The Science of the Total Environment 214 Ž1998. 211]219

The treatment of metals in urban runoff by constructed wetlands L. Scholes a,U , R.B.E. Shutes a , D.M. Revitt a , M. Forshaw b , D. Purchase a a

Urban Pollution Research Centre, Middlesex Uni¨ ersity, Bounds Green Road, London N11 2NQ, UK b En¨ ironment Agency, Thames Region, Kings Meadow House, Reading, Berks, RG1 8DQ, UK

Abstract The use of constructed wetlands for the treatment of domestic wastewater is now well established in the UK and their ability to treat a range of industrial wastewaters is now being investigated. However, their ability to treat urban runoff is relatively untested despite the fact that this application could have important environmental and operational benefits, in both industrial and developing countries. In response to this, the Environment Agency have developed constructed wetland treatment systems at two selected sites in south-east England, both of which receive large volumes of urban runoff. The sites are located at Brentwood and Dagenham and were completed in April 1995. Water and sediment samples have been collected at bi-monthly intervals at each site since October 1995 and analysed for a range of parameters including the total concentrations of six trace metals } cadmium, copper, nickel, chromium, lead and zinc. Similar analysis has been carried out on plants collected from both sites in the spring of 1997. Results show a wide variation in pollutant levels, reflecting the highly variable quality characteristics of urban runoff. Mean removal efficiencies of metals in the water vary between sites in dry weather conditions, with maximum removal efficiencies being recorded at the Dagenham wetland during a storm event. Analysis of plant tissues indicates that the reeds bioaccumulate trace metals and that metal uptake is greatest in the roots. Sediment metal concentrations are typical of a site receiving urban runoff. At both sites the highest sediment concentrations are consistently recorded in samples collected from the settlement tanks. Q 1998 Elsevier Science B.V. Keywords: Runoff; Wetlands; Metals; Environment Agency; Pollutants

1. Introduction The development of buildings, roads and other surfaces using impermeable materials results in the loss of natural water retention provided by

U

Corresponding author.

soils and vegetation ŽMerritt, 1994.. Such urbanisation alters the natural hydrological cycle, changing peak flow characteristics and the volume and quality of the runoff. Traditionally in the UK, development sites have been engineered so that surface water is directly drained to the closest watercourse as quickly as possible to prevent flooding. However, such a system ignores the po-

0048-9697r98r$19.00 Q 1998 Elsevier Science B.V. All rights reserved. PII S0048-9697Ž98.00072-2

212

L. Scholes et al. r The Science of the Total En¨ ironment 214 (1998) 211]219

tential pollutant loads generated from urban runoff and their impacts on receiving waters ŽMungur et al., 1995., hence this form of pollution dilution treatment is now being discouraged. The pollutant load of urban drainage waters tends to be highly variable even within a single catchment area. Concern about the quality and management of urban runoff increased in the 1970s following several studies in the USA which showed that higher pollutant levels are associated with more intensive development and that urban runoff pollutant levels may be comparable to secondary treated wastewater effluent ŽLivingston, 1989.. The quantity of pollutants depends on a variety of factors, such as land use, characteristics of the drainage system and catchment area, the nature and frequency of storms and the weather conditions between storms ŽMerritt, 1994.. The principal pollutants in urban runoff are BOD, suspended solids, heavy metals, hydrocarbons, deicing salts, faecal coliforms and particulate pollution originating from road and vehicle wear. All urban surface runoff generates a pollutant load, but although highways may occupy only 5]8% of a catchment area, they can contribute 50% of suspended solids, 16% of hydrocarbons and 35]75% of heavy metals ŽEllis and Revitt, 1991.. The concept of treating urban runoff is still very new and most studies have focused on the development and use of detention basins for both flood control and removal of pollutants ŽEllis and Revitt, 1991.. However, natural wetlands worldwide are recognised to play an important role in mitigating the effects of severe storms ŽLivingston, 1989.. They can perform a variety of functions including storage of stormwater, reduction of flood flows and velocity, reducing erosion and increasing sedimentation and modifying pollutants. This information, in combination with the success of constructed wetlands in the UK in treating other forms of wastewater ŽCooper et al., 1996. and their introduction to residential developments, resulted in the Environment Agency ŽThames Region. developing constructed wetlands at two selected surface water outfall ŽSWO. sites in outer London, UK.

