Abstract

To investigate the impacts of inflow on overflow suspended solids (SS) concentration in rain gardens, field experiments were carried out in Guangming New District, Shenzhen, China. The pollutant degradation process was assumed to follow the first-order decay theory. Inflow rate and SS concentration data measured from a mild rainfall event were first used to calculate the decay constant based on the continuously stirred tank reactor (CSTR) theory, in which sediment remobilization was assumed to be negligible due to small inflow. Then, SS release rates of other rainfall events were calculated based on the known decay constant and the CSTR equations. Results show that maximum SS release rate has a strong correlation with the maximum inflow rate. Furthermore, sediment remobilization was observed in general, especially in cases of large inflow rate. Analyses show that the SS release rate is proportional to the inflow velocity. However, in the later stage of rainfall events, SS release rate drops more rapidly than the inflow velocity. This indicates the existence of a critical point, at which the inflow velocity does not have enough momentum to remobilize sediments. The study also demonstrates that small sediment particle sizes and uneven distribution of inflow could trigger sediment remobilization even with low inflow velocity.

INTRODUCTION

Widespread recognition of the negative impacts of urban stormwater (Hatt et al. 2004) has resulted in the identification of two important management goals, that is, maintaining stormwater quantity and quality. Low impact development (LID) practices, including many small scale green infrastructures, have proven effective in reducing stormwater runoff through in-situ retention and infiltration processes (Valinski & Chandler 2015).

As a typical measure of LID, a rain garden (also called bioretention pond) is a green infrastructure that can be used to control and utilize rainwater on the basis of mimicking the natural hydrological cycle and source control concepts. Prior research (Blecken et al. 2009; Zhang & Guo 2014) has demonstrated that rain gardens can effectively reduce the amount of runoff and remove suspended solids (SS), heavy metals, phosphorus, fats and oils, as well as hydrocarbons and pathogenic bacteria. Liu et al. (2015) concluded that compared with other operating parameters, type of medium and pollutant plays a critical role in the efficiency of processing. Jennings et al. (2015) stated that rain gardens give priority to penetration as a means of reducing runoff compared with evaporation and evapotranspiration.

Research has gradually shifted to focus on studying the internal operation theories and phenomena of rain gardens. For instance, the leaching phenomenon in the deeper medium of rain gardens has attracted wide attention (Brown & Hunt 2011; Chahal et al. 2016). Mullane et al. (2015) found that bioretention systems that have been enhanced with compost can be a source of nutrients in the discharge water. Søberg et al. (2017) observed dissolved Cu and Pb leaching from the deeper filter layers, but the majority of metals were captured in the surface of the filter material.

At the same time, research on the characteristics of sediments accumulated on the surface of LID practices progressed in many regions. McNett & Hunt (2011) evaluated the toxicity of accumulated sediments in forebays of stormwater wetlands and wetponds. They found that more than half of the sediment metal concentration in the samples taken from monitoring sites exceeded the limit for aquatic health. Jenkins et al. (2010) conducted a study on a rain garden to determine whether the sediments accumulated were associated with time and permeability and found that sediments had no significant effect on infiltration rate although may hinder the infiltration rate at specific points. While increased sediment accumulation implied that LID performed as intended, i.e., occurrence of accumulation, the available storage volume decreased. Hence, sediment removal rate also decreased, thereby increasing the risk of flooding and pollutant transfer downstream (Heal et al. 2006).

In spite of the need to understand the sediment movement mechanism, available field data on quantitative evaluation of sediment movement are still very limited due to impulsive and unpredictable occurrence of the phenomenon. Moreover, sediment settled inside LID practices does not imply pollutant reduction because the accumulated pollutants can still be remobilized from the sediments back to the water column during natural (e.g., hydrodynamics and bioturbation) or anthropogenic (dredging operations) perturbations. Even minor distribution would generate additional shear stresses at the bottom of LID practices and probably remobilize the surface sediments, which will eventually be transported by the flow to the outlet (Bentzen et al. 2009). The surface sediments that are remobilized with low energies (commonly present in estuarine environments) are usually richer in heavy metals compared to deeper sediments (Kalnejais et al. 2007).

