Understanding the Role of Constructed Wetlands in Stormwater Management

Abstract

Constructed wetlands have been utilized for some time in the treatment of wastewater and have been recognized for the treatment of stormwater runoff and flood protection in the last couple of decades. Constructed wetlands are built to remove sediment and nutrients, primarily phosphorus and nitrogen, from contaminated water. However, with increased urbanization and enhanced climate change, these constructed wetlands need to be managed and their treatment effectiveness monitored and maintained especially at the post-construction phase. In addition, a greater understanding of the role of these systems in the urbanized environment and how they treat wastewater are needed to optimize their performance. As more advanced computer modeling is developed there is a need to ascertain what parameters and how these changes overtime and what skills are required to enable the adoption of constructed wetlands for future planning and management. There has been limited research into constructed wetlands for flood mitigation and with some receiving inflows larger than their design intent, it is necessary to determine if these systems would still be able to treat pollutants. This chapter involves a review of the literature to address these concerns relating to constructed wetlands.

1. Introduction

Whilst constructed wetlands have been utilized for some time in the treatment of wastewater, they only gained popularity for the treatment of stormwater runoff and flood protection in the last couple of decades [1, 2]. Constructed wetlands are employed to remove sediment and nutrients, primarily phosphorus and nitrogen, from contaminated water [1]. However, with increased urbanization and enhanced climate change, these constructed wetlands need to be managed and their treatment effectiveness monitored and maintained once these are established.

Constructed wetlands remove sediments using large ponds which allow for dissipation of water velocities, making sediment particles drop out of the water column, settling at the bottom of the basin which is then supposed to be cleaned regularly [1]. The removal of nutrients by constructed wetlands is varied and complex and is understood to be primarily undertaken by anoxic microbial degradation processes within the wetland environment [3]. Secondary to this, plants also capture nitrogen and remove it from the system through various methods such as volatilization and phytodegradation [1]. Wetland performance in treating stormwater is generally a function of hydraulic loading rate and detention time and these two parameters are in turn functions of the runoff volume, storm intensity and the wetland size itself [4]. Sizing the wetland is crucial for the health of the wetland treatment system. The hydrodynamic criteria of wetlands such as the inundation depth, wetness gradient, base flow and hydraulic regime are crucial for wetland sizing. If these hydrodynamic characteristics receive inadequate attention, the performance of treating stormwater is likely to be reduced.

Constructed wetlands also provide a retarding function, and therefore can be utilized to assist in flood protection in urban areas. As wetlands are generally controlled by a pit and a piped outlet, they can act under the same principle as a retarding basin by discharging flood flows at a controlled rate. With more wetlands being constructed and restored widely, flood storage capacity is increased, and the flood peak is reduced. Constructed wetlands and the restoration of wetlands have come into wide practice as they have the potential to act as an effective water treatment basin as well as provide essential flood control [5].

In Australia, particularly in metropolitan areas, constructed wetlands have become a common occurrence [3]. There is a multitude of reasons why wetlands have increased in popularity over the years, some reasons include their effectiveness in treating stormwater, their ability to treat large areas, scalability, their cost-effectiveness, ease of maintenance, and because they can act as a feature [6, 7]. Perhaps the main driver for their increase in popularity in Victoria, Australia is the legislation that requires all new developments to treat the additional runoff caused by the newly installed impervious surfaces to acceptable levels so that the concentrations of nutrients in receiving waters are not compromised [8].

The Best Practice Guidelines recommend the removal of 80% TSS (Total suspended solids) [9]. This is generally managed practically by sizing a constructed wetland to treat 100% of rainfall flows up to 1 in 3 months ARI rainfall intensity, which in Victoria represents on average 80% of the total annual rainfall [10]. Depending on the design, i.e., if the wetlands are constructed “online” as opposed to “offline”, constructed wetlands may experience inflows from larger rainfall intensities. However, the best practice guidelines do not consider detention characteristics of the wetlands.

To date, there has been limited research into constructed wetlands treating flows larger than its design intent. Whilst there have been studies which investigate the capabilities of constructed wetlands over long periods, e.g., 2 years, which cover a range of storms [11, 12], there appears to be little research directly focusing on the treatment that occurs from wetlands receiving flows larger than their design capacities.

