Influences of riverbed siltation on redox zonation during bank filtration: a case study of Liao River, Northeast China

The upper part of riverbed sediment is one of the key interfaces between surface water and groundwater, and biogeochemical process in this interface has a profound influence on the chemistry of infiltrated water. The lithology and permeability of bed sediment is mainly controlled by variation in river hydrodynamic conditions. However, there have been few studies of the effect of riverbed siltation on the hydrochemistry and redox reactions of infiltrated water due to the high variability in these processes and challenges associated with sampling. This study selected and examined a river channel near a site of riverbank filtration by drilling on the floating platform and conducting microelectrode testing and high-resolution sampling. The hydrodynamic and chemical characteristics of pore water in and lithologic characteristics of riverbed sediment, the siltation, and redox zone were examined and compared. Differences in hydrodynamic conditions changed the lithology of riverbed sediment, consequently affecting redox reactions during the process of river water infiltration. Variations in siltation changed the residence time of pore water and organic matter content, which ultimately resulted in differences in extension range and intensity of redox reactions. This study provides a valuable reference for understanding the effect of riverbed siltation on water quality of riverbank infiltration.


INTRODUCTION
Surface water infiltration is an important category of surface water-groundwater interaction, and in particular, river bank filtration (RBF) increases groundwater recharge. RBF is widely used around the world as a sustainable source of drinking water (Ray et al. ; Tufenkji et al. ), as this technology reduces environmental problems resulting from overexploitation of water resources and alleviates water supply crises in some areas (Ray et al. ; Hu et al. ). RBF not only changes the hydraulic connection between the river and groundwater, but also dilutes the concentrations of pollutants as a large proportion of dissolved and suspended contaminants are removed from water during the infiltration process. The redox environment of oxygen-carrying surface water, which is believed to promote redox reactions and the degradation and removal of pollutants (Romero-Esquivel et al. ; Guo et al. ).
Developing appropriate management to realizing the full benefits of RBF requires an understanding of the evolution of hydrogeochemistry during riverbank infiltration (Henzler et al. ).
The riverbed sediment zone constitutes the uppermost zone during water infiltration, and this zone shows the greatest variations in physical, chemical, and biological gradients between river water and groundwater. These characteristics of the river sediment zone make it a key interface within surface water-groundwater interactions (Sophocleous ; McLachlan et al. ). The biogeochemistry of riverbed sediment plays a vital role in maintaining groundwater quality and ecological security (Tufenkji et al. ). During surface water infiltration, microbial fermentation, aerobic respiration, denitrification, iron-manganese reduction, sulfate reduction, and other reaction processes act to degrade bioavailable organic matter in riverbed sediment, resulting in sequential redox zonation (Yuan ). The permeability of sediment to water infiltration is influenced by physical, chemical, and biological factors, such as flow conditions, dissolved organic matter, and the activity of microorganisms (Brunke & Gonser ). Siltation not only reduces riverbed sediment permeability and recharge intensity, but also changes the hydrodynamic conditions of bank infiltration, drives changes to environmental conditions at the sediment-water interface, and affects the evolution of The study area of the current study was a riverside well field in Shenyang, northeast China, in which groundwater quality is characterized to be of poor quality with high con- The Liao River is a highly seasonal river, with low flow periods characterized by siltation of the channel, whereas incoming water flow and sediment concentration increase greatly during the flood period, resulting in an increase in the sediment transport capacity of the channel, channel scouring, and over-bank siltation. Water flow gradually reduces subsequent to the flood, with a concurrent decrease in the ability of water to carry sediment, resulting in a return to channel siltation conditions. Previous studies have shown that rivers under the influence of upstream runoff and a downstream reservoir show obvious seasonal scouring and a decrease in the permeability of bed sediment, resulting in a variability in bed sediment depth of as much as 10 cm (Hu et al. ; Huang ). Porewater concentrations of Fe, Mn, and As far exceed drinking water standards, and changes in concentrations of these elements in porewater are related to the siltation and scouring of bed sediment.
The objectives of the present study were to: (1) analyze the spatial structure characteristics of the riverbed sediment of the Liao River; (2) identify differences in infiltration permeability of sediment within a transverse segment of the river; and (3) analyze the spatial distribution of pore water chemical components and identify variation in redox zoning of scoured and silted riverbed sediment.