However, there is a general absence of research quantifying the sediment remobilization process and its impact on overflow flow and SS concentration from rain gardens. Sediment remobilization occurs frequently because of irrational rain garden design parameters such as uneven distribution of inflow. Sediment remobilization is generally undesirable as it can contribute to various environmental problems such as raising overflow SS and other pollutant concentrations. While important, runoff source control is difficult due to the variety of runoff pollutant sources. Accordingly, stormwater quality treatment is an essential component in urban drainage to protect receiving waters (Walsh 2000). However, as with many other LID technologies, biofilters were mostly developed without specific consideration for their hydrological parameters, which may be critical, based on the work by Roseen et al. (2009).

Additionally, it has been observed that global climate change has increased the occurrence frequencies of intense rainfall events. As LID practices are generally designed to cope with mild or medium rainfall events, it is important to study quantitatively the potential flow conditions that could trigger the remobilization process as well as the impacts on the overflow water quality. Hence, quantitative characterization of remobilized sediments is the key to maintaining a good performance for LID facilities. Then, as a critical step to achieve good performance, there is a need to study the source and the mitigation approaches (Autixier et al. 2014).

Accordingly, a pilot site of the nationwide sponge city construction program in Guangming New District of Shenzhen in China was chosen as a case study to investigate the sediment remobilization processes. The continuously stirred tank reactor (CSTR) theory was applied to calculate the decay constant of a constructed rain garden in a mild rainfall event. Then, the decay constant was used to determine the SS release rate caused by sediment remobilization in rainfall events with different intensities, in order to explore the impacts of inflow velocity on overflow SS concentration from a rain garden and recommend general design preferences to mitigate the negative impacts.

METHODOLOGY

Site description

The sponge city pilot area, one of the green building communities in Guangming New District, Shenzhen, was selected as the study site. The city's average annual rainfall of 1,924.7 mm is concentrated in the rainy season from March to September. Guangming New District is located in the west of Shenzhen, China, which is the national Green Building demonstration zone and LID demonstration site.

This test device collected rainwater from three different land use types: roads, squares, and roofs (Figure 1). The profile of the land use types is shown in Table 1. The rainwater from the catchments was collected by separate plastic rainwater drainage pipes. All the rainwater reached an open channel before entering the device.

Table 1

Distribution of the land use types of the study catchment

Land use types Material Runoff coefficient Area (m2
Road Concrete 0.9 60 
Square Concrete 0.9 200 
Roof Concrete 0.9 190 
Land use types Material Runoff coefficient Area (m2
Road Concrete 0.9 60 
Square Concrete 0.9 200 
Roof Concrete 0.9 190 
Figure 1

Land use type.

Figure 1

Land use type.

Four rainfall events with different intensities were monitored between August 16 and September 13. The four events represent heavy, medium, and small rainfall events (Table 2).

Table 2

The rainfall event statistics

Date Rainfall (mm) Rainfall duration (min) Max rainfall intensity (mm/h) Rainfall categories 
2013.08.16 61.0 160 22.9 Heavy 
2013.09.02 18.0 45 24.0 Medium 
2013.09.04 37.9 140 16.2 Medium 
2013.09.13 8.2 20 24.6 Small 
Date Rainfall (mm) Rainfall duration (min) Max rainfall intensity (mm/h) Rainfall categories 
2013.08.16 61.0 160 22.9 Heavy 
2013.09.02 18.0 45 24.0 Medium 
2013.09.04 37.9 140 16.2 Medium 
2013.09.13 8.2 20 24.6 Small 

Experimental setup

The test device intercepts runoff by containing rainfall within the concrete wall. The interior is an ecological rainwater treatment measure constructed under a LID concept. The device was 3.5 m in length and 2.7 m in width. The width and depth of a rectangular open channel at the inlet of the device are both 0.1 m. The structure layers from the top to the bottom are set as follows: clearance, grass, improved soil, fine sand, filler, and pebble layer. A schematic view of the experimental setup is shown in Figure 2. The volume ratio of sand:clay is 1:1–1:1.5 to improve the soil permeability coefficient. Considering the climate conditions and rainfall pattern, Taiwan grass was selected as the plant vegetation (Figure 3). A layer of fine sand was used as a barrier between the sand and pebble layer. Slag, with permeability coefficient of 1.0*10−4 m/s, was chosen as filler in the filler layer. Perforated PVC pipes of 50 mm in diameter were installed at the bottom to collect the treated water.