2. Why are constructed wetlands utilized in the urban environment?

Urban Stormwater: Best Practice Environmental Management (BPEM) Guidelines [9] outline the principles and objectives behind the use of water treatment devices such as wetlands by referencing SEPPs (State Environmental Protection Policies) of Victoria, Australia. These objectives are clear: to preserve the beneficial uses of local waterways, including:

• natural aquatic ecosystems and associated wildlife;

• water-based recreation;

• agricultural water supply;

• potable water supply;

• production of mollusks for human consumption;

• commercial and recreational use of edible fish and crustacea; and

• industrial water use.

As urbanization increases, so too does the anthropogenic environmental impact, which includes pollutants occurring from human lifestyle activities, building and infrastructure, construction activities [9], urban vehicular emissions and increased stormwater runoff volume peaks and nutrient loads due to hard surface runoffs inherent in typical urban development [13]. Constructed wetlands, as part of a SUDS (Sensitive Urban Drainage System) can be used to mitigate these impacts [14], and within Victoria, Australia they are typically installed as part of large-scale urban developments to comply with the BPEM guidelines.

Wetlands are cost-effective treatment systems that can be used to treat urban stormwater runoff. The advantages for installation of constructed wetlands, including “(1) low cost of construction, and, especially, maintenance; (2) low energy requirements; (3) being a ‘low-technology’ system, they can be established and run by relatively untrained personnel; and (4) the systems are usually more flexible and less susceptible to variations in loading rate than conventional treatment systems” [3]. In addition, “the major disadvantages of constructed wetland treatment systems are the increased land area required, compared to conventional systems, and the possible decreased performance during winter in temperate regions” [3]. Under Melbourne Water’s drainage Schemes, landowners and developers are fairly compensated for land use dedicated to wetlands, thus mitigating the disadvantage of the increased land area required [15]. As wetlands are integrated into large-scale urban development design and construction, they are typically designed to receive flows up to the 1 in 3-month ARI rainfall to minimize land take whilst treating to best practice.

3. How do constructed wetlands treat stormwater?

To enable efficient operation and maintenance of constructed wetlands, the current practice is to split the treatment processes into separate zones; the sediment pond and macrophyte zone. Figure 1 shows the design layout for constructed wetlands within Melbourne as per the guideline and Figure 2 indicates the sediment storage pond.

Figure 1.

Constructed wetland diagram [16].

Figure 2.

Sediment pond storage [17].

3.1 Sediment removal

The sediment pond is utilized to remove sediments before the wetland. Current design guidelines require that sediment ponds shall be sized to retain 95% of sand particles (i.e., particles down to 125 μm) during a 1 in 3-month ARI rainfall event [16]. Regular cleanout of retained sediment is required whenever retained sediment levels reach 500 mm below the normal water level (NWL) [17]. There is conjecture as to how well this maintenance is carried out, as many existing constructed wetlands do not receive regular sediment maintenance.

Secondly, the Macrophyte Zone (shallow water-filled ponds planted with aquatic plants) is utilized to contain the remaining sediments not captured by the sediment pond by the velocity reducing and filtering effect of Macrophyte (aquatic plant) stems and root systems [18].

3.1.1 Nitrogen removal

The macrophyte zone removes nitrogen through the Nitrification–denitrification process [19, 20]. Sediment ponds also provide limited nutrient treatment. This includes an attachment of approx. 15% of Total Nitrogen (TN) and Total Phosphorus (TP) to particles larger than 300 um [21], and small amounts via anoxic biochemical processes [3]. Studies on the Prado wetlands in California USA suggest that constructed wetlands have, on average, 50–60% TN removal [22].

3.1.2 Phosphorous removal

Phosphorus removal is primarily through sorption by clay, and typically will reach a saturation point after which phosphorus can leach back into the water it should be treating [23], and upon reaching this point, wetland clay materials should be replaced. Additional sorption can be achieved via decomposed plant matter [23]. Studies on the Prado wetlands in California USA suggest that constructed wetlands have, on average, 40–50% TP removal [22].

3.2 On-line/offline wetland operation

Standard wetland designs in Victoria include two typical layouts: online wetlands with the sediment pond receiving unrestricted flows and the macrophyte zone receiving controlled flows, and alternatively, offline wetlands where flows into the wetland can be restricted using a diversion pit [16]. Online flows include the total runoff from a catchment area. Offline flows divert a volume of the flow up to a certain amount, and the remaining flow bypasses the wetland. If a wetland does not include a sediment pond, then the bypass can be managed via a pit arrangement with suitable baffle, invert levels and pipe sizes to suit the urban catchment area.

