1. Introduction
Nitrogen pollution is an important environmental problem that rivers are facing, and it has become a critical objective for river ecological restoration [
1]. It was found that more than half of the nitrogen entering the river can be finally converted into N
2 emission by biochemical processes. The main biochemical nitrogen processes occur in the hyporheic zone (HZ), a saturated water area under or beside the riverbed with mixed shallow groundwater and surface water [
2], which makes the entire physical, chemical, and biological processes more complex [
3]. Research on the different influential processes of the HZ has been drawing a great deal of interest.
Transport and reactions of nitrogen or other species are mainly influenced by the distribution of some environmental factors, which are controlled by hyporheic exchanges [
4,
5]. The downwelling flow in the HZ is an important mechanism for carrying oxygen-rich surface water and pollutants into sediments and groundwater, which provides abundant dissolved oxygen and organic matter [
6]. The upwelling flow in the HZ releases pore water containing low oxygen-reducing species back into the overlying surface water [
7]. In addition, the water exchange path and residence time play an important role in the biochemical processes in the HZ [
8]. In previous research, study methods including indoor flume, field experiments, and numerical simulation have been conducted to analyze the impact of natural factors, such as riverbed morphology, riverbank morphology, and sediment distribution, on the hyporheic exchange process [
9,
10,
11,
12]. Hester et al. used the coupled model of surface water and groundwater to analyze the factors including riverbed permeability characteristics, riverbed slope, and groundwater head for the nitrate removal in the HZ [
13,
14]. Zheng et al. found that the effect of temperature variation on the daily average nitrate removal efficiency is not obvious compared with steady-temperature systems [
15,
16]. Ping et al. used COMSOL to study the impact of natural riverbed fluctuation on nitrification and denitrification [
17]. The results showed that large morphology fluctuation can accelerate the nitrogen reaction. Further, Ping et al. studied the development of the bioclogging effect on the nitrate source and sink function in HZ by a numerical model. In this research, they set a single triangular dune streambed and the ANSYSY Fluent was used to calculate the sediment–water interface pressure, and COMSOL was used for flow and reaction in porous media [
18]. Liu et al. also adopted COMSOL to build a two-dimensional numerical model with a series of dunes for evaluating the nitrogen removal rate. A new model coupling the basic biochemical reaction and genetic programming was proposed to delineate the boundary between nitrification and denitrification [
19]. Mendoza et al. analyzed the influence process of hydraulic conductivity in the hyporheic sediments on the nitrogen uptake from both the reach and hyporheic scale. It found that hydraulic conductivity is important for determining the contribution of hydrological exchange in different scales and thus influences nitrogen uptake [
20]. Pescimoro et al. discussed the hyporheic exchange in physically and chemically heterogeneous sediments using the simulation method and found that a higher volume proportion of silt is a more efficient nitrate removal. In the paper, they computed the hydraulic head distribution of idealized channel geometry for the sediment–water interface first and then adopted MODFLOW to discuss the process in the HZ [
21].
Focusing on the improvement of the river environment, more and more river ecological restoration projects have been established, which can change the natural conditions of surface water and sediment and will affect the hydrodynamics and solute migration process in the hyporheic zone [
22,
23,
24,
25]. Mutz et al. conducted an experiment to change the river channel morphology by adding wood. It was found that flow resistance can increase the vertical water flux through the riverbed by 1.8 to 2.5 times compared with structures with no wood [
26]. Sawyer et al. verified that weir structure can improve the path length of the hyporheic exchange and residence time by a flume experiment [
27]. Rana et al. constructed a sequence of weirs in the river to simulate the NaCl movement process with artificial structures, such as log dams, pebble dams, and debris dams [
28]. The results showed that weirs can result in increases in both the surface water flow and the cross-sectional area of the transient storage area. Hester and Doyle analyzed the effect of different structures on hyporheic flow and showed that channel crossing structures are more prominent than partial crossing structures [
13]. Li et al. conducted an indoor plume experiment and used a numerical model to study the NaCl solute transport in the HZ with an in-stream structure [
29]. Non-reactive solute transport depth, hyporheic vertical exchange flux, and solute flux can increase with the ratio of ground height to the underground part and structure number. Ward et al. found that hyporheic flux can be impacted by the structure design parameters [
30]. Liu and Chui built models with different weir heights to consider the impact on denitrification in the HZ. It indicated that a possible optimal weir height exists when some assumptions are met [
31]. Tewari et al. summarized the current research development of the engineered hyporheic zone and discussed the limitations of the main three methods including the field experiment, flume experiment, and numerical models [
32]. Moreover, numerical simulation has been an important study method from previous research on hydrodynamic and reaction processes in the HZ. The coupling of the surface water and flow in porous media with different governing equations is the key process. Generally, the sediment–water interface pressure is assumed as the connecting boundary, and the coupling process is computed by two simulation software or methods for surface water and groundwater, respectively, such as ANASYS Fluent and some empirical functions for surface water, MODFLOW, FEFLOW, and other software for porous media. In comparison with these separation forms, COMSOL can be used for both processes and is convenient for the coupled simulation, which has been adopted in some research [
17,
18,
19,
29,
33].