Figure 2

Schematic of experimental setup.

Figure 2

Schematic of experimental setup.

Figure 3

Appearance of the device.

Figure 3

Appearance of the device.

The rainwater drained through perforated pipes underneath the rain garden for reuse while excess runoff was detained at the top. At the same time, when the rainwater exceeded the accumulation and infiltration capacity of the device, it overflowed through an overflow weir. The whole process mimicked a natural rainwater treatment process. Pollutants were mainly disposed by physical chemical and biological processes through a soil-microorganism-vegetation system.

The rainfall was monitored by automatic gauge (JDZ-1) automatic recording near the test point. The runoff volumes at the inlet and outlet were measured by a volume-time method. Flow gauging and sampling at the inlet and outlet during the rainy season controlled the runoff fluctuation process. Water samples were collected at least three times at the rising limb, peak, and recession parts, respectively. Since total suspended solids (TSS) are known as the main vector for pollutant transport in runoff, this study focused on the variation of that variable within rain events (APHA et al. 2005).

The water samples were immediately taken back to the laboratory after collection. The SS concentration of each sample was measured according to the Chinese national criteria (GB 11901-8 2009) using the gravimetric method.

Governing equations

During rainfall events, rainwater in and out of the rain garden is continuous and follows the operating conditions in a CSTR reactor. Furthermore, because of an open channel setting in the inlet, rainwater from the inlet can rapidly mix with rainwater in the rain garden and form the same pollutants' concentration in the reactor, which corresponds with the definition of a CSTR reactor. As a result, in the process of water quality routing, the rain garden is assumed to behave as a CSTR in which the effect of flow dilution in steady flow conditions progressively reduces the pollutant concentration at the outflow. The first-order decay function is a comprehensive model that combines different removal mechanisms including settling, bio-degradation, etc. It expresses the rate at which pollutant concentration moves towards an equilibrium or background concentration (C*), with distance along the treatment measure, as a linear function of the concentration. It has been widely documented that this model, known as the ‘k–C* model’, can be applied to a wide range of pollutants, including TSS, total phosphorus, and total nitrogen (Wong et al. 2006). Thus, the pollutant, i.e., SS degradation, is assumed to follow the first-order decay theory in the governing equation. The governing equation (Li et al. 2016) is shown below:  
formula
(1)
where = water volume, m3; = concentration in the mixed volume, mg/L; = inflow rate, L/s; = concentrations of the influent, mg/L; = outflow rate, L/s; = concentrations of the effluent, mg/L; = first-order decay constant, s−1; = source (or sink), mg/s.

Equation (1) is solved to predict the performance of a rain garden, in which the clearance layer behaves as a CSTR. At first, the rainwater infiltrates through porous media and is drained by a perforated pipe underneath the rain garden while excess runoff is detained at the top. Once the rainwater exceeds the accumulation and infiltration capacity, it overflows through the overflow outlet. The whole process of one rainfall event can be divided into three stages: stage 1, from initial time to the moment when overflow occurs; stage 2, from the time when overflow occurs to the moment when overflow ends; stage 3, from the time when overflow ends to the moment the rainfall event ends. In stage 2, part of the inflow infiltrates through the porous media while the rest overflows simultaneously.

In stage 1, inflow just fills the clearance layer with no overflow effluent. In stage 3, overflow ends. So we model stage 2 by assuming the infiltration concentration and concentration in the mixed volume to be equal to overflow concentration . Also, in this stage, inflow rate is equivalent to the outflow rate . Considering that the area of the rain garden is relatively small, the rainwater taken by itself is assumed to be negligible. Hence the source (or sink) of pollutants only includes those from the internal part of the rain garden.  
formula
(2)
Equation (3) can be obtained by plugging the above equations into Equation (1):  
formula
(3)

RESULTS

During the monitoring period, flow data for four storm events were collected. The field rainfall and water quality data of the four events are discussed in the following. The field data of September 4 is used to calculate the first-order decay constant. The field data of August 16, September 2, and September 13 are placed in case two to analyze sediment remobilization in each rainfall event.