Wetlands that are receiving an online flow passing through the Macrophyte zone are at risk of having collected sediments and nutrients dislodged and washed downstream during peak flows [16]. Standard wetland design practice includes a bypass channel to prevent this occurrence as shown in Figure 3.

Figure 3.

Schematic diagram of a constructed wetland, adapted from [16].

4. How do wetlands provide flood control?

Severe flood events can be distinguished as a natural disaster as its effect includes damages to properties and agricultural lands and in some cases loss of life. Flooding may be caused by dam failures, snow melts and when a large amount of rainfall occurs, and the natural waterways do not have enough capacity to convey excess water and result in overland flow. Overland flow can result from two hydrological processes: the first process is through a big storm event, where the rainfall intensity is large and it cannot infiltrate into the soil and the other process is when the soil is oversaturated - where there is no more capacity for the soil to hold extra rainwater [5]. Rain that falls onto the surface can either go through the evaporation process, get infiltrated into the soil, run along the impervious surfaces or get captured in hollows surface of the ground or wetlands. In the last 150 years, we have lost almost 70% of the capacity of the soil to hold water due to developments being built and more impervious areas installed [24]. This capacity needs to be restored to reduce the amount of overland flows and the risk of properties and people getting inundated. To minimize the risk of flooding and to protect assets and properties downstream, engineers have manipulated the use of constructed wetlands for flood mitigation and control.

In the past, engineers used structural restraints such as levees, which is not always the optimal solution to prevent flooding. Unfortunately, levees tend to hold up water which can significantly increase the level of the river stage and increase the flood velocity. There are a few flood cases where flood water would surge and overtops levees like the flood event in 1993 along the Mississippi River and Missouri River [24]. The flooding in mid-July 1993 from the Mississippi River and Missouri River exceeded the 100-year average recurrence interval (ARI) and had caused major properties damage cost of between US$ 12 billion–US$ 16 billion and approximately 32 losses of life [25]. As more areas are being inundated by overland flow and more properties are damaged, engineers have re-evaluated the situation and have considered other alternatives for flood mitigation plans and developing effective floodplain management programs. The aftermath of the great flooding of the Mississippi and Missouri River, wetlands have been considered as part of the flood management and flood mitigation process [26]. Wetlands have been installed and restored in many areas by the US federal government and the US Army Corps of Engineers in the United States and flooding at the downstream of the rivers was reported to have been reduced [27].

Constructed wetlands and the restoration of wetlands have come into wide practice as they have the potential to act as an effective water treatment basin as well as providing essential flood control. Wetland can alter flooding in many ways such as reducing the peak flood water level, the timing of flood water can be delayed, or the flows can be reduced by providing flood storage [5]. The location of wetlands is important for implementing flood protection. Wetlands that are located at an upstream location, the wetland will mainly be affected by headwater from rainfall, whereas wetlands that are located at the downstream location will mostly be affected by river flow. Constructed wetland at the downstream locations is often dry before a storm event therefore it has the potential to store more water during a storm event [5].

The sizing of wetlands is also equally important to the location of wetlands where continuous hydrologic modeling must be undertaken to simulate wetland storage during major storm events. Moreover, the time of concentration that a large flood needs to reach the wetlands must be taken into consideration. With the right computed concentration time, it can help reduce flooding at the downstream location just before the flood peaks [26]. Like a retarding basin, flood water that is stored in wetlands helps to delay the peak time of downstream hydrograph by releasing its water slowly and in a controlled manner. According to [26], the drainage of wetlands can influence flood levels where the storage of water attenuates and potentially delay downstream flood peaks.

Constructed wetlands are designed to replicate natural wetlands such as meadows, saltwater marshes, forested wetlands, and bogs. Aquatic or wetlands plants are used in constructed wetlands not only to reduce the amount of pollutants for stormwater quality but also to offer an ecological habitat to a wide range of wildlife species. Moreover, wetlands plants can create great landscape features and recreational amenities for the community [28]. Wetlands not only act as a flood storage but also have the capability to reduce the velocity of flood water with the influence of wetlands plants or vegetation. According to [29], velocities of flood water through wetlands are usually delayed by a friction factor which is influenced either by the slope, depth and/or the vegetation type (density and height of plants). This friction factor also known as the Manning’s roughness coefficient developed by Chow in 1959 is widely applied by engineers to calculate the resistance of flow in open channels.

