Variation characteristics of water and sediment of managed aquifer recharge with the Yellow River water in the piedmont sand gravel channel in the North China Plain

The North China Plain (NCP) is a typical case of groundwater depletion due to agricultural development. Groundwater depletion has attracted the attention of many researchers (Cao et al. 2016; Zhao et al. 2019). Managed aquifer recharge (MAR) is the purposeful recharge of water in aquifer mediums for the recovery of the ecological environment (Dillon 2005). The water from the Yellow River flowing through the NCP can provide recharging water for improving the fragile groundwater environment (Rong et al. 2017). Hydrogeological zones in the NCP primarily include the piedmont region and plain area, of which the piedmont channel consists principally of sand and gravel (Wu et al. 1996; Zhao et al. 2021).

Due to the large slope of the river channel, the Yellow River water flowing through the study area mainly forms three runoff processes under the influence of gravity and its components along the riverbed, namely infiltration deep runoff (IDR), surface runoff (SR) and lateral shallow runoff (LSR), among which the IDR and LSR belong to underground runoff (Figure 1(c)). Only IDR can actually be recharged to the limestone aquifer, which is an effective recharge. Both SR and LSR eventually flow downstream, on account of the riverbed slope. However, the Yellow River water has an extraordinarily high silt content. Suspended particles deposited in riverbed medium can easily cause aquifer clogging.

Previous research has shown that the risk of porous medium clogging occurs in an aquifer when the suspended solids turbidity is more than 25 mg/L or 3 NTU (Pavelic et al. 2011; Cui et al. 2021; Rong et al. 2017; Zheng et al. 2020). During recharging, water is the carrier of suspended particles, and the seepage process can affect the suspended particle movement in the porous medium. Additionally, the process of suspended solids migration in the medium further determines the seepage process variation. The clogging type of MAR projects mainly includes physical, chemical, biological, and combined clogging (Wang et al. 2012; Cui et al. 2018, 2021; Xian et al. 2019). Notably, previous surveys have indicated that aquifer clogging caused by suspended particles is the primary clogging case (Fetzer et al. 2017; Xie et al. 2020). Physical experiments based on the one-dimensional sand column model are widely used to simulate the aquifer clogging process in groundwater recharge. Zheng further studied the movement characteristics of exogenous and endogenous particles in aquifers during recharging through sand column infiltration experiments (Zheng et al. 2020). Through a series of sand column tests, Wang studied the migration mechanism of different sized particles in porous media (Wang et al. 2012). Wang investigated the combined effects of suspended particles and bacteria on the permeability of porous media by sand column experiments (Wang et al. 2020). However, the sand column model for studying particle blockage still has limitations, especially when describing aquifer clogging in the piedmont areas with large slopes. Furthermore, laboratory experiments focused on the clogging mechanism have frequently occurred in previous articles. Studies on the variation in suspended solids and seepage flow are in short supply.

Zhao explored the mechanism of water distribution and silt clogging in the ‘open window’ river channel consisting of upper gravel and lower karst layer in the recharge of the piedmont plain (Zhao et al. 2021). According to the permeability coefficient of the medium, the double layer medium composed of upper gravel and lower karst layer belongs to the combination of ‘upper coarse and lower fine’. Owing to the existence of low permeability limestone medium, the maximum capacity of piedmont sand and gravel riverbed to distribute water and sediment has not been adequately studied (Zhao et al. 2021). On the basis of previous research, in this study, the variation characteristics of water distribution and sediment movement were studied by sand tank experiments when the lower boundary of the sand gravel layer was not constrained by the low permeability limestone medium. The conclusions of this research were an effective supplement to the operation theory of MAR projects in the Yufuhe River. Furthermore, in the piedmont region of the NCP, the riverbed surface was generally covered by a sand gravel medium.

The research results of this paper can effectively improve the existing pattern of the Yellow River water used to recharge groundwater in the NCP, especially in the piedmont area of the river channel consisting of sand gravel. Therefore, in the study, the primary study objectives were as follows: (1) to reveal the variation characteristics of water in the sand gravel channel infiltration process of the piedmont area, (2) to explore the variation characteristics of water and sediment in the piedmont channel recharge process, and (3) to simulate the influence of engineering measures such as rubber dams on water and sediment variation in the piedmont sand gravel reach infiltration.