Constructed wetlands, or reed beds, are natural wastewater treatment systems, providing an efficient, low-cost, easily operated alternative to conventional treatment systems. They are capable of modifying, removing or transforming a variety of water pollutants by a combination of biological, chemical and physical processes, whilst, depending on their area, also providing the wildlife and recreational benefits commonly associated with natural wetland systems. Furthermore, wetlands provide an example of sustainable environmental development and fulfil the objectives of Agenda 21 ŽUNCED, 1992.. 2. Locations and methods 2.1. Site descriptions Both constructed wetlands were constructed during January]April 1995. The first site is at Brentwood, a small town situated to the northeast of London. The site includes a horizontal sub-surface flow system under normal conditions and surface flow system in storm conditions within a constructed wetland and also an adjacent area of natural wetland Žsurface flow system; Fig. 1.. The constructed wetland is built on the site of a flood basin which receives flow from a SWO, draining a 400-ha catchment area before rejoining the River Ingrebourne close to its source. At this point urban runoff constitutes the majority of flow within the River Ingrebourne. Background water analyses carried out by the Environment Agency identified elevated BOD levels Žup to 75 mgrl. and heavy metal concentrations, especially lead and zinc Žmaximum concentrations of 195 m grl and 132 m grl, respectively.. The stormwater discharge initially enters a settlement zone which reduces the water velocity and encourages settlement of suspended solids. Under normal flow conditions water passes through the constructed wetland planted with Phragmites australis. Under storm conditions flow will also pass through the natural wetland, colonised by Typha latifolia, which exerts less hydraulic resistance. The water level is controlled throughout the wetland by stoplogs. Both the

L. Scholes et al. r The Science of the Total En¨ ironment 214 (1998) 211]219

constructed wetland and the natural wetland then discharge into a combined outlet chamber before rejoining the River Ingrebourne. The site at Dagenham is located on the Wantz stream, east London. The Wantz is a small watercourse which receives substantial discharges from the surrounding urban catchment area. Water quality data collected by the Environment Agency prior to the construction of the wetland indicated elevated levels of BOD Žup to 69.4 mgrl. and of heavy metals Žtotal Pb 285 m grl and total Zn 550 m grl.. The watercourse exhibits flashy characteristics during storm events owing to the highly impermeable nature of the catchment area resulting in a visible deterioration in water quality. The wetland is 250 m long and is built in a specifically widened area of the stream. A series of weirs control the flow into three separate beds

213

to prevent hydraulic short-circuiting and to increase aesthetic appeal ŽFig. 2.. In front of the first weir is a settlement zone for the initial removal of suspended solids. The first bed is planted with Typha latifolia followed by two beds planted with Phragmites australis } all at a planting density of 4rm2 . A surface flow system was chosen because a sub-surface flow system would require a large cross-sectional area perpendicular to the flow and there was insufficient land available. A sub-surface flow system would in addition present a physical barrier preventing upstream movement of fish and other aquatic life. 2.2. Sampling programme Twelve sampling sites have been identified, seven at Brentwood and five at Dagenham ŽFigs.

Fig. 1. Diagrammatic representation of Brentwood wetland.

214

L. Scholes et al. r The Science of the Total En¨ ironment 214 (1998) 211]219

Fig. 2. Diagrammatic representation of Dagenham wetland.