Velocity rate can be altered depending on the surface roughness where a higher Manning’s value can significantly reduce the velocity of water. This can be seen in Manning’s equation where velocity is respectively proportional to the roughness coefficient [29]. Wetland’s riparian or vegetation have higher Manning’s value (“n”) in comparison with “n” value of a concrete path (Vegetation, n = 0.06 and concrete path, n = 0.02). Therefore, the vegetation of wetlands can influence flows in which the roughness of wetlands channel bed can potentially reduce the velocity of flood water and reduce the peak flow discharge [5]. According to [30], with the presence of wetland plants, the velocity profile is uniform in the vertical direction. This is illustrated in Figure 4. The velocity of flows through wetlands may be reduced significantly depending on the vegetation density and the height of the wetland plants increase friction and drag by the vegetative stems [30].

Figure 4.

Schematic elevation view of a wetland emergent plant (adapted from [30]).

The importance of wetlands for flood mitigation is now recognized and understood widely. Ogawa and Male [31] performed a hydraulic simulation on evaluating the flood mitigation potential of wetlands for the “Charles River, Neponset River and Ten Mile River in Massachusetts”. From the simulation, the results implied that both upstream and downstream wetland locations altered peak flows and reduced flooding. Another study of wetland’s role for flood mitigation was done for the Red River Valley major flood in 1997 and the damage cost by the flood that year was US$ 3.5 billion. The study concluded that a 5% increase in wetland area would significantly reduce flood volume by 5.6% for the 1997 flood event and also reduce the amount of damage cost [32]. Moreover, the restoration of wetlands within the Devils Lake basin of North Dakota could potentially store 72% for the 2 Year ARI storm event and 41% for the 100 Year ARI storm event of total runoff and the US Corps of Engineers calculated that flood damage cost (Approximately US$ 17 million) can be prevented each year with the use of wetlands for flood mitigation [5]. Wetlands are not only used to treat stormwater but also play an important part in reducing flood peaks and flooding at the downstream locations.

5. Are existing wetlands capable of treating flows larger than what they are designed for?

The most crucial part of designing wetlands is the sizing of wetlands. It is recommended that the capacity of a wetland should be at least 3% of its catchment size or can take 1 in 3 months flow to remove pollutants within the guidelines for stormwater quality treatment [17]. Since there is limited research on wetlands receiving flows larger than their design intent, this section will review the following fundamental questions:

5.1 How do depth, duration, and frequency of flooding influence wetlands plants?

The duration and frequency of overland flow (water regime) can be a major influence on the development of wetland plants. In Australia, wet to dry seasonal changes cause different water levels each season which can potentially affect plant growth and responses [33]. Wetlands plants are utilized to prevent erosion, capture fine particles and to trap pollutants from runoff [3, 17].

An experiment was done to determine whether the depth, duration and frequency of flooding influence the development of wetlands plants. The experiment was done with different combinations of depth, duration, and frequency of flooding with different types of wetlands plants collected from (a) wetlands that are rarely dry (near permanent) and (b) poorly drained wetlands (intermittent) in New South Wales [33]. All 17 treatments were placed outdoor in uncovered tanks to allow the plants to be exposed to rainfall and some tanks were filled with water depending on the treatment. According to [33], the experiment was conducted for more than 16 weeks using different combinations to maximize results as plants are well established after this period of time. The results show that the depth of flooding does have a significant impact on the biomass and the different species of wetland plants for both wetland types (near permanent and intermittent) and the results of the durations of inundated plants varied between plants from the two wetlands.

The results indicated that the inundated plants from the near-permanent wetlands show no difference between different durations of flooding (4, 8 and 12 weeks) whereas plants from the intermittent wetlands show there was a decline in some species of plants. As for the flooding frequency, no significant impact on plants establishments for all 17 treatments for plants that are from the near-permanent wetland but for the plants from the intermittent wetlands some decline of plants species are reported. The highest biomass and the vast plant species can be seen from the treatment tank that were never flooded [33]. According to [34] some wetland plants species are sensitive to the change of water level and could cause a distinct loss in its species which are caused by oxygen depletion and the ability to go through the photosynthesis process when it is fully submerged. Therefore, depending on the wetland type (near permanent and intermittent) wetlands plants that are frequently flooded can have an adverse effect on the treatment process.