Most previous studies focused on the hydrodynamic process and nitrogen migration and transformation under natural conditions. Relatively few studies considered the influence of ecological structure design on the entire nitrogen reaction process, and the research results cannot guide practical applications well and restrict the further optimization of ecological engineering. Moreover, in our previous research [
29,
33], we only limited our scope to analyzing the hydrodynamics process and the diffusion of conserved substances with a weir structure by the flume experiment and numerical simulation. The main factors controlling the hydrodynamic process and the diffusion of conserved substances have been well understood, but the effects on the more complex processes of nitrogen migration and transformation with the in-stream weir structure have not yet been discussed, which we considered our ultimate goal for river ecological restoration design. In this paper, the same indoor flume experiment as that used in our previous research was conducted and a hydrodynamic reaction HZ model with a weir structure was considered. The basic model was verified by the real data of the flume experiment through a NaCl solute. A hypothetical nitrogen transport and reaction model considering three main reaction processes and four species was established. It aimed to answer two fundamental questions. The first was how the special weir structure impacts the nitrogen transport and transformation in the HZ and how the main structure design parameters influence the process. Second, we discussed the impacts of the homogeneous permeability characteristic in porous media on nitrogen transport and transformation in the HZ. The research can help deepen the understanding of nitrogen transport and transformation with in-stream structures and optimize the design parameters.
4. Discussion
As shown above, the weir structure can impact the hydrodynamic process and thus solute transport and reaction. Further, we analyzed some main impact factors by simulating the basic model to enhance the understanding of the structure in ecological restoration engineering in the hyporheic zone.
4.1. Effects of Structure Height above the SWI
Pressure distribution along the SWI depends on the height of the weir structure above the porous medium, which can influence the flow and reaction processes in the porous medium. Focusing on the height factors, five height cases were selected including 1, 2, 3, 4 (basic model), and 5 cm, while the other parameters remained unchanged.
As the height increased, the pressure difference at the SWI between upstream and downstream locations increased. It can be concluded that the flow velocity increased with the height in the porous medium. The flow velocity increased from 5.9 × 10−5 m/s (Case 1) to 8.93 × 10−4 m/s (Case 5). The larger heights can promote the occurrence of hyporheic exchange.
The concentration distributions of the four species (DOC, DO, NH
4+, NO
3−) in the five cases are shown in
Figure 6. As the surface water entered the hyporheic zone, the DOC solute transported and diffused downward under the five operating conditions. After the concentration reached the maximum value in
Figure 6a, it started to decrease continuously. The reason for the concentration of DOC decreasing after reaching the maximum value is that the denitrification reaction with NO
3− occurred in the anoxia zone. With the height increasing, the concentration front expanded deeper due to the larger flow velocity. The concentration distribution of DO is shown in
Figure 6b. The trends of transport and diffusion under the five operating conditions were similar to those of DOC. The diffusion depth of DO at the downstream location was smaller to that at the upstream location. With the increase in structure height, the diffusion depth of DO also increased, but the bottom of the porous medium was not reached. As the DO flow moved downward, nitrification consumed O
2. The concentration of NH
4+ was illustrated in
Figure 6c. The variation in NH
4+ was mainly caused by diffusion and nitrification. Under the condition of H = 1 cm, a ring zone appeared below the structure with a concentration value of about 0.22 mol/m
3, and the lowest concentration value was about 0.18–0.2 mol/m
3 near the right boundary. When H = 2, 3, 4, and 5 cm, with the increase in the structure height H, the location of the maximum concentration of NH
4+ migrated downward from the interface, and the NH
4+ concentration decreased from the maximum value to the minimum value of 0.25 mol/m
3. The NH
4+ front shape was similar to DO due to nitrification. As shown in
Figure 6d, the NO
3− the front became larger as the height increased. The maximum value of the NO
3− concentration varied from 0.14 to 0.16 mol/m
3 in the five cases. The concentration of NO
3− was related to DO and NH
4+ distributions.