1 and 2.. Sites are sampled at bi-monthly intervals, collecting duplicate samples of water and sediment at each sample site. Plants were collected in spring 1997 for tissue metal analysis. This involved collection of Typha latifolia from the inlet of the first bed at Dagenham and collection of Phragmites australis from the inlet of the constructed wetland at Brentwood. Sediment samples were oven dried at 1008C for 24 h, sieved to the fraction - 250 m m and digested with concentrated nitric acid. Duplicate samples were analysed for Pb, Zn, Cu, Cd, Cr and Ni by inductively coupled plasma atomic emission spectroscopy. Plant samples were thoroughly washed with tap water, separated into three parts Žroots, rhizomes, leaves and stems. and dried for 24 h at 1008C. After grinding, they were digested using concentrated nitric acid. Water samples were also treated with concentrated nitric acid and duplicates of both plant and water samples were analysed for Pb, Zn, Cu, Cd, Cr and Ni using inductively coupled plasma atomic emission spectroscopy and graphite furnace atomic absorption spectroscopy, respectively. 3. Results and discussion 3.1. Sediment metal concentrations The results of the sediment metal analyses

Table 1 Ranges of sediment metal concentrations at the Dagenham and Brentwood wetlands, in comparison to unpolluted wetland soils Metals

Dagenham wetland Ž m grg.

Brentwood wetland Ž m grg.

Unpolluted sediment Ž m grg.

Cr Zn Cd Pb Ni Cu

3]167 21]830 3.0]9.6 38]332 23]187 17]178

6]50 39]675 3.3]8.7 41]350 17]147 25]122

7]71 23]50 0.1]2.0 4]40 2]23 4]20

indicate that trace metals are present in varying concentrations, which decrease in the following order Zn ) Pb ) Ni) Cu ) Cr ) Cd. Sediments at the Dagenham wetland tend to contain higher concentrations of Zn, Ni, Cu and Cr compared to the Brentwood site, whereas Pb and Cd are present in similar concentrations at both sites. Table 1 shows the ranges of metal concentrations at each site Žbased on eight sample sets. in comparison to concentrations reported for unpolluted wetland soils ŽKadlec and Knight, 1996.. These results suggest that metals are being actively taken up by wetland substrates. Concentrations are comparable to those reported by other studies of sites receiving urban runoff ŽSchiffer, 1989; Zhang et al., 1990; Mungur et al., 1995..

L. Scholes et al. r The Science of the Total En¨ ironment 214 (1998) 211]219

Maximum concentrations for Zn, Pb, Cu and Cr are consistently recorded in the settlement tanks at both sites Žsettlement tanks are represented by B3 and B4 in Fig. 3 and D2 in Fig. 4., clearly demonstrating both the strong association of these metals with suspended matter and that the settlement of such particles and associated micropollutants is an important process in removing heavy metals from the water column. However, this is not meant to imply that the reeds themselves are unimportant in metal removal, as discussed later. At the Brentwood site there are two settlement tanks } the initial settlement tank ŽB3. and a final settlement tank ŽB4.. Overall the final settlement tank tends to show higher concentrations of metals than the initial tank. This could be due to the direct discharge of the SWO into the initial settlement tank causing the resuspension and mobilisation of some sediments Žand associated pollutants. particularly during storm events. Before reaching the final settlement tank flow passes through either the constructed wetland or the natural wetland, which greatly reduces flow velocity enhancing conditions for the settlement of suspended solids. In addition, due to gravity the largest particles settle out first, with the settlement of progressively finer particles occurring further down the system. As the concentration of pollutants increases with decreasing particle size ŽSalomons and Forstner, 1984., the fact that this settlement tank is located at the end of the system is probably also an important factor. Unlike the sediment concentrations of Zn, Cu, Cr and Pb, concentrations of Ni and Cd appear relatively constant throughout each system suggesting that no individual section is dominating the removal of these metals. A possible explanation for this is that both metals and in particular Cd, are readily leached from sediments, being sorbed and desorbed to substrates throughout the system and therefore demonstrating a highly mobile behaviour. At the Brentwood site, the lowest sediment concentrations of all metals tend to be recorded at the upstream site ŽB1 in Fig. 3. which does not receive any urban runoff, emphasising that runoff is the primary source of metals in the system. At Dagenham the minimum sediment metal concen-

215

Fig. 3. Mean sediment metal concentrations at each sampling point, Brentwood.

trations tend to be from D3 Žat the end of the Typha bed. or D5 Žthe final sample point at the end of the second Phragmites bed.. The settlement tank ŽD2. is obviously removing a significant proportion of the trace metals from the water column, so the concentration of metals available for removal is significantly reduced after this section. However, a further explanation for the relatively lower sediment concentrations at D3 is

Fig. 4. Mean sediment metal concentrations at each sampling point, Dagenham.

