A multi-functional and multi-compartment constructed wetland to support urban waterway restoration

Urban stormwater nutrient is considered as an important nuisance for the introduction of non-point source pollutants, both organic and inorganic, into downstream receiving waterways (Pitt et al. 1995). In Western Australia (WA), urban drainage contributes around 160 tonnes of nitrogen (N) and 16 tonnes of phosphorus (P) to the Swan-Canning River system annually (SRT 2008). Water regulatory organisations and their partners have invested substantial resources to protect river water quality as part of Healthy Rivers Action Plan, and then to achieve water sensitive and livable cities under the Water Sensitive Urban Design (WSUD) framework. One of the WSUD elements is the implementation of constructed wetlands (CWs) before a watercourse to retain and treat stormwater nutrient. CWs are engineered green bioreactors (Mitsch & Jorgensen 2003) commonly employed in Australia for stormwater management (Lloyd 2001). CWs can manage, treat and transform stormwater nutrient by utilizing natural processes that take place within wetland standing water, soils/sediments, filter media, aquatic macrophytes and microbial assemblages (Mitsch & Gosselink 2007; Vymazal 2007). CWs can be different types, but surface flow (SF) and subsurface flow (SSF) systems are the most commonly used. Their treatment capacity can change over time through feedback loops like design modification, restoration, etc.

The focus of this work was on the Wharf Street Constructed Wetland (WSCW) that is an approximately 1 ha hybrid CW with three vegetated SF and two laterite-based SSF compartments. The WSCW receives seasonal stormwater from a 129 ha urban catchment before entering to the downstream Canning River. The paper described the attenuation of stormwater nutrient in the WSCW during different hydrological regimes and design settings, particularly how the attenuation differs within SF and SSF compartments. This study also aimed to investigate the potential scope of non-market services from the WSCW to improve urban liveability and local amenity.

The WSCW (Figure 1) was initially constructed in Cannington, WA in 2009 and laterite-based SSF compartments were incorporated in 2012. For simplification and comparison, we divided the WSCW into four different compartments (Adyel et al. 2016): (a) compartment 1: SF1 with inlet (W1) and outlet (W2); (b) compartment 2: laterite-based SSF1 and associated inlet (W3) and outlet (W4); (c) compartment 3: laterite-based SSF2 and associated inlet (W5) and outlet (W6); and (d) compartment 4: SF2 with inlet (W7) and outlet (W8). Each compartment was planted with aquatic macrophytes.

The Department of Water WA (DoW; now the Department of Water and Environmental Regulation) monitored water discharge and level at the WSCW inflow and outflow stations at 5-min intervals. A structure comprising a rock riffle (primary flow control) with a set of three by-pass pipes (under the riffle) and a hydrostatic water level probe to provide water stage information was used to capture inflow data. A concrete V-notch weir using a water level transducer in a float well system was used for outflow measurement. A detailed water balance of the WSCW has been previously described elsewhere (Adyel et al. 2016).

Stormwater nutrient data such as ammonium (NH₃), filtered total oxidised nitrogen (NO_x)_, total Kjeldahl nitrogen (TKN), dissolved organic nitrogen (DON), unfiltered total nitrogen (TN), filtered reactive phosphorus (FRP) and unfiltered total phosphorus (TP) at different sampling points of the WSCW were gathered from the DoW, who sampled fortnightly or monthly from 2009 onwards, particularly during base flow and high flow conditions. During storm event sampling, the DoW collected water samples only at main inlet (W1) and outlet (W8) at 2–3 h intervals. Additional water samples were collected from 2013 onwards specifically for this research, and analysed using the Lachat Quick Chem procedure (APHA 2012) as described elsewhere (Adyel et al. 2016).

The site experiences strong seasonal hydrology, and hydrological variability shapes nutrient loads attenuation. The site experiences prolonged low flows during the dry summer and episodic high flows during the wet winter. Therefore, stormwater can be stored within different compartments during the summer and the CW becomes a lentic system. Therefore, a recirculated system pumps water from the main outlet (W8) to the main inlet (W1) to make sure a minimum flow is maintained throughout the year for aesthetic reasons. Interestingly, during the high flow conditions the system was designed in such way that around 10% of water travels from the SF1 to the SSF2 compartment, by-passing the SSF1 compartment, which might reduce the attenuation capacity of the overall system.

The site was designed to promote higher attenuation by extending the retention time during low to medium flow conditions. Higher flow rates showed increased attenuation of particulate nutrients, probably due to dilution. Overall, the WSCW attenuated up to 96% NH₃, 90% NOx and 88% TN during typical dry weather conditions (DWCs) or low flow (when flow rate ≤0.005 m³/s) (Figure 2). SDC for DON was low (up to 15%) compared to other parameters. Average attenuation of FRP and TP in the system was 58% and 65%, respectively (Figure 2).

During DWC, the WSCW captured about 1 to 55% and 10 to 99% TN and TP load, respectively, preventing it from being delivered to the downstream waterway. However, the load attenuation for TN and TP was 54 to 79% and 27 to 68%, respectively during the high flow conditions (when flow rate >0.005 m³/s). We also assessed how the WSCW attenuates nutrient during episodic storm events, using high-resolution sampling. During six studied storm events, the WSCW retained about 41% of TN and 66% of TP loads. Over the last six years, the WSCW attenuated about 45% of TN and 65% of TP loads from being delivering to the Canning River. The performance was even better than other typical CWs (Collins et al. 2010; Vymazal 2011). SF and SSF compartments showed higher attenuation of N and P-based nutrients, respectively (Figure 3).