Case 1: September 4 rainfall event

The inflow and outflow hydrographs for September 4 are shown in Figure 4(a).

Figure 4

(a) Inflow and outflow hydrograph on September 4. (b) Relationship of inflow velocity and decay constant.

Figure 4

(a) Inflow and outflow hydrograph on September 4. (b) Relationship of inflow velocity and decay constant.

According to the inflow and overflow concentration of four rainfall events, SS removal rate was −17.90% on August 16, 7.04% on September 4, 39.94% on September 4, and −22.89% on September 14. Compared with the other three rainfall events, the degradation effect on the September 4 rainfall event was better. Given the great degradation effect, relatively stable inflow change, and smaller inflow concentration, it is desirable to ignore sediment remobilization in this rainfall event, that is L = 0.

Equation (4) can be obtained by transformation of Equation (3):  
formula
(4)
Because of the monitoring data interval, is treated as the ratio of concentration difference to time difference between two adjacent monitoring time points. The derivative of the concentration and time at t is defined as the ratio of the value of the function at time t + 1 minus the function at time t and Δt. Thus, Co stands for the overflow concentration at time step t.  
formula
(5)
Consequently, the decay constant during the monitoring period was calculated by the above equations. To figure out the impact of inflow conditions on sediment remobilization, the inflow velocity was also calculated. The calculated decay constant and SS concentration of inflow and outflow are shown in Table 3.
Table 3

The decay constant calculation

(min)  (mg/L)  (mg/L)   K (s−1
379 – – – – 
10 165 – – – – 
15 125 – – – – 
20 113 – – – – 
25 98 – – – – 
30 93 – – – – 
35 93 71 −10.67 12.06 – 
45 50 61 −13.34 14.3 0.000387 
55 46 49 −16 2.88 0.000361 
65 45 40.5 −11.34 −1.2 0.000333 
95 44 27 −6 1.49 0.000342 
125 40 21.83 −2.30 5.1 – 
155 – – – – 
(min)  (mg/L)  (mg/L)   K (s−1
379 – – – – 
10 165 – – – – 
15 125 – – – – 
20 113 – – – – 
25 98 – – – – 
30 93 – – – – 
35 93 71 −10.67 12.06 – 
45 50 61 −13.34 14.3 0.000387 
55 46 49 −16 2.88 0.000361 
65 45 40.5 −11.34 −1.2 0.000333 
95 44 27 −6 1.49 0.000342 
125 40 21.83 −2.30 5.1 – 
155 – – – – 

The relationship between inflow velocity and the decay constant is shown in Figure 4(b).

For this rainfall event, the decay constant was relatively stable with average value of 0.00035 s−1. As for the same rain garden, the factors affecting the decay constant are underlying surface type and pollutant particle size generally. In this study, considering the same catchments and short monitoring time interval, this decay constant could reasonably be applied to other rainfall events.

Case 2: Sediment remobilization analysis

The inflow and outflow hydrographs for each rainfall event are shown in Figures 5(a), 6(a), and 7(a).

When is higher than , it means that sediment remobilization occurred in the rainfall event. Quantitative analysis of sediment remobilization can be obtained by solving the following equation (Equation (6)), which can be obtained by transformation of Equation (3):  
formula
(6)
Consequently, the pollution release rate was calculated by the above equation based on the known decay constant and CSTR theory. To figure out the impact of inflow conditions on sediment remobilization, the inflow velocity was also calculated. The SS release rate and flow concentration of the rainfall events are shown in Tables 46, respectively.
Table 4