Study area
The study area of the present study is located 40 km north of The chemistry of shallow groundwater of the study area is characterized as a Ca-Mg-HCO 3 or Ca-HCO 3 type, with iron and manganese concentrations of 2,809-4,261 mg kg À1 and 1,401-1,805 mg kg À1 , respectively. During bank infiltration, organic matter acts as the main electron donor, and the order of acceptors in terms of the degree of reactability is: O 2 , NO À 3 > Mn (IV), and Fe (III) oxides or hydroxides > SO 2À 4 . In fact, there is little difference in the horizontal extent of the redox zone between the shallow and deep flow paths, and the redox zone is controlled by the permeability of riverbed sediment and the aquifer at different depths (Su et al. ). The infiltration of river water from the Liao River into the groundwater acts as a vital water resource that is characterized by stable and good water quantity and quality.

Sampling and analysis
The riverbed section labeled 'I-I' (see Figure 1(c)) across the Liao River was selected as the control monitoring section in the present study as this section is consistent with the direction of the river infiltration and passes through the center of the groundwater drawdown funnel. Riverbed sediment and pore water samples were collected in September 2019 from the sampling locations shown in Figure 1(c).
For the collection of sediment samples, a mobile drilling platform was established over the Liao River and a portable high-frequency vibration drill (Wink S5 Vibracore, Canada) was used to collect sediment samples from a depth 5 m below the riverbed (see Table 1). The collected undisturbed sediment was packed in polyvinyl chloride (PVC) pipe, which was cut into 0.5 m segments. The segments were sealed with sample plugs and sealing film and then placed in an air-tight box for storage in refrigerator at a low temperature. The samples were transported to the laboratory as soon as possible for analysis and testing (Figure 4).  River water samples were collected in sampling tubes through siphoning, which ensured that the surface layer of the sediment remained undisturbed. The physico-chemical parameters of each sample were measured, after which the sample was transferred to a clean polyethylene bottle, sealed with a protective agent according to the test requirements, stored at low temperature (4 C) and transported back to the laboratory for analysis and testing as soon as possible.
Pore water samples were collected from the upper 0-20 cm layer of sediment using an HR-Peeper sampler

Hydrodynamic conditions
Hydrodynamic conditions of a river segment are driven by drops in the height of the river upstream and downstream,  Sampling points are shown in Figure 1 There was an obvious correlation between water depth and hydrodynamic condition, with the depth of water being positively related to the hydrodynamic intensity of the river. The riverbed section could be divided according to velocity and water depth into three zones of hydrodynamic intensity: (1) weak; (2) medium; and (3) strong (see Table 2).

Lithologic distribution of riverbed sediment
Riverbed sediment at different points showed stratification according to various characteristics, such as color and particle size ( Figure 6). The lithology of the section could be described as mainly silty sand and fine sand, with silty soil and medium sand appearing at some points.
There were differences in the lithology of riverbed sediment among the different hydrodynamic zones, which was manifested as the thickness of the silt layer and average sediment particle size. The thickness of the silt layer was greater in the weak hydrodynamic zone compared with that in the medium hydrodynamic zone, while the silt layer in the strong hydrodynamic zone remained undeveloped. The rank of the zones according to the overall average particle size of sediment was: weak hydrodynamic zone < medium hydrodynamic zone < strong hydrodynamic zone. Therefore, there was a positive correlation between hydrodynamic condition and average particle size, which was consistent with the degree of riverbed scouring and clogging.

Environmental indices
The DO of overlying water was 5.0-5.5 mg L À1 , whereas that of pore water was reduced to <1 mg L À1 at a sediment depth of 5 mm. There were differences in the depths at which DO reached the detection limit among the different siltation conditions (Figure 7(a)), with the higher the degree of siltation, the shallower the depth at which DO reached the detection limit. Strong siltation resulted in an increased pore water travel time and a decrease in the depth at which DO decreased below the detection limit.
The ORP of overlying water was ∼400 mV, indicating a relatively strong oxidation environment. The changes in ORP with depth were similar among different siltation zones (Figure 7(b)). Aerobic respiration occurred at a riverbed sediment depth 1 cm, and the decline in DO resulted in pore water moving from a relatively strong oxidizing environment to a weak oxidizing environment, with a rapid decrease in ORP. ORP fluctuated slightly within a depth range of 1-50 cm, whereas the redox environment remained basically stable.
The pH range of overlying water ranged between 7.5 and 8.5 with no obvious differences among the three siltation zones (Figure 7(c)). The pH of pore water remained basically stable between 7.0 and 8.0 with increasing sediment depth. There was a significant decrease in the pH of the surface layer of strong siltation zone. This observation can be explained by the longer residence time and sufficient biological respiration, which resulted in a greater production of CO 2 for release into the water and a consequent decrease in pH.

