that about 95% of the volume of water in a core exchanges in 3 times the hydraulic residence time. According to Miller-Way and Twilley (1996), the time until steady state is typically less than 2 times the residence time. In a steady state, SOD can be calculated using the following equation:Intact sediment cores were collected at a central point of Lake Shinji, a shallow eutrophic lagoon in Japan, using Plexiglas cores (85 mm diameter, 50 cm height). The cores were immediately transported to the laboratory and placed in a water bath to maintain the water-temperature at $15°\text{C}$ , the temperature at the sampling site. Overlying water in the sample core was agitated using a plastic propeller positioned about 11 cm above the sediment-water interface, whose rotation speed was controlled by a speed control motor. The flow velocity in the core was controlled by changing the propeller’s rotation speed. In this case, rotation speeds were 64, 93, and 135 rpm. They were low enough to avoid resuspension of the sediment particles and resulted in mean flow velocities of 1.47, 1.82, and $3.22\text{cm}{\text{s}}^{-1}$ , respectively. In this system, “mean velocity” is the spatially averaged tangential velocity in the bulk region. The overlying water, collected near the sediment core collection site, was filtered with a glass fiber filter (Whatman, GF/C) and used as the water supply, and its DO concentration was controlled by ${\text{N}}_2$ and ${\text{O}}_2$ bubbling. For each experiment, two supply rates and three stirring conditions were chosen, which enabled us to perform six simultaneous experiments. Duplicate cores were prepared for each condition. One reference core, which was half filled with glass beads, was also used. After a 1-day-incubation, Experiment 1 was commenced. The DO concentration in the water supply was measured at 30 min intervals, and the concentration in each exiting water stream was measured at about 6 h intervals. After Experiment 1 was completed, the supply rates to cores 1–6 were changed, and Experiment 2 was commenced after one day had elapsed. The experimental conditions are listed in Table 1.

The second experimental method was a batch system that involved a gastight rectangular flume (250 cm length, 12.5 cm width, 15 cm height) in which the flow velocity could be controlled (Fig. 2). Although sediment structures are more or less disturbed, fluid motion in the field (a uniform flow, not a rotary motion) can be simulated. The middle section of the flume bed has a cavity (100 cm length, 12.5 cm width, 10 cm depth) in which the sediments taken from a eutrophic river and estuary beds are placed. In this type of flume, the sediment samples will be more or less disturbed.

Both ends of the flume are connected by pipes made of vinyl chloride (100 mm diameter), and a pump, valve, flow meter, and a water-temperature controller (ORION, LPA3) are installed between the pipes. Water in the flume is circulated through the pipes in one direction using the pump. The rotation speed of the pump and the flow velocity in the flume are regulated by a controller. In our experiment, water-temperature was also controlled ( $\pm 0.1°\text{C}$ accuracy). Water in the flume was sampled from the sampling stopcock without being exposed to the air. The flow velocity profiles had been previously measured using a laser Doppler velocimeter (TSI, Model 9710), and the relationship between the cross-sectional mean flow velocity and the shear velocity was obtained based on a logarithmic profile of velocity. The DO concentration is monitored using a DO meter (TOA, DO-25A), whose probe was installed at the downstream end of the flume.

SOD is obtained by considering the rate of DO concentration change in the overlying water using the following equation:where $V=\text{entire}$ volume of recirculating water in the system; $C=\text{DO}$ concentration in the overlying water; $A=\text{surface}$ area of the sediment; and $r_w$ is the consumption rate in the overlying water.

The sediment used in this experiment was collected from a river bed using a shovel and was immediately transported to the laboratory. The sediment contained benthic organisms such as river crabs and nereimorpha. The sediment was placed in the cavity of the flume bed and fresh water was carefully poured into the flume so as not to disturb the sediment surface. Next, the pump was started to produce a constant flow in the flume, and the water-temperature controller was set to $30°\text{C}$ . The experiment was started after as many of the benthic organisms had been removed as possible using tweezers. The DO concentration in the overlying water was monitored every hour during the experiment using a DO meter. Water sampling was carried out at the beginning and end of each experiment. The experiment was performed in the dark with a thick black curtain covering the entire flume to suppress photosynthesis generating DO. Three flow velocities, 1.47, 3.93, and $6.09\text{cm}{\text{s}}^{-1}$ , were used. After each experiment, air bubbles were passed through the overlying water in the flume to increase the DO concentration for the next experiment. The DO meter was calibrated with a 2-point-method that used the DO concentration measured by the Winkler method at the starting and finishing time of each experiment.