The SS release rate calculation of August 16 rainfall event

(min)  (mg/L)  (mg/L)   L (mg/s) 
110.79 – – – – 
107 – – – – 
11 104.96 – – – – 
16 102.92 – – – – 
21 101.75 – – – – 
26 99.71 – – – – 
31 99.42 – – – – 
36 97.23 152.20 −19.83 −135.52 115.69 
46 90.67 137.33 −34.21 −172.83 138.62 
56 84.40 111.67 −15.56 −109.81 94.22 
66 84.20 100.00 −2.09 −72.24 70.15 
96 72.73 95.29 −4.99 −72.48 67.49 
126 75.07 98.07 −37.36 −31.91 −5.45 
156 – – – – 
(min)  (mg/L)  (mg/L)   L (mg/s) 
110.79 – – – – 
107 – – – – 
11 104.96 – – – – 
16 102.92 – – – – 
21 101.75 – – – – 
26 99.71 – – – – 
31 99.42 – – – – 
36 97.23 152.20 −19.83 −135.52 115.69 
46 90.67 137.33 −34.21 −172.83 138.62 
56 84.40 111.67 −15.56 −109.81 94.22 
66 84.20 100.00 −2.09 −72.24 70.15 
96 72.73 95.29 −4.99 −72.48 67.49 
126 75.07 98.07 −37.36 −31.91 −5.45 
156 – – – – 
Table 5

The SS release calculation of September 2 rainfall event

(min)  (mg/L)  (mg/L)   L (mg/s) 
100 – – – – 
90 – – – – 
14 70 – – – – 
19 69 – – – – 
24 66 – – – – 
29 64.71 90 −5.33 −48.72 43.39 
34 55 88 −10.67 −65.23 54.56 
44 54.51 80 −13.33 −52.99 39.65 
54 54 70 −2.67 −33.04 30.37 
64 54 68 −30.22 −25.20 −5.02 
94 – – – – 
(min)  (mg/L)  (mg/L)   L (mg/s) 
100 – – – – 
90 – – – – 
14 70 – – – – 
19 69 – – – – 
24 66 – – – – 
29 64.71 90 −5.33 −48.72 43.39 
34 55 88 −10.67 −65.23 54.56 
44 54.51 80 −13.33 −52.99 39.65 
54 54 70 −2.67 −33.04 30.37 
64 54 68 −30.22 −25.20 −5.02 
94 – – – – 
Table 6

The SS release calculation of September 13 rainfall event

(min)  (mg/L)  (mg/L)   L.(mg/s) 
131 – – – – 
11 115 – – – – 
16 72 – – – – 
21 66 – – – – 
26 64 84 −5.33 −62.12 56.79 
31 62 82 −16 −59.56 43.56 
36 60 76 −1.33 −43.84 42.51 
46 55 75 −100.00 −30.00 −70.00 
56 – – – – 
(min)  (mg/L)  (mg/L)   L.(mg/s) 
131 – – – – 
11 115 – – – – 
16 72 – – – – 
21 66 – – – – 
26 64 84 −5.33 −62.12 56.79 
31 62 82 −16 −59.56 43.56 
36 60 76 −1.33 −43.84 42.51 
46 55 75 −100.00 −30.00 −70.00 
56 – – – – 

The relationship between inflow velocity and SS release rate of each rainfall event is shown in Figures 5(b), 6(b), and 7(b), respectively. Since the inlet channel area remains the same, the change of inflow velocity is the same as inflow rate. In general, sediment remobilization was observed, especially in cases of large inflow rates. The maximum inflow rate and maximum SS release rate for each rainfall event are shown in Table 7 and the correlation between them is shown in Figure 8.

Table 7

The maximum SS release rate and inflow rate

Monitoring time August 16 September 2 September 13 
Max SS release rate (mg/s) 138.62 54.56 56.79 
Maximum inflow rate (L/s) 2.88 1.23 2.38 
Monitoring time August 16 September 2 September 13 
Max SS release rate (mg/s) 138.62 54.56 56.79 
Maximum inflow rate (L/s) 2.88 1.23 2.38 
Figure 5

(a) Inflow and outflow hydrograph on August 16. (b) Relationship of inflow velocity and SS release rate on August 16.

Figure 5

(a) Inflow and outflow hydrograph on August 16. (b) Relationship of inflow velocity and SS release rate on August 16.

Figure 6

(a) Inflow and outflow hydrograph on September 2. (b) Relationship of inflow velocity and SS release on September 2.

Figure 6

(a) Inflow and outflow hydrograph on September 2. (b) Relationship of inflow velocity and SS release on September 2.