5.2 Does the residence time distribution get affected with the change of wetland depth and flow rate?

According to [35], “A distribution of times that parcels of water spend in a constructed wetland is known as a residence time distribution”. The residence time distribution (RTD) is a tool that has been used widely by engineers to measure wetland’s characteristic that could affect its treatment capabilities. Retention or residence time can be calculated as retention time = (area x depth)/flow per day where flow is the average inflow and outflow of the wetland noting that the velocity profile in subsurface flow wetlands can influence the distribution of the hydraulic residence time. This is due to the velocity profiles that are influenced by the surface friction that potentially slows the water [36]. The RTD can be modeled on a computer and using tracer tests. A tracer test involves a dyed chemical tracer that gets dissolved into the wetland which can easily be detected to measure RTD [35]. Holland et al., [36] has conducted a study on RTD characteristics and how wetland depth and flow rate can influence it. The study was performed by using a series of dye tracers in a small-scale constructed wetland with a constant and controlled flow rates and water levels. Different water levels and flow rates are then used throughout the study to determine its effects on RTD.

Twelve experiments were conducted with different sets of water levels and flow rates, over a period of 13 weeks by adding 15 l of dye to the inlet of the constructed wetland. The average depth of water is 166 and 398 mm for low and high-water levels respectively. The results of this study indicated that there was no significant difference for the RTD values between the high and low flow rates (Table 1). There was however a significant difference for the RTD between the high and low water depth, the low water level result in distribution with one clear peak value whereas the higher water level result in a continuous probability with two different peaks. According to [36], the value differences are reflected in Table 1 for the water levels are significant. The mean RTD spread, o²e, values changed significantly when the water level changed from low to high. However, the change for the RTD centroid of the first moment, λ_t, shows not much of a change in the values.

	Flow rates		Water levels
Parameters	High flows	Low flows	High flows	Low flows
Peak concentration time. λp	0.25	0.22	0.19	0.29
Min. travel time of tracer dye, tm	0.12	0.11	0.15	0.077
RTD centroid of first moment, λt	0.53	0.51	0.49	0.55
Normalized variance of RTD, σ^2θ	0.65	0.55	0.73	0.47

Table 1.

Summarized results for flow rates and water level (using high and low flows) [36].

All values are unitless.

This study concluded that the residence time distribution changed significantly with the change of water levels but not so much with different flow rates. Moreover, a little change in the volume can influence the RTD characteristic in a constructed wetland because the volume is a function of depth. Depth or volume - of water effects the hydraulic efficiency of a constructed wetland and should be considered during the designing of wetlands [36]. Poor designing or sizing of wetland can affect its treatment performance. From this study, it can be stated that with the change of depth and volume in wetlands, the treatment process will be affected therefore the water quality can be assumed to be poor.

6. Modeling of existing wetlands

6.1 Accuracy of modeling

The modeling of stormwater pollutants in runoff and the modeling of Stormwater Quality Treatment (SWQT) assets has developed significantly over the years. The first modeling primarily consisted of relatively simple mathematical equations which were considered somewhat crude, now software programs model many of the complex interactions which occur through the stormwater runoff and treatment process, via user-friendly interfaces [37]. Within Australia, the most widely used SWQT modeling program is MUSIC [38]. MUSIC is a stochastic model which utilizes probability to help determine the pollutants in stormwater runoff and the performance of SWQT assets [39]. Consistent with the rest of Australia, in the Melbourne region MUSIC is also the SWQT modeling software of choice. To approve new SWQT assets and to assess whether a new development is meeting best practices, Melbourne Water and Councils throughout Melbourne require a MUSIC model [40].