To conform to the characteristics of the medium under actual conditions, the sand and gravel used in the sand tank model were obtained from the channel surface in the study area in the experiment. The sand samples were dried and sieved to demonstrate an average granularity (d₅₀), effective grain size (d₁₀), coefficient of permeability (k), and porosity of 8 mm, 0.2 mm, 90 m/d, and 23.9%, respectively. Tap water and suspension were used as reinjection water sources in the experiment. The turbidity of tap water was lower than 0.5 NTU, and it was regarded as a liquid without suspended particles, which is used in the water quantity experiment of the sand tank model. The suspension was used to simulate the Yellow River water. Montmorillonite is the main component of the Yellow River sediment, accounting for more than 40% of the sediment. Montmorillonite and tap water were used to prepare three suspensions with concentrations of 100, 150 and 200 NTU, respectively, for the suspended particles experiment in the sand tank.

To eliminate the interference of fine particles with particle sizes less than 0.074 mm in the original sand, the sand samples must be cleaned and dried before filling the sand tank. After completing the sand filling work, the subsequent operation steps were followed:

• (1)

  The experiments were carried out under the conditions of flow rates of 2.5, 3 and 3.5 L/min respectively. When the water pump was turned on, and tap water was injected into the sand tank from the water inlet hole, it marked the beginning of the experiment. The flow of the outlets and the head of the pressure measuring pipe were both measured every 10 minutes. The outlets flow was stabilized, and the experiment was immediately stopped.

• (2)

  A 3 cm-high plexiglass plate was set on the surface of the sand tank device to simulate the impact of the rubber dam on infiltration deep runoff (IDR), surface runoff (SR) and lateral shallow runoff (LSR) during recharging. On this basis, the operation of step (1) was repeated.

• (3)

  A two-dimensional saturated-nonsaturated steady-state flow model established by using HYDRUS-2D software was used to simulate the movement characteristics of the water flow in the sand gravel medium during the sand tank water release experiment. The numerical model simulated the water distribution process under the condition of maximum permeability of the sand gravel medium. The governing equations and the numerical model have been described in detail in previous studies (Zhao et al. 2021).

• (1)

  Reinjection water with turbidity of 100, 150, and 200 NTU was respectively injected into the sand tank, and the inlet hole flow was always controlled at 3 L/min through the pump. After the experiment had run for 160 minutes, the suspension was replaced by tap water, and the experiment continued. The flow and turbidity of all outlets were measured every 10 minutes while recording the water head value of the pressure measuring pipe. In a previous article, the measurement method of suspended particle mass in a sand tank is comprehensively described (Zhao et al. 2021).

• (2)

  A 3 cm-high plexiglass plate set on the sand tank surface was used to simulate the impact of the rubber dam built on the riverbed surface on surface runoff, underground runoff, and suspended particle movement during recharging. On this basis, the operation of step (1) was repeated.

With the existence of the rubber dam, IDR, SR, and LSR occurred concurrently. Compared with before the rubber dam was set, the value of IDR did not increase significantly (Figure 3(d)–3(f), and Table 1). But Table 1 showed that the conversion efficiency of SR to LSR was increased by approximately 25–30%, and the larger the water discharge was, the higher the conversion efficiency. This was of great significance for retaining surface water in the piedmont areas where water resources are scarce. In the piedmont channel with a large slope, the river water flowed downstream quickly. The establishment of a rubber dam can promote the conversion of high-velocity SR into low-velocity LSR, and protect ecological base flow. According to the results of the sand tank experiment, considering the impact of rubber dams on surface runoff and underground runoff, it was necessary to demonstrate repeatedly before building water retaining structures in the riverbed.

Whether the clogging process of the sand tank occurs in advance depends on the time at which the intersection of IDR and SR appears. Apparently, with an increase in the suspended particle concentration in the recharge water, the blockage time of the sand tank occurred earlier. The establishment of a rubber dam did not lead to the early occurrence of the medium clogging process (Figure 6). In the suspended particles experiment, the rubber dam can effectively increase the IDR, but due to clogging, this value was only 24–31% of water release flow of the sand tank (Table 2). Consequently, controlling the suspended particle concentration was the primary factor in improving the MAR operation efficiency.

The higher the suspended particle concentration in the reinjection water source, the more particles were deposited in the sand tank (Table 3). When the suspended particles experiment starts, many suspended particles follow the water flow and enter into the sand gravel medium. And the medium pores of the sand tank are blocked due to the deposition of suspended particles. The higher the suspended solids concentration increased in the recharge, the faster the sand gravel medium was blocked. After the experiment the distribution of the suspended particles in the sand tank was analyzed and suspended particles were mainly deposited in the surface and upper layers of the sand tank. A thin clay layer caused by suspended particle deposition was clearly observed on the sand tank surface, which is the main reason for the sharp reduction in IDR.