Figure 7

(a) Inflow and outflow hydrograph on September 13. (b) Relationship of inflow velocity and SS release on September 13.

Figure 7

(a) Inflow and outflow hydrograph on September 13. (b) Relationship of inflow velocity and SS release on September 13.

Figure 8

Correlation between maximum SS release rate and inflow rate.

Figure 8

Correlation between maximum SS release rate and inflow rate.

DISCUSSION

Relationship between maximum inflow rate and maximum SS release rate

The results demonstrated that maximum SS release rate has a strong correlation with the maximum inflow rate, and is in line with the exponential growth pattern, which equation can be described as . And a, b, c are undetermined parameters that depend on the particular rain garden designs. The maximum SS release rate (138.62 mg/s) corresponds to the maximum inflow rate (2.88 L/s) during the whole monitoring period. When the maximum inflow rate is relatively small (i.e., 1.23 and 2.38 L/s), the maximum SS release rate (i.e., 54.56 and 56.79 g/s) remains almost the same. However, when the maximum inflow rate increases from 2.38 L/s to 2.88 L/s, the upward trend of the maximum SS release rate increases even more. It can be predicted that if maximum inflow rate continues to increase, the rate of maximum SS release will increase faster. This relationship can be used to predict the maximum SS release rate in any of the maximum influent flow conditions.

Relationship between inflow velocity and SS release rate

Additionally, it is possible to draw the conclusion that SS release rate is proportional to the inflow velocity. The greater inflow velocity, the greater SS release occurs. The SS release rate ranges from −70.00 mg/s to 138.62 mg/s in all rainfall events. The positive and negative values that can also be seen in a certain rainfall event means the source and sink of pollutants can occur in one rainfall event. It is also worth mentioning that the moment maximum SS release rate matches with the moment maximum inflow velocity. However, in the later stage of rainfall events, the decrease of SS release rate is more rapid than the inflow velocity. This is most likely due to the fact that there is a critical condition for triggering sediment remobilization. In the later stage of rainfall events, the exterior circumstances, including inflow velocity and particle size, are insufficient to attain the critical condition.

Relationship between uneven distribution of inflow and SS remobilization

Sediment resuspension is presumably caused by increased shear stress on sediments (Lovstedt & Bengtsson 2008). Increased shear stress is often caused by relatively large inflow velocity. In general, a minimum critical velocity for light sand particles to get resuspended is 0.1 m/s. In this study, some of the inflow velocities are under the critical value which is less likely to create sediment resuspension. However, prior research has found that typical sediments accumulated on the top of biological treatments are atmospheric sedimentations that are composed of small particle sizes and light-mass lawn clippings. In the meantime, the open channel section in the inlet of the rain garden creates a flow short-circuit in the rain garden, which increases the potential of sediment remobilization even with small inflow velocity.

Generally, the design of rain gardens includes an open channel section at the inlet or direct runoff from roadway curbsides to rain gardens without sophisticated considerations of flow short-circuiting issues. For LID practices, the most effective treatment takes place when the rainwater flows through the entire surface area, with no stagnant region and minimal short-circuiting conditions. As a result, the possibility to remobilize sediments would decrease. The risk of pollutants transferred to downstream would decrease accordingly.

LID effectiveness and recommendations

According to the results, it can be concluded that the degradation effect of sediment for this rain garden is not satisfactory. Marutani et al. (1999) reported an 8- to 25-year lag time for reducing sediment delivery and stored sediment removal in streams in New Zealand. This transitional-period condition recognizes that LID implementation and effectiveness are not usually concurrent. There is a certain time period required for a system with LID practices to become stabilized and begin to provide effective non-point source pollutant removal. This rain garden has not been established for long so the operation effect is still unstable, and as a result, it is easy to produce an outflow pollutant concentration higher than the inflow. At the same time, the design of this rain garden lacks consideration of inflow distribution and hydraulic conditions.