However, even though MUSIC models are a requirement by Councils and the major water authority within the Melbourne region, some research suggests that MUSIC models are not completely accurate and may over-treat or under-treat depend on the situation [38, 41]. In their paper Modeling stormwater treatment systems using MUSIC: Accuracy [41] undertook a series of comparisons between existing SWQT assets (located in Australia, Sweden, and New Zealand} and MUSIC models which had been created, with modified parameters to represent the actual conditions (e.g., inflows concentrations) and existing assets. They found that depending on the type of treatment asset, the accuracy of the MUSIC model may vary; in some cases, the MUSIC models overestimated treatment whereas in other cases the modeling underestimated the treatment. It should be noted that the study did not investigate MUSIC’s ability to accurately model wetlands, a topic which requires further examination, however, it does call into question MUSIC’s ability to create accurate models and may provide a portion of the answer to the research question.

The uncertainty of the accuracy of MUSIC SWQT modeling could be due to setting up and calibrating the model. Several studies have found that some of the parameters in MUSIC, e.g., soil storage and field capacity, are crucial for obtaining accurate results and require calibrating based on local data [38, 39]. When creating MUSIC models, errors may occur in the modeling when the user uses the default MUSIC parameters, accidently inputs the incorrect parameters, is unaware of the correct parameters to input, or intentionally inputs the wrong values. To mitigate these potential human errors and improve the quality of MUSIC modeling, Melbourne Water produced MUSIC guidelines, which state the parameters to be used when creating a MUSIC model and general information about SWQT modeling elements. Whilst Melbourne Water’s MUSIC Guidelines provide some recommended parameters, [38, 42] recommend further research into assessing the parameters required in MUSIC for catchments with “similar land use, climatic characteristics and hydrological behavior”.

6.2 Condition of the wetland

The condition of the wetland could contribute to why modeling results may differ from wetlands. Wetland conditions that may influence the performance of the wetland include bad construction, outlet blockages, modification of the terrain by animals, etc. [6, 37]. Models must make some assumptions and they generally assume that assets will function as in intended. However, this is not always the case as there are many operational factors that affect the functioning of a wetland.

6.3 Timing of water sampling

To assess whether a comparison between the modeling and real-life conditions should occur, the first aspect that has to be determined is if the wetland is fully developed. Kadlec and Wallace [37] state that it can take more than 2 years for the wetland to develop fully. This is the duration required for the bio-system to mature, which requires amongst many other things the build-up of a layer of plant detritus over the base of the wetland so that congregations of periphyton and bacteria can form. The congregations of periphyton and bacteria are essential as these organisms’ form part of the nutrient removal process [37]. Thus, if sampling is undertaken before the wetland is fully developed, it may not be reflective of the future potential of the wetland as the bio-system has not matured and is not working to full capacity.

Another aspect that is of importance to the timing of the sampling, is the change of seasons. As plants are seasonal and sprout and perish on an annual basis, the natural biological process dictates that there will be fluctuations in concentrations of nutrients due to the cycling, uptake and release of nutrients, by the plants [37]. Spring generally produces higher uptake of nutrients as the plants are growing and absorb more nutrients in this period whereas in autumn plants are generally dyeing-off and their decaying litter releases nutrients in the waterways [37]. As a result, depending on the timing of the sampling, the same fully developed wetland may produce significant results. However, [12] found that in their two-year study covering all seasons, there was no increase in nitrogen concentrations over the autumn and winter period, which raised the question that there may be other nitrogen removing mechanisms at play. Although, to mitigate this potential error, it is suggested that long-term sampling occurs so that a baseline performance can be determined which takes into account seasonal fluctuations [43].

6.4 Maintenance

One aspect which plays a crucial role in the condition of the wetland is maintenance. Regular maintenance is vital to the performance of a wetland as it facilitates the correct functioning of the wetland [37]. One maintenance task which enables the proper functioning of a wetland, is the cleaning-out of the sediment pond [16]. If the sediment pond fills up beyond its designed depth, the sediment pond may not have the required depth for sediment to settle and therefore the sediment will remain suspended in the water [37]. This suspended sediment may flow into the macrophyte zone and settle, or it may remain suspended and resist treatment. This has various implications to the effectiveness of the wetland, one being that if sediment settles on the base of the macrophyte zone and builds up, the water velocities in the macrophyte zone may increase due to the smaller flow area, which may cause erosion or impact on detention times [37]. Additionally, the makeup of plants in the macrophyte zone may change due to the lack of habitat, e.g., there may be no deep marsh plants, due to sediment changing the makeup of this zone to the shallow marsh.