The third experimental method was a combination of the first two methods (Fig. 3). The method was originally designed by Sayama (1990). A cylindrical core with an inner diameter of 8 cm is screwed beneath the floor of the rectangular main flume (19 cm length, 12 cm width, 5 cm height), and the sediment surface is adjusted to the level of the flume surface. Water in the main flume is recirculated, and its flow velocity can be controlled by a variable pump. This system is advantageous because it can simulate natural flow conditions and use intact sediment cores. Before the experiment, the relationship between the vertical velocity profile and the rotation speed of the variable pump was obtained using a laser Doppler velocimeter (TSI, Model 9710). In addition, the relationship between the rotation speed and shear velocity was also obtained by fitting measured velocity profiles to a logarithmic profile. The feeding water inlet and the overlying water outlet are mounted on the main flume, making continuous flow experiments possible.

In this experiment, precise distributions of the DO concentration were measured in the vicinity of the sediment surface using a DO microelectrode. In the 1980s, a DO microelectrode with a high spatial resolution was developed, enabling measurement of precise DO profiles near the sediment-water interface and SOD estimation from concentration gradients (e.g., Jørgensen and Revsbech 1985). The DO microelectrode makes it possible to observe directly the effects of flow velocity on the DO profile and SOD. The glass-coated electrode used in the measurement has a platinum tip with a $5\mu \text{m}$ diameter and a very high spatial resolution. In this microsensor, the cathode is situated behind an electrically insulating membrane of silicone rubber, which is extremely permeable to oxygen (Jørgensen and Revsbech 1985). The platinum wire is covered with an inner glass pipe excluding the tip and is bathed in an electrolyte solution of KCl into which a Ag/AgCl reference electrode is immersed. The electrode is attached to a micromanipulator whose precision is $1\mu \text{m}$ , so that measurement of detailed DO profiles near the sediment-water interface is possible. A control device that automatically operates the micromanipulator enables movement of the microelectrode at different time and vertical space intervals. The micromanipulator is also attached to another traverse device that can move the microelectrode horizontally with a precision of $100\mu \text{m}$ . These two traverse devices make it possible to precisely reach any measurement point. The output from the microelectrode signals is amplified by a picoammeter and recorded.

Intact sediment cores were collected with Plexiglas cores (8 cm diameter, 30 cm height) from Lake Shinji. Water-temperature and DO concentration at the sampling location were measured. An oxic brown layer was observed at the top 2 or 3 mm of the sediment, and an anoxic layer was observed underneath. Sediment cores were immediately transported to the laboratory and placed in the experimental device as described above. The overlying water collected at the sediment core sampling location was filtered using a glass fiber filter (Whatman, GF/C), and carefully poured into the main flume, so as not to disturb the sediment surface. The silty sediment collected from Lake Shinji had a distinct sediment-water interface, allowing easy identification of the position of the sensor tip relative to the interface using a stereoscopic microscope. Since the DO concentration is directly proportional to the electrode’s output signal, the DO electrode calibration was done in the bulk region of the overlying water and the anoxic part of the sediment (Bakker and Helder 1993; Glud et al. 1994).

Vertical DO concentration profiles near the sediment-water interface were measured for various flow velocities and locations. Before making the main measurements, we checked the horizontal DO homogeneity to find out a representative measurement point. Given our premeasurement results, we used profiles measured on the downstream side of the core, as discussed below. Moreover, since Inoue et al. (2000) predicted that SOD values would reach almost steady values within 20 min, DO profiles were measured more than 30 min after velocity changes.