Based on the simulation results of the CSTR model, the following two mechanisms (i.e., an increment of water depth and a buffer zone pool) are prevailing in the resuspension reduction of the deposited materials. An increment of the water depth of the pond can be used to minimize the effect of the flow speed. An increase in water depth will give a reduction of the yearly discharge mass (Bentzen et al. 2009). Also, when considering the sediment accumulation and resuspension effect on LID performance, it has the necessity to consider an optimal depth that takes into account the negative effect of a shallow depth on sediment resuspension. A buffer zone pool set in front of LID practices can also be considered. It is generally known that a well-designed buffer pool can reduce the rainwater speed within distances. This practice also leads to a significant drop in sediment resuspension. As an additional comment, it can be mentioned that short-circuiting flow can be avoided by dividing the inflow area into multiple-influent points.

All technologies employed will be beneficial to increase hydraulic residence, decrease runoff velocity and sediment remobilization. As a consequence, LID performance can be improved at ecosystem level.

This quantitative analysis provides information about the impacts of sediment remobilization on overflow SS concentration. In addition, it demonstrates that external conditions such as inflow distribution and hydraulic conditions play significant roles in the performance of LID practices which should be paid more attention in the design of further engineering practices.

CONCLUSIONS

Conclusions are drawn as follows:

  • (1)

    Results show that sediment remobilization presents in general, especially in cases of large inflow rates. The maximum SS release rate has a strong correlation with the maximum inflow rate. The maximum SS release rate (138.62 mg/s) corresponds to the maximum inflow rate (2.88 L/s) during the whole monitoring period. Also, the variation of SS release rate is consistent with the change of inflow velocity and the moment maximum SS release rate matches with the moment maximum inflow velocity.

  • (2)

    The SS release rate ranges from −70.00 mg/s to 138.62 mg/s in all rainfall events. The positive and negative values can also be seen in a certain rainfall event.

  • (3)

    In the later stage of rainfall events, the decrease of SS release rate is more rapid than the inflow velocity. This is more likely due to the fact that sediment remobilization has a critical triggering point. In the later stage of rainfall events, the initiating conditions including inflow velocity and particle size are insufficient to attain the critical condition.

  • (4)

    Open channel section in the inlet of rain gardens or direct runoff from roadway curbsides to rain gardens results in stagnant zones and flow short-circuiting. Small sediment particle sizes and uneven distribution of inflow of rain gardens could trigger sediment remobilization even in low inflow velocity.

  • (5)

    According to the conclusions from the results, the recommendations for further engineering practice are as follows: (1) an increment of the water depth of the pond can be used to minimize the effect of the flow speed; (2) consider the water distribution of water in the rainwater garden such as dividing the inflow area into multiple-influent points; (3) a buffer zone pool not only creates uniform water distribution but also intercepta a part of SS to avoid the risk of sediment remobilization.

ACKNOWLEDGEMENTS

The research reported here is supported by the National Key R&D Program of China, non-point source pollution control and stormwater management technology in hilly cities, the Ministry of Science and Technology, PR China (No.2017YFC0404704).