Another way in which maintenance can impact the performance of a wetland is through the outlet structure. The outlet structure may become blocked due to litter or the natural decay of plant species. This blockage may result in the water level to rise for extended periods of time which can kill off a number of the plant species that inhabit the wetland [37]. Similarly, without regular maintenance of the wetlands, certain plants such as Typha may grow rampant and effect the hydraulic efficiency of the wetland. In doing so it may cause the water level to rise and kill off several plants [37]. To prevent this, it is suggested that scheduled maintenance of the wetland is to occur.

6.5 Frequency of water quality sampling and testing

Another area that may add to the differences between modeling and existing wetlands is in the sampling and testing. There are many aspects that need to be controlled to obtain accurate results. An important factor which must be taken into consideration to obtain an accurate result is frequency of the sampling. In one study it was found that “to sample TSS adequately within a storm event, at least 12 flow-weighted samples were required, and that polluto graphs of seven storm events needed to be sampled within a year to estimate mean annual loads at a reasonable level of accuracy” [43]. Building on this research [43] recommend that to have an error of less than 10% for sediment sampling, sampling must occur every three days or less for TSS–TN and TP will be different.

6.6 Samples’ contamination

Differences between the modeling and actual results may result from the contamination of samples. This could occur in a multitude of ways hence the samplers must be vigilant and follow the protocols when proceeding to take samples. The following are some examples of ways that contamination may occur [44]:

1. Through disturbing the sediment/wetland base when sampling by placing the sample container too deep.

2. Dirty hands or instruments that take the sample may contaminate the water when sampling;

3. Sampling equipment is contaminated either through poor cleaning or coming into contact with other media before sampling.

7. How does government policy influence the sizing of wetlands?

Whilst there are no formal policies requiring wetlands there are policies that promote stormwater quality treatment to meet mandatory sediment and nutrient concentrations defined in statutory policies. The Environmental Protection Agency (EPA) has created the State Environment Protection Policies (SEPP), and of relevance to wetland policy is the SEPP Waters of Victoria [8]. The SEPP Waters of Victoria outline the required concentrations of sediment and nutrients in waterways and larger receiving bodies, such as Port Phillip Bay and Western Port, for waters to be considered healthy. In addition to the main policy, there are various Schedules that provide unique requirements for specific catchments, e.g., SEPP Schedule F6 relates to the water quality requirements of waters for Port Phillip Bay, SEPP Schedule F7 relates specific requirements of the Yarra catchment, etc. These policies are statutory under Section 16 of the Environment Protection Act 1970 [8]. The information provided in the SEPPs indicates what the receiving water’s concentrations of pollutants should be [45]. Whilst this information is useful for providing guidelines for testing it provides little guidance on what stormwater quality treatment is required for new urban developments to maintain these concentrations in the receiving bodies [45]. To try and bridge this gap, the EPA and a panel of stakeholders and experts, which included the Department of Sustainability and Environment, Melbourne Water, Municipal Association of Victoria and local government, were engaged to develop the Best Practice Environmental Management (BPEM) guidelines, which provides a pragmatic methodology for maintaining the concentrations of sediment and nutrients listed in the SEPPs [8, 9].

Rather than produce guidelines that have a focus on concentrations, the BPEM guidelines promote performance objectives that utilize a sediment and nutrient load reduction procedure, and if followed, should maintain the concentrations listed in the SEPPs [9, 45]. Table 2 defines the required load reductions by SWQT assets to meet the BPEM objectives and subsequently meet the SEPP (Waters of Victoria) concentration requirements.

Pollutant	Receiving water objective	Current best practice performance objective
Post-construction phase
Suspended solids (SS)	comply with SEPP (e.g., not exceed the 90th percentile of 80 mg/L)1	80% retention of the typical urban annual load
Total phosphorus (TP)	comply with SEPP (e.g., base flow concentration not to exceed 0.08 mg/L)2	45% retention of the typical urban annual load
Total nitrogen (TN)	comply with SEPP (e.g., bate flow concentration not to exceed 0.9 mg/L)2	45% retention of the typical urban annual load
Litter	comply with SEPP (e g No litter in waterways1	70% reduction of typical urban annual load3
Flows	Maintain flows at pre-urbanization levels	Maintain discharge for the 1.5-year ARI at pre-development levels
Construction phase
Suspended solids	comply with SEPP	Effective treatment of 90% of daily run-off events (e.g., <4 months ARI). Effective treatment equates to a 50%ile SS concentration of 50 mg/L
Litter	comply with SEPP (e.g., No litter in waterways)1	Prevent litter from entering the stormwater system.
Other pollutants	comply with SEPP	Limit the application, generation and migration of toxic substances to the maximum extent practicable

Table 2.