The existence of a rubber dam can promote the deposition of suspended particles in the sand tank (Table 3). Because of the retaining surface water effect of the rubber dam, the suspended solids were heavily intercepted, and flowed into the sand tank with water flow. After the clogging of the sand tank was formed, the scouring of the medium was simulated by the sudden removal of the rubber dam, but the clogging degree of the medium did not change significantly. After the suspended particles test, the surface sand gravel was washed and then backfilled to the sand tank model for the recharge test again, under the condition that the recharging water is tap water. And the monitored value of the IDR can almost recover to the critical value of the medium IDR.

Riverbed filtration can not only improve groundwater, but also purify water quality, which is a simple and efficient MAR type (Bertelkamp et al. 2016; Hamann et al. 2016; Cartwright & Irivine 2020; Yenehun et al. 2020). However, during the recharge, the river located in the piedmont area had completely different characteristics compared with the plain river. In this study, the permeability of the riverbed, and suspended particle turbidity had a serious influence on effective recharge of groundwater in the piedmont sand gravel channel infiltration. According to the research results of this paper, when IDR reached the critical seepage capacity of the riverbed, an excessive increase in water release would reduce the recharge efficiency. Furthermore, the suspended particles in the recharge water can also cause a greater reduction in the recharge efficiency of the gravel channel. In terms of the plugging position, it was easy to form plugging on the surface of the medium, especially physical clogging, whether it was a two-dimensional tank or one-dimensional sand column (Wang et al. 2012; Zheng et al. 2020; Zhao et al. 2021). To improve the recharge efficiency of groundwater, the rubber dam built on the riverbed surface had no obvious effect. However, it can promote the conversion of surface runoff with high speed into lateral shallow runoff with low speed, which has a positive significance for storing surface water in areas with low rainfall. Additionally, particles were carried into the riverbed by water flow, which is a potential hidden danger for the operation efficiency of recharge projects, due to the existence of the rubber dam. Therefore, before the construction of a rubber dam on riverbed surfaces, it should be fully demonstrated.

The piedmont channel in the NCP is dominated by sand gravel, which is an excellent recharge site for restoring groundwater because of its high permeability. One of the most effective ways to control groundwater overexploitation in the NCP is to recharge groundwater by the seepage of water from the Yellow River in the piedmont sand gravel channel. In this paper, a sand tank model filled with sand and gravel was designed to simulate the characteristics of water distribution and sediment migration during managed aquifer recharge by Yellow River water in the piedmont sand gravel channel of the NCP. The research results of this paper can provide a new idea for groundwater restoration in the NCP. The Yellow River is an important water source flowing through the NCP. Clogging of the recharge projects caused by suspended particles in the Yellow River water seriously affects the recharge efficiency of groundwater. The way to study aquifer clogging should not be limited to physical models, and the COMSOL Multiphysics software has great advantages in simulating the movement of water and suspended particles in porous medium. The next step is to focus on the biological clogging and physical-biological clogging mechanisms in the sand gravel infiltration of the piedmont area through numerical models.

A two-dimensional sand tank experiment was carried out to investigate the variation characteristics of water and sediment during the Yellow River water recharge in the piedmont sand gravel channel of the North China Plain. The water variation process was simulated using HYDRUS-2D software to verify the rationality of the sand tank model compared with the results of the physical model. The main conclusions were as follows:

• (1)

  Only infiltration deep runoff and surface runoff existed in the sand tank experiment. The IDR of the sand gravel medium accounted for 65–85% of the total recharge water. A reasonable released water plan was necessary in advance because the remaining released water was considered an invalid recharge when the IDR reached the critical value of the medium IDR. The rubber dam had no obvious effect on improving IDR, but could promote the transformation from SR to LSR.

• (2)

  The thin clay layer formed by suspended particle deposition on the medium surface reduced the IDR value. After clogging, the surface sand gravel was washed and then backfilled into a sand tank for the tap water recharge test again. The IDR value can almost recover to the critical value of the medium IDR. The increase in suspended particle concentration promoted the suspended particle deposition in the sand tank. After filtration by the sand tank, the recharge water turbidity was reduced by at least 96%. The rubber dam can increase the suspended particle deposition mass in the sand tank.

• (3)

  By comparing the physical and numerical model results, the fitting degree of the pressure head with time change was more than 0.9. Thus, the sand tank model can be used to simulate the water variation in the piedmont sand gravel channel.

• (4)

  The research conclusions of this paper can provide a new recharge method by using the Yellow River water in the piedmont areas of the North China Plain.

This paper was supported by the Natural Science Foundation of Shandong Province (No. ZR2021ME069). We sincerely thank the editors and anonymous reviewers for their detailed and constructive comments on this article, which has greatly improved the quality of the article.

This paper was composed by collaboration among all authors. W.W. and S.L. designed this study and helped improve its progression and clarity. W.Z. wrote this paper. S.Q., D.X., X.S. and Y.M. helped in revising the paper.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.