REFERENCES

REFERENCES
APHA, AWWA, WEC
2005
Standard Methods for the Examination of Water and Wastewater
,
21st edn
.
American Public Health Association
,
Washington, DC
.
Autixier
,
L.
,
Mailhot
,
A.
,
Bolduc
,
S.
,
Madoux-Humery
,
A.-S.
,
Galarneau
,
M.
,
Prevost
,
M.
&
Dorner
,
S.
2014
Evaluating rain gardens as a method to reduce the impact of sewer overflows in sources of drinking water
.
Science of the Total Environment
499
,
238
247
.
Bentzen
,
T. R.
,
Larsen
,
T.
&
Rasmussen
,
M. R.
2009
Predictions of resuspension of highway detention pond deposits in interrain event periods due to wind-induced currents and waves
.
Journal of Environmental Engineering
135
(
12
),
1286
1293
.
Blecken
,
G. T.
,
Zinger
,
Y.
,
Deletic
,
A.
,
Fletcher
,
T. D.
&
Viklander
,
M.
2009
Influence of intermittent wetting and drying conditions on heavy metal removal by stormwater biofilters
.
Water Research
43
,
4590
4598
.
Brown
,
R. A.
&
Hunt
,
W. F.
2011
Impacts of media depth on effluent water quality and hydrologic performance of under sized bioretention cells
.
Journal of Irrigation & Drainage Engineering
137
(
3
),
132
143
.
Chahal
,
M. K.
,
Shi
,
Z.
&
Flury
,
M.
2016
Nutrient leaching and copper speciation in compost-amended bioretention systems
.
Science of the Total Environment
556
,
302
309
.
Hatt
,
B. E.
,
Fletcher
,
T. D.
,
Walsh
,
C. J.
&
Taylor
,
S. L.
2004
The influence of urban density and drainage infrastructure on the concentrations and loads of pollutants in small streams
.
Environmental Management
34
(
1
),
112
124
.
Heal
,
K.
,
Hepburn
,
D.
&
Lunn
,
R.
2006
Sediment management in sustainable urban drainage system ponds
.
Water Science and Technology
53
(
10
),
219
227
.
Jenkins
,
J. K. G.
,
Wadzuk
,
B. M.
&
Welker
,
A. L.
2010
Fines accumulation and distribution in a storm-water rain garden nine years postconstruction
.
Journal of Irrigation & Drainage Engineering
136
(
12
),
862
869
.
Jennings
,
A. A.
,
Berger
,
M. A.
&
Hale
,
J. D.
2015
Hydraulic and hydrologic performance of residential rain gardens
.
Journal of Environmental Engineering
141
(
11
),
04015033
.
Kalnejais
,
L. H.
,
Martin
,
W. R.
,
Signell
,
R. S.
&
Bothner
,
M. H.
2007
Role of resuspension in the remobilization of particulate-phase metals from coastal sediments
.
Environmental Science and Technology
41
,
2282
2288
.
Li
,
J.
,
Li
,
Y.
&
Li
,
Y.
2016
SWMM-based evaluation of the effect of rain gardens on urbanized areas
.
Environmental Earth Sciences
75
(
1
),
17
.
Liu
,
A.
,
Jiang
,
Y.
&
Guan
,
S. D. Y.
2015
Characterizing stormwater treatment efficiency at the laboratory scale for effective rain garden design
.
Desalination & Water Treatment
54
(
4–5
),
1334
1343
.
Marutani
,
T.
,
Kasai
,
M.
,
Reid
,
L. M.
&
Trustrum
,
N. A.
1999
Influence of stormrelated sediment storage on the sediment delivery from tributary catchments in the Upper Waipaoa River, New Zealand
.
Earth Surface Processes and Landforms
24
,
881
896
.
Ministry of Environmental Protection of the People's Republic of China (GB 11901-8)
2009
Water Quality-Determination of Suspended Substance-Gravimetric Method
.
Mullane
,
J. M.
,
Flury
,
M.
,
Iqbal
,
H.
,
Freeze
,
P. M.
,
Hinman
,
C.
,
Cogger
,
C. G.
&
Shi
,
Z.
2015
Intermittent rainstorms cause pulses of nitrogen, phosphorus, and copper in leachate from compost in bioretention systems
.
Science of the Total Environment
537
,
294
303
.
Roseen
,
R. M.
,
Ballestero
,
T. P.
,
Houle
,
J. J.
,
Avellaneda
,
P.
,
Briggs
,
J.
,
Fowler
,
G.
&
Wildey
,
R.
2009
Seasonal performance variations for storm-water management systems in cold climate conditions
.
Journal of Environmental Engineering
135
,
128
137
.
Søberg
,
L. C.
,
Viklander
,
M.
&
Blecken
,
G. T.
2017
Do salt and low temperature impair metal treatment in stormwater bioretention cells with or without a submerged zone?
.
Science of the Total Environment
579
,
1588
1599
.
Valinski
,
N. A.
&
Chandler
,
D. G.
2015
Infiltration performance of engineered surfaces commonly used for distributed stormwater management
.
Environmental Management
160
,
297
305
.
Wong
,
T. H. F.
,
Fletcher
,
T. D.
,
Duncan
,
H. P.
&
Jenkins
,
G. A.
2006
Modelling urban stormwater treatment – A unified approach
.
Ecological Engineering
27
(
1
),
58
70
.
Zhang
,
S. H.
&
Guo
,
Y. P.
2014
Stormwater capture efficiency of bioretention systems
.
Water Resource Management
28
(
1
),
149
168
.