BPEM reduction load targets [9].

¹

An example using SEPP (Waters of Victoria 1988), general surface waters segment.

²

SEPP Schedule F7—Yarra Catchment—urban waterways for the Yarra River main stream.

³

Litter is defined as anthropogenic material larger than five millimeters.

The reduction loads were determined by the Cooperative Research Centre for Catchment Hydrology, in their research publication Best Practice Environmental Management Guidelines for Urban Stormwater. This research publication was based on data analysis within the Background Report to the Environment Protection Authority, Melbourne Water Corporation and the Department of Natural Resources and Environment, Victoria [10]. The primary intention was to create performance objectives that helped achieved the SEPP however, [10] believed that the BPEM performance objectives should be:

• Simple to use,

• Practical and cost-effective

• Prescriptive

• encouraging innovation

• flexible and

• justifiable and defensible (based on sound scientific method)

• “equitable and applicable to all organisations or communities who discharge to urban stormwater”

With these considerations in mind, one of the influencing factors on the performance objectives was land-take. Mudgway [10] found that an asset footprint of approximately 1% of the catchment was sufficient to produce reasonable reductions, e.g., TSS (40–80%) and T.P (35–45%). Hence the performance objectives were created in an attempt to satisfy the SEPP and the ideals above and not necessarily to prescribe treatment for certain flow frequencies. SEPP and BPM have an influence on the sizing of SWQT assets however it is indirectly and not through explicit statements. Most wetlands are sized for the 1 in 3 months flow, however, this appears to be more of a rule of thumb, which can be enforced by local authorities such as Melbourne Water, rather than a statutory requirement [17]. The 1 in 3-month flow is nominally sized flow frequency which has generally been utilized to satisfy the BPEM performance objectives (retention of the typical annual load: TSS = 80%, TP = 40%, TN = 45%) [9]. The primary objective of wetland sizing is to meet the BPEM performance objectives, and it just SQ happens that the 1 in 3-month flow meets this objective.

Despite the fact the BPEM performance targets are the primary drivers for sizing wetlands, local authorities such as Melbourne Water may enforce that wetland be sized for 1 in 3-month flows. This may not necessarily be due to treatment meeting the SEPP concentrations but rather for maintenance reasons, e.g., plant protection [17]. The authority with which Melbourne Water may influence the sizing of the wetland is somewhat convoluted as they receive their authority from the Water Act 1989, and the Environment Protection Act 1988 via the SEPP (Waters of Victoria). Through these acts and policies, Melbourne Water has the authority to dictate the design parameters of wetlands and in their most recent wetlands manual they state “All flows ≤ the peak three-month ARI event is transferred into the macrophyte zone when the EDD in the macrophyte zone is at Natural Water Level (NWL) (Figure 2)” [17].

8. Conclusions and recommendations

Based on the review of literature it can be concluded that:

1. Constructed wetlands are cost-effective treatment systems that can be used to treat urban stormwater runoff.

2. Wetlands plants that are frequently flooded can have an adverse effect on the treatment process.

3. The change of depth and volume of water in constructed wetlands affects the treatment process.

4. It can take more than 2 years for the constructed wetland to fully develop, when the bio-system matures, and form part of the nutrient removal process. Thus, sampling and monitoring before the wetland is fully developed should be taken into consideration to reflect the future potential of the wetland.

5. To have an error of less than 10% for sediment sampling, sampling must occur every three days or less for TSS.

6. Maintenance is important for the ongoing effectiveness of constructed wetlands in water treatment and in flood mitigation.

7. Whilst there are no formal policies requiring constructed wetlands, in Victoria, Australia they are typically installed as part of large-scale urban developments to comply with the BPEM guidelines.

8. Further research into assessing the parameters required in MUSIC for catchments with “similar land use, climatic characteristics and hydrological behavior” is recommended.