The DO consumption rate per unit volume of the overlying water, $r_w$ , was measured by a quasi-BOD method at the experimental temperature. The porosity and sediment oxygen consumption rate per unit volume were also measured immediately after each experiment. Known volumes of particular layers of sediments were removed in narrow sediment cores using end-cut syringes from the sediment employed in each experiment. Some portions of the sediment were dried at $60°\text{C}$ for 2 days, and their porosities were estimated from the weight difference before and after the drying process. Other portions were used to measure the volumetric oxygen consumption rate of the sediment, $r_s$ , which was determined by the method of Hosoi et al. (1992); the DO concentration of water clouded with sediment was measured. In the following analysis, $r_s$ values obtained during the first half day were employed to avoid negative effects due to a decrease in the sediment organic-matter content after a half day. This parameter will be discussed below.

From the results of the continuous flow system, it is evident that the DO concentration decreases when the flow velocity increases (data not shown). This is probably caused by a decrease in the thickness of the diffusive boundary layer and an increase in the DO concentration gradient at the sediment-water interface as the flow velocity increases (e.g., Boudreau 2001). The relationship between SOD calculated from Eq. (2) and the rotation speed is given in Fig. 4. In Fig. 4, sediment cores are divided into two groups, namely Core 1 to Core 6 and Core 7 to Core 12, because cores in the same group experienced almost the same experimental conditions except for rotation speed. Each experiment continued for about 30 h. Since the DO concentration in each water outlet stream was measured at about 6 h intervals, we calculated SOD for 5 times. As the experiments were performed in duplicate, standard deviations were calculated from the 10 values. It can be seen that SOD increases when the rotation speed (flow velocity) increases. In addition, SOD increases with an increasing supply rate because the DO concentration in the overlying water strongly depends on the supply rate (i.e., turnover time). Therefore, the DO concentration gradient at the sediment-water interface might be less under low supply rate conditions.

In the rectangular flume system, the DO concentration decreases with time because of DO consumption in the sediment. However, the decreasing rates of the DO concentration varied depending on the flow velocities. Fig. 5 shows that high flow velocity conditions corresponded to high SOD, and that SOD showed a monotonically increasing tendency as the DO concentration increased, except at $1.47\text{cm}{\text{s}}^{-1}$ . These results qualitatively reveal the same tendency as results obtained from the continuous flow system and indicate that the limiting process for SOD in these experiments could be considered the diffusive transport of DO in the boundary layer around the sediment-water interface (e.g., Nakamura and Stefan 1994).

Fig. 6 shows vertical DO microprofiles measured in the combined system around the diffusive boundary layer with different flow velocities. Theoretically calculated DO profiles discussed in the following analysis are also presented. Flow velocities changed from 0.24 up to $10.1\text{cm}{\text{s}}^{-1}$ , where no visible sediment resuspension was observed. When the flow velocity was $10.1\text{cm}{\text{s}}^{-1}$ , however, it was visibly confirmed that some sediment particles were rolling on the sediment surface as a kind of bed load. The figure clearly shows that as flow velocity increases, the thickness of the diffusive boundary layer, $\delta$ , decreases (0.62 mm for $0.24\text{cm}{\text{s}}^{-1}$ and 0.28 mm for $10.1\text{cm}{\text{s}}^{-1}$ ), the DO concentration at the sediment-water interface increases ( $4.1\text{mg}{\text{L}}^{-1}$ for $0.24\text{cm}{\text{s}}^{-1}$ and $6.0\text{mg}{\text{L}}^{-1}$ for $10.1\text{cm}{\text{s}}^{-1}$ ), and the oxygen penetration depth increases (2.0 mm for $0.24\text{cm}{\text{s}}^{-1}$ and 2.9 mm for $10.1\text{cm}{\text{s}}^{-1}$ ). Reduction of the diffusive boundary layer thickness leads to an increased concentration gradient in the layer, which indicates an increase in SOD.