The relative extent of nutrient attenuation varied in the different compartments, and alternative SF and SSF compartments made the whole system more effective for multiple nutrient species than a single stage CW that might effective for a particular nutrient (Malaviya & Singh 2012). The WSCW showed higher N attenuation that other CWs that contained an initial or more SSF systems (Vymazal 2013). Nutrient attenuation changed in compartments due to variation in design and surface area, presence of macrophytes and filter media, available aerobic/anaerobic conditions, seasonal pattern and interaction with light and air. Laterite in the SSF compartments are potentially known for P adsorption via ligand exchange reactions, where phosphate displaces water or hydroxyls from the surface of iron (Fe) and aluminum (Al) hydrous oxides (Wood & McAtamney 1996). Interestingly, the addition of Fe to the substrate improved P retention significantly (from 55 to 66%) in CW of semi-arid condition (Cerezo et al. 2001). We observed higher P retention when Al or Fe attenuation was higher in the WSCW, with co-precipitation probably being the cause. The WSCW is a large-scale setting in an urban catchment and is exposed to different hydro-metrological conditions. The diversion of Bebington Court Drain's water to the inlet of the SSF2 compartment could improve overall nutrient attenuation as this water will pass through one SF and one SSF compartment over a longer time period. We can expect more benefit by incorporating a laterite filter at the main outlet.

Sediments absorbed one-third of the incoming nutrient loads, but occasionally acted as a source at the SF compartments. Sedimentation area determined the nutrient pools as the larger the sedimentation area, the larger the nutrients pools. The net accumulation of sediment nutrients over time (t = days since 29 June 2010) was TN = 0.0027t²-3.2448t + 2397 (R² = 0.39); TP = 0.0005t²-0.8548t + 478.59 (R² = 0.56) and TOC = 0.0314t²-40.446t + 24649 (R² = 0.55) (Adyel et al. 2016). Sediments accumulated about one-third of incoming TN and TP loads. The Al and Fe levels increased in the WSCW sediment along with P pools, suggesting that P co-precipitated with Al/Fe minerals (Vymazal 2007). Sediments released nutrient loads during abrupt flow conditions and seasonal drying-wetting of the riparian zone. Nutrient flux from sediments therefore needs to be estimated for the optimisation of the system.

Aquatic macrophytes such as B. articulata, B. rubiginosa, B. preissii and S. validus accumulated about a quarter of incoming nutrient loads. Below-ground biomass (roots and rhizospheres) accumulated about 2–3 times more nutrient mass than that in above-ground biomass (shoots and leaves). Nutrient accumulation was higher during early to mature growth stages, particularly for the first two years of operation. Macrophytes in the open water bodies or SF compartments showed higher nutrient pools than those found in temperate climates (Vymazal 2011; Vymazal 2013). However, macrophytes showed seasonal senescence that can return nutrients to the system. Moreover, the presence of un-sampled macrophytes and the absence of an up to date species map make the estimation of macrophyte nutrient pools challenging. Periodic species coverage mapping/counting is essential for better estimation of nutrient pools of macrophytes, while harvesting mature macrophytes prior to summer senescence is required to ensure permanent removal of nutrients from the CW.

The WSCW delivered non-market ecosystem services; providing a recreational amenity to the public as well as a habitat for native species. Careful design of this site not only considered the nutrient attenuation, but also the equally important need for it to integrate with residential areas, the regional nature park and downstream sensitive waterways, providing a passive eco-recreation and educational asset consolidated by safety, amenity and accessibility (Figure 4). The site also enhanced the local habitat by returning endemic species and habitat types for flora and fauna. This site experienced mosquito problems (Russell 1999) before establishment of the CW and subsequent CW design ensured a water velocity or enough recirculation to discourage development of mosquito larvae. The WSCW landscape and green infrastructure design concept can be implemented in other areas to enhance urban livability and ecological sustainability. However, a careful quantification of the non-market value of the ecological services provided by the WSCW is required to compare it with other man-made ecosystems.

The WSCW site experienced strong seasonal hydrology and hydrological variability that shaped nutrient load attenuation. The relative extent of nutrient attenuation differed in the different compartments. Sediments stored about one-third of the incoming nutrient loads, but however occasionally acted as source, particularly in the SF compartments. Macrophytes accumulated about a quarter of the incoming nutrient loads, while seasonal macrophytes senescence may return some nutrient to the system. The overall concept of this CW design and implementation can show the pathways of protecting downstream waterways, ensuring urban liveability and providing other non-market benefits.

This study was funded by the Cooperative Research Centre for Water Sensitive Cities under Project C4.1 ‘Multi-functional urban water systems’ and Project B2.4 ‘Hydrology and nutrient transport processes in groundwater/surface water systems’. The first author was supported by a Scholarship for International Research Fees (SIRF-UWA). The authors wish to acknowledge Peter Adkins and Christie Atkinson (Department of Parks and Wildlife, WA) for their continuous assistance with data, site information and comments. Authors also acknowledge Carlos Ocampo, Hasnein Tareque, Gayan Gunaratne and Taj Sarker for their assistance during the fieldwork, experiments and data analysis.