The reaction rate of net denitrification is demonstrated in
Figure 7. Nitrification was dominant in the aerobic region. Because NH
4+ and DO were abundant at the SWI, the nitrification reaction rate had the maximum value near it. With the consumption of the two reactants, the nitrification reaction rate decreased with depth. The denitrification reaction was dominant in the anoxic zone, so the denitrification rate peaked below the aerobic–anoxic boundary and then decreased to 0 mol/(m
3·s) with depth, forming a narrow but clear denitrification zone. In addition, with the increase in height, the depth of the nitrification zone increased from 0.02 m when H = 1 cm to 0.22 m when H = 5 cm, and the depth of the denitrification zone increased from 0.05 m to 0.28 m. At the same time, the nitrification and denitrification reaction area also increased with the increase in structure height. As shown in
Figure 8, the spatial mean rates of nitrification and denitrification also increase with the increase in structure height. The mean rate of the nitrification reaction increased from 1.2126 × 10
−7 mol/(m
3·s) to 2.04676 × 10
−6 mol/(m
3·s), while the denitrification rate increased from 2.508 × 10
−7 mol/(m
3·s) to 3.0646 × 10
−6 mol/(m
3·s). The mean net denitrification rate increased as the structure height increased in Case 1–4. Although it decreased in Case 5, it was still larger than in Case 3. It can be seen that the increasing structure height was advantageous to the removal of nitrate in the hyporheic zone and a possible optimal height exists.
4.2. Effects of Burial Depth of the Structure
The burial depth of the structure in the porous medium can impact the flow and solute distribution. Five burial depths in the porous medium, 4, 8, 12, 16, and 20 cm, were considered. The height above the SWI was the same as in the basic model and the other parameters remained unchanged.
With the increase in the burial depth of the structure, the maximum value of fluid velocity in the porous medium decreased from 2.47 × 10−4 to 1.3 × 10−4 m/s. The decrease in velocity was caused by the larger blocking effect of the increased burial depth of the structure in the porous medium.
The concentration distributions of DOC for different burial depths are shown in
Figure 9a. The difference in concentration distributions in different cases was not obvious compared to the cases with different structure heights. In the five cases, the concentration front extended to the bottom of the porous medium. The concentration distributions of DO are shown in
Figure 9b. When the burial depth exceeded 12 cm, the concentration diffusion of the DO solute in the porous medium can only occur at the upstream location of the structure. In addition, for the depths of 4, 8, 12, and 16 cm, the diffusion depth of the DO solute was located at 17 cm. However, when the burial depth was 20 cm, the structure depth exceeded the DO solute diffusion maximum depth, and the diffusion depth of the DO solute became shallower and dropped to 13 cm. The concentration distributions of NH
4+ and NO
3− are shown in
Figure 9c,d, respectively. The pattern of variation was similar to that of DO, except for the extension area. With the increase in structure depth, especially when it exceeded the solute peaks of NH
4+ and NO
3−, the solute diffusion only occurred in the upstream region of the structure and the diffusion depth became shallow, while the solute could not diffuse downward in the downstream region. The reason for the changes in the concentrations of the DO, NH
4+, and NO
3− solutes in the hyporheic zone is that the streamlined distribution in the porous medium was changed with the increase in the burial depth of the structure, which led to the decrease in the concentration value near the structure and the smaller maximum diffusion depth.
The reaction rates of net denitrification are illustrated in
Figure 10. The nitrification reaction area began to change at a burial depth of 12 cm due to the blocking by the structure and the nitrification reaction area in the downstream area becoming smaller. For the depths of 16 and 20 cm, the nitrification reaction area occurred only in the upstream area of the structure, and only a small number of reactions occurred in the downstream area at the surface water–sediment interface. The denitrification reaction area began to change at a burial depth of 16 cm. With the further increase in the burial depth of the structure, when the depth was 20 cm, the structure passed through the reaction area that occurred around the structure. The denitrification area became smaller compared to the other four cases.
Figure 11 shows that the spatial mean reaction rate of nitrification decreased with the increase in the burial depth of the structure, from 8.4938 × 10
−7 to 5.7553 × 10
−7 mol/(m
3·s). The denitrification reaction rate increased from 1.9174 × 10
−6 mol/(m
3·s) to 2.1769 × 10
−6 mol/(m
3·s) with the increase in burial depth. The mean net denitrification rate increased with the structure burial depth increasing, except at a depth of 20 cm. It can be concluded that a larger structure burial depth is beneficial to the removal of nitrate in the hyporheic zone, and a possible optimal depth exists.