Nakamura and Stefan (1994) expressed the theoretical relationship between SOD and flow velocity under smooth surface conditions when the volumetric rate of oxygen consumption is constant as follows:where ${\text{SOD}}_{\ast }$ and $U_{\ast }=\text{nondimensional}$ SOD and flow velocity, respectively, as defined bywhere $D_s=\text{apparent}$ diffusion coefficient of DO in the sediment; $r_s=\text{DO}$ consumption rate per unit volume of the sediment; $C_O\left(z=\infty \right)=\text{DO}$ concentration in the bulk region in the overlying water; $n=\text{numerical}$ constant ( $=0.124$ , Deissler 1955); $C_f=\text{friction}$ coefficient; $\text{Sc}=\text{Schmidt}$ number; and $U=\text{mean}$ flow velocity.

Fig. 7 shows the relationship between nondimensional SOD and velocity as defined by Eq. (4). The experimental data obtained using the continuous flow system were scattered. Although the theoretical curve shows an increasing tendency of ${\text{SOD}}_{\ast }$ as $U_{\ast }$ increases, theoretical SOD values were considerably lower than the experimental values. As a large number of spiomorpha were observed in the sediment cores, the effect of their respiration and bioturbation on SOD may have been substantial (Andersen and Helder 1987; Bakker and Helder 1993). However, there was no clear surface area increase due to the existence of benthic organisms. In addition, the estimation of $U$ might be open to further discussion, because estimating $U$ from the measured data of flow velocity distribution in the core was difficult. Fig. 8 compares theoretical values with experimental values obtained from the rectangular flume system. The theoretical values qualitatively reproduced the experimental values. Fig. 8 shows a better estimation than Fig. 7. The theoretical predictions show that hydrodynamic control is significant when nondimensional velocity is less than 5. In this region, diffusive transfer in the boundary layer controls the total flux. The closer agreement between theoretical and experimental values in Fig. 8 shows that the diffusive transport model is reliable for studying the boundary layer. However, the scatter in the experimental results in Fig. 8 shows the need for further refining the oxygen kinetics model in the sediment. As shear stress in the developing boundary layer is expected to be larger than that in the fully developed boundary layer, experimental SOD values should be larger than the theoretical values. This might cause the discrepancy shown in Fig. 8.

Figs. 7 and 8 show that all SOD values calculated from the experimental results exceed the theoretical estimates, by factors of 2–3 in Fig. 7 and 1–1.5 in Fig. 8. This finding agrees with the report of Arega and Lee (2005) that SOD calculated from the DO profile by Fick’s law of diffusion was smaller than the value obtained from continuous flow and batch operations. In the continuous flow experiment, the discrepancy shown in Fig. 7 was considered to be due to the respiration of a large number of spiomorpha observed on the sediment surface (Rasmussen and Jørgensen 1992). Calculating SOD only from the concentration gradient at the sediment-water interface might be inappropriate when the macrofaunal effect is high. Another possible cause of the discrepancy was the higher effective surface area, as pointed out by Gundersen and Jørgensen (1990). Such discrepancies have been also reported for metabolism rates of coral reef algal turfs (Carpenter and Williams 2007). Precise measurements are necessary for further refinement of these results, using, for example, oxygen microelectrodes that have a high resolution in space (e.g., Jørgensen and Des Marais 1990).

DO concentration profiles are calculated using a diffusion equation. Assuming that: (1) there is a steady state; (2) the sediment surface is a fixed, smooth, and flat bed; (3) velocity and DO concentration are homogeneous except in the vertical direction; (4) the boundary layer is thin enough for its DO consumption to be negligible and shear stress in it to be approximated as a constant, according to Fick’s second law of diffusion, the diffusion equation of DO can be expressed as follows:where $D_z=\text{diffusion}$ coefficient of DO in the overlying water and $\epsilon =\text{porosity}$ . $D_z$ is expressed as follows (Deissler 1955), assuming that the turbulent Schmidt number is unity:where $\nu =\text{kinematic}$ viscosity $u_{+}$ and $z_{+}=\text{nondimensional}$ velocity and vertical coordinate, defined bywhere $u=\text{flow}$ velocity and $u_{\ast }=\text{shear}$ velocity. On the other hand, $D_s$ is expressed as follows (Berner 1980):where $D_m=\text{molecular}$ diffusion coefficient of DO; $\theta =\text{tortuosity}$ ; and $m=\text{numerical}$ constant ( $=1.8$ , Berner 1980). Eq. (7) represents the DO flux across the diffusive boundary layer, whereas Eq. (8) represents the DO flux at the sediment-water interface. Assuming that the diffusive boundary layer is thin enough for its DO consumption to be negligible, the two fluxes must coincide. Therefore, Eqs. (7) - (8) are combined by eliminating the concentration at the interface.