Redox reaction sensitivity indices
Monitoring of the redox sensitivity indices (O 2 , NO À 3 , Fe 2þ , Mn 2þ , SO 2À 4 , etc.) or the contents of reducing products (NH 4 þ , HS À , Fe 2þ , CH 4 , etc.) in pore water (Appelo & Postma ) can be used to identify redox zones during surface water infiltration. As shown in Figure 8, the present study constructed curves showing the changes in redox products at different depths below the riverbed as an approach to identify the redox zones, with the results showing that the index changed regularly with the infiltration distance ( Figure 8).
The overlying water of the control section showed a high concentration of NO À 3 which was unevenly distributed (Figure 8(a)). The spatial variation in NO À 3 in the Liao River is due to the influence of fish and other aquatic organisms, and NO À 3 accumulates in the surface layer of riverbed sediment. The concentration of NO À 3 in pore water decreased significantly at a depth of 10 cm, with the depth of the sediment zone in which the decrease occurred different among the different siltation zones. The depths of the zones in which NO À 3 in pore water decreased significantly, which could be regarded as the positive reaction zones of nitrate reduction, were 5, 4, and 7.5 cm within the strong, medium, and weak siltation zones, respectively.
The concentration of Mn 2þ in the overlying water was generally low, ranging from 0.051 to 1.854 mg L À1 , with the highest concentrations found in the center of the riverbed (Figure 8(b)). The concentration of Mn 2þ peaked at a certain depth at points near the south bank (HC1, HC2, HC5), which could be attributed to the reduction of manganese oxide or hydroxide. However, Mn 2þ at points furthest from the south bank (HC6, HC7, HC10) remained basically unchanged with depth, with no reduction of Mn within 50 cm.
The concentration of Fe 2þ in the overlying water was low overall, with the highest Fe 2þ concentration at HC2 with a value of as much as 0.283 mg L À1 (Figure 8(c)). The concentrations of Fe 2þ appeared the peak only at some points (HC1, HC2) at a depth of 50 cm. There was no significant change in Fe 2þ content far from the south bank, with no clear reduction in Fe.
The concentrations of SO 2À 4 in the overlying water ranged from 20.347 to 56.435 mg L À1 , and were different among the different zones (Figure 8(d)). The degree of variation in SO 2À 4 with depth was greatly different among the showed that O 2 on the surface layer is the preferred electron acceptor during the initial stage of infiltration, and therefore, O 2 is the first electron acceptor to participate in the reaction.
At the point at which oxygen is consumed up to a certain threshold, denitrification occurs, and there is a rapid decline in NO À 3 concentration. Since there will generally be overlaps between O 2 and NO À 3 in the reaction area, they will concurrently act as electron acceptors in an O 2 /NO À 3 reduction zone. At a point at which the concentrations of O 2 and NO À 3 reduce to a lower value, reactions involving manganese and iron oxides or hydroxides in the riverbed sediment initiate, resulting in higher concentrations of Mn 2þ and Fe 2þ , i.e., the Mn (IV) reduction zone and the Fe (III) reduction zone. Since the reducibility of manganese exceeds that of iron, the Mn (IV) reduction zone will appear first. As infiltration continues, SO 2À 4 will also participate in the redox reaction, forming an SO 2À 4 reduction zone. The present study determined the thresholds of redox zoning (Table 3)   Within the strong siltation zone, O 2 /NO À 3 reduction, Mn (IV) reduction, Fe (III) reduction, and SO 2À 4 reduction were identified from the surface of the riverbed to a depth of 50 cm. Redox zoning was relatively dense, with some overlapping of zone points and a variety of simultaneous reduction reactions (Figure 9(a)). The strong siltation zone was characterized by fine lithology of the surface sediment, slow infiltration velocity, long residence time, and abundant organic matter, which provided enough electron donors to result in complex and abundant redox reactions.
Within the medium siltation zone, O 2 /NO À 3 reduction and Mn (IV) reduction occurred in the riverbed sediment up to a depth of 50 cm, and there was a trend of Fe (III) reduction. However, no SO 2À 4 reduction zone was evident  ( Figure 9(b)). Compared with the strong siltation zone, the silt layer of the medium siltation zone was thinner, the water infiltration rate was faster and the travel time was shorter in the middle part of the siltation zone, resulting in a decrease in organic matter content, which limited the degree of strength of redox reactions.
Within the weak siltation zone, only the O 2 /NO À 3 reduction and Mn (IV) reduction zones appeared within the riverbed sediment depth up to 50 cm (Figure 9(c)).
In contrast with the medium siltation zone, there was no complete Mn-reduction zone, indicating the presence of a large range of Mn (IV) reduction below 50 cm depth. Almost no silty soil was found on the surface layer, the overall lithology was coarse, water travel time was shorter, and there was a lower organic matter content, resulting in weaker redox reactions.
As shown in Table 4, the redox zonal ranges were ident-