4.3. Effects of Permeability Characteristics in Porous Media
Permeability characteristics are an important factor for the flow process in porous media, which results in nitrogen transport and transformation. In this section, we focus on the influence of permeability characteristics on the hyporheic zone with a weir structure, which can help guide in selecting the structure located on the target river channel. To this end, different homogeneous porous media are studied.
The effects of different homogeneous permeabilities on nitrogen transport and transformation are studied based on the basic conceptual model. The four permeability coefficient cases are described in
Table 5.
Figure 12 shows the concentration distributions of the four species for different homogeneous permeability coefficients. With the permeability coefficient increasing, the diffusion depths of DOC, DO, NH
4+, and NO
3− in the porous media also increased. The reason was that a larger flow velocity in the porous medium was obtained with better permeability characteristics when the pressure difference on the SWI was unchanged in the four cases. The maximum velocity can increase from 1.82 × 10
−5 m/s to 2.22 × 10
−4 m/s. As the DO and NH
4+ fronts moved downward, the nitrification areas extended deeper, which enlarged the NO
3− concentration area and increased the peak concentration.
Figure 13 and
Table 6 showed the distributions of the reaction rates of net denitrification and the mean spatial reaction rates of nitrification, denitrification, and net denitrification in different cases. With the permeability coefficient increasing, both the nitrification and denitrification reaction areas became larger. The mean spatial reaction rates of nitrification and denitrification increased from 1.8141 × 10
−7 mol/(m
3·s) to 8.3481 × 10
−7 mol/(m
3·s) and 4.0259 × 10
−7 mol/(m
3·s) to 19.0241 × 10
−7 mol/(m
3·s), respectively. Better permeability characteristics can promote nitrification and denitrification reactions. However, the mean spatial reaction rate of net denitrification decreases from −2.2118 × 10
−7 mol/(m
3·s) to −10.676 × 10
−7 mol/(m
3·s), indicating that the denitrification reaction is stronger with the permeability coefficient increasing. It can be concluded that the zone with a larger homogeneous permeability coefficient has the advantage of reducing nitrogen pollution and serves as a better location for the ecological weir structure.
5. Conclusions
In this research, nitrogen transport and reaction in the hyporheic zone with an ecological weir structure were analyzed. A non-reactive indoor flume experiment was conducted. The coupled model with the SWI pressure boundary was built, simulated with COMSOL Multiphysics, and validated by experimental data. The hypothetical nitrogen reaction involving three main reaction equations was studied and the important factors of the structure were discussed. The following conclusions can be obtained.
The model coupling surface water and flow in the porous media by the SWI pressure boundary is an appropriate method to describe the flow in the HZ and was verified by the flume experiment. Considering the main nitrogen reactions, the convective diffusion can decline the four species’ concentrations with the depth increasing. The DO and NH4+ were the main control boundary for nitrification and denitrification. The height of the weir structure above the sediment changed the pressure distribution at the SWI and influenced the velocity field. With the increasing height, the velocity, both in the overlying water and the porous media, can increase, which makes the exchange occur more quickly. The solutes can diffuse much wider and deeper, which results in influencing the entire nitrogen transport and reaction processes. In most cases, increasing the height can promote nitrification and denitrification. The reaction area and concentration front can move deeper with a larger flow velocity. A larger height can better reduce nitrogen pollution. In addition, larger burial depths below the sediment can be also beneficial for nitrogen pollution elimination. The reduced velocity is caused by the larger blocking effect due to the increase in the burial depth of the structure in porous media. However, the rates of nitrification and denitrification are larger, and the net nitrogen elimination effect is better. For future practical engineering, the findings can help optimize the structure design.
Moreover, considering the homogeneous sediment with different permeability coefficients, larger permeability coefficients can accelerate the flow exchange, which results in the species breaking through deeper. DO and NH4+ can move and diffuse deeper and enhance both the nitrification and denitrification reactions. With the permeability increasing, the effect of nitrogen elimination was promoted. It indicated that the zones with better permeability characteristics are better suited for the weir structure locations.
Furthermore, a few potential research questions related to both the indoor experiment and the coupled model remain to be investigated. First, the scale of the indoor flume may limit the flow and the no-flux boundary can impact the more realistic flow. In the next step, experiments of larger scales or cycle boundaries can be conducted, and realistic nitrogen reactions should be considered to validate the simulation model. Second, more detailed reactions should be considered and simulated in the porous medium. Third, the coupled surface–groundwater model should be improved with more complex ecological structures.