A comparison of the theoretical DO concentration profile with the measured profile of the present study is shown in Fig. 6. The theoretical calculations can qualitatively reproduce the decrease in diffusive boundary layer thickness and the increase of oxygen penetration depth with the increase of flow velocity. However, the estimate under the bed load condition was unsatisfactory. Some phenomenon such as pore water advection may have occurred near the sediment-water interface to cause this result.

In order to estimate SOD the concentration gradient in the diffusive boundary layer must be precisely known. A method of estimating SOD by multiplying the DO gradient and molecular diffusivity has been considered to be reliable as long as the diffusive boundary layer thickness is larger than the depth resolution. For a Schmidt number, Sc, of 500, typical for DO, the diffusive boundary layer thickness would be $\sim 2n/u_{\ast }$ . Therefore, the DO concentration should be measured at intervals of much less than $2n/u_{\ast }$ to obtain an accurate estimate of SOD. In that case, the DO concentration gradient could be reproduced accurately. However, quantitative estimation of total SOD is still difficult even without benthic fauna. One possible explanation of this is that the surface area was underestimated because the microscopic topography at the sediment surface was neglected.

As nondimensional parameters such as $U_{\ast }$ and ${\text{SOD}}_{\ast }$ are not commonly used, comparing these values with results of other papers is difficult. Therefore, we introduce and discuss the mass transfer coefficient, $K_c$ . $K_c$ is basically defined by Eq. (13)where $\delta =\text{effective}$ diffusive boundary layer thickness, obtained from extrapolating the linear gradient at the sediment-water interface to the bulk DO concentration (Jørgensen and Revsbech 1985). $K_c$ was calculated using Eq. (13) and the experimental results for the combined system, because $\delta$ could be directly measured in this system. However, $\delta$ could not be directly measured in the other two systems, so $K_c$ was calculated using Eq. (14), which was mathematically derived from Eq. (13) for the continuous flow system, the rectangular flume system, and continuous flow measurements in the combination systemwhere $C_O\left(z=0\right)=\text{DO}$ concentration at the sediment-water interface. In the process of the model formulation in Nakamura and Stefan (1992), $C_O\left(z=0\right)$ was expressed using Eq. (15)Arega and Lee (2005) showed that $K_c$ is related to the shear velocity, $u_{\ast }$ , and the Schmidt number, SC, as,where $c=\text{numerical}$ coefficient.

Fig. 9 shows the relationship between $K_c$ and $u_{\ast }{\text{Sc}}^{-2/3}$ . These $K_c$ values are of the same order as the $0.0139\text{mm}{\text{s}}^{-1}$ of Steinberger and Hondzo (1999) and the $0.002–0.013\text{mm}{\text{s}}^{-1}$ reported by Arega and Lee (2005). Fig. 9 allows us to calculate the coefficient for each system, giving $c_{\text{CFS}}=0.179$ for the continuous flow system, $c_{\text{RFS}}=0.090$ for the rectangular flume system, $c_{\text{CSE}}=0.089$ for DO microelectrode measurements in the combination system, and $c_{\text{CSC}}=0.326$ for continuous flow measurements in the combined system. These $c$ values are larger than the value of 0.0645 obtained from empirical coefficients by Nakamura and Mikogami (1994). It is worth noting that $c_{\text{CSC}}$ is much larger than $c_{\text{CSE}}$ by a factor of 3.5. Arega and Lee (2005) carried out an experiment using a cylindrical SOD chamber, and they also obtained different coefficients of $c=0.199$ for the total SOD and $c=0.106$ for the diffusive SOD. According to $c_{\text{RFS}}$ and $c_{\text{CSE}}$ in this paper and $c$ for the diffusive SOD in Arega and Lee (2005), $c$ values under a diffusion-oriented condition seem to be nearly 0.1. However, other experiments show larger values, and the differences may be caused by processes other than diffusion, such as bioturbation and/or bioirrigation. Rasheed et al. (2006) also pointed out the same tendency of release rates for inorganic nitrogen, phosphate, and silicate from the sediment. Further studies are required to clarify the reason for such a discrepancy.

The first method was used for experiments in a continuous flow system with sediment core samples. A stirring device was employed to agitate the overlying water. The advantage of this system is that it can simultaneously produce a steady state and multiple environmental conditions. As controlling water-quality conditions for each core is simple with this system, it is suitable for reproducing field experiments with multiple samples. As can be seen clearly from Eq. (1), DO concentrations in the overlying water can be controlled by changing the discharge rate. This is advantageous for studying the anaerobic release of chemicals by controlling DO and nitrate concentrations, which are responsible for the redox potential. In order to evaluate the actual mass flux in the environment, intact sediment cores should be used, but undisturbed sediment is not always necessary for fundamental studies on the effects of hydrodynamics and water-quality on fluxes at the sediment-water interface. However, the hydraulic condition in the cores is artificial.

The second method was performed with a rectangular, long flume in which the flow velocity was controlled. It is suitable for fundamental study of processes, such as hydrodynamic control of mass transfer, rather than assessment of the actual environmental flux. For instance, rectangular roughness elements can be placed on the flume floor to investigate the effects of surface roughness in a long flume in which the turbulent boundary layer is fully developed downstream. In practice, however, shorter flumes are usually used to reduce cost. Experimental results using them could be influenced by nonuniform shear stresses and shear velocity in the downstream direction. Higashino and Stefan (2005b) presented a model of SOD for a sediment bed of finite length. According to their model predictions, a sediment bed of $10,000\nu /u_{\ast }^{\prime }$ length in the downstream direction is needed to make the effect of diffusive entry length on the mean SOD negligible. $u_{\ast }^{\prime }$ is the shear velocity, calculated by Eq. (17)where $g=\text{acceleration}$ gravity; $l=\text{Manning}$ ’s roughness coefficient ( $=0.024$ for fine gravel, Arcement and Schneider 1989); and $R=\text{hydraulic}$ radius. From this point of view, the sediment bed lengths required were about 436 cm for $U=1.47\text{cm}{\text{s}}^{-1}$ , about 163 cm for $U=3.93\text{cm}{\text{s}}^{-1}$ , and about 105 cm for $U=6.09\text{cm}{\text{s}}^{-1}$ . Since the sediment bed length was 100 cm in this study, experiments under high flow velocity conditions were considered to be acceptable. However, under low flow velocity conditions, the experimental results should be treated carefully and could be improved.

The third method consisted of a combined system of the aforementioned two methods: a cylindrical, intact core sample was attached beneath the main rectangular flume. This system is advantageous because it can simulate natural flow conditions and use intact sediment cores. It also is more appropriate for studying actual exchange rates, as it can reproduce natural physicochemical processes not only in the sediment but also in the overlying water.

Consequently, the rectangular flume system and microprobe measurements in the combined system are suitable for fundamental studies of the diffusion process around the sediment-water interface, especially with respect to hydrodynamic control of diffusive transfer rate. On the other hand, the continuous flow system and the continuous flow measurement in the combined system are advantageous for measuring actual exchange rates including other processes. In practice, a system should be appropriately chosen by considering the condition of collected sediment, the number of samples, the aim of the work, and so on. For example, if one intends to measure microprofiles around the sediment-water interface, one should use a system with a microelectrode. If hydrodynamic control is the main object, we recommend using a rectangular flume. On the contrary, intact sediment cores should be prepared for biochemical assessment of mass fluxes in the field, because biochemical conditions in the surface layer of the sediment are easily disturbed, and it is quite important to preserve vertical distribution of particles and solutes. In addition, for multiple measurements, a system such as the continuous flow system introduced this paper is convenient.