216

L. Scholes et al. r The Science of the Total En¨ ironment 214 (1998) 211]219

hydraulic short-circuiting in the Typha bed. Short-circuiting reduces both the treatment area of the wetland actually in use and the overall retention time of the system, both of which greatly reduce treatment efficiency. Similarly the lower concentrations at D5 could be due to metals being removed earlier in the system or be a reflection of initial establishment problems of the Phragmites resulting in a higher flow rate and reducing treatment efficiency in these sections. 3.2. Plant metal concentrations Metal analysis of plant tissues indicates that both species of reed, Typha latifolia and Phragmites australis, bioaccumulate trace metals ŽFig. 5.. Mean concentrations recorded are comparable to those reported by other researchers for Typha and Phragmites in wetlands receiving urban runoff and support several studies which have shown that a range of plants can accumulate metals to high levels ŽZhang et al., 1990; Mungur et al., 1995, in press; Kadlec and Knight, 1996.. Typha and Phragmites were collected from the inlet of the first bed at Dagenham and the constructed wetland at Brentwood, respectively ŽFigs. 1 and 2.. Overall Typha appears to accumulate more Zn, Pb, Cr and Cd than Phragmites, whereas Phragmites tends to contain higher concentrations

of Cu ŽNi concentration was determined in Typha only.. This could indicate that Typha is more efficient in the uptake of trace metals than Phragmites, or may simply be a reflection of the higher sediment and water metal concentrations at the Dagenham site in comparison to the Brentwood site. Similar studies of reeds in both a natural and a constructed wetland receiving urban runoff reported higher metal concentrations in Phragmites ŽMungur et al., 1995, in press.. However, it should be emphasised that this is the first plant sample set to be analysed and therefore there is as yet insufficient data to establish a trend. Analysis of the various sections of the plants Žroots, rhizomes and leaves. found that metal uptake decreases in the order of roots ) leaves ) rhizomes for both species. This is not consistent with other studies which report metal uptake strongly preferential in the order of roots ) rhizomes) leaves ŽKadlec and Knight, 1996.. A possible explanation for this is that the dry weight of the roots and rhizomes, in comparison to that of the stems and leaves, was extremely low, therefore biasing results. Further samples and analysis will clarify this. 3.3. Total metal concentrations in the water Total metal concentrations in the water reflect sediment metal concentrations in that the Dagenham runoff is more contaminated than the BrentTable 2 Total metals concentrations in the water at Dagenham and Brentwood, in comparison to E.C. Water Quality Standards ŽEC Dangerous Substances Directive 76r464rEEC, 1976.

Fig. 5. Mean metal concentrations in roots, rhizomes and leaves of Typha ŽDagenham wetland. and Phragmites ŽBrentwood wetland..

Metal

Dagenham wetland Ž m grl.

Brentwood wetland Ž m grl.

E.C. Water Quality Standards Ž m grl.

Zn Cd Pb Cu Ni Cr

2.5]632 0.3]9.6 1.5]63.1 2.6]75.4 12.9]173 0.5]9.5

3.90]382 0.3]8.9 1.5]92.0 1.6]44.0 3.3]26.0 0.5]4.1

250AT 5AT 125AD 10AD 150AD 200AD

Notes: AT, annual total mean; AD, annual dissolved mean; Thames Region water hardness ; 100 mgrl CaCO 3 .

L. Scholes et al. r The Science of the Total En¨ ironment 214 (1998) 211]219

wood runoff and that metals decrease in a similar order of abundance. In general, metal concentrations do not exceed the European Community Water Quality Standards, although Zn, Cu, Ni and Cd have all done so occasionally. Water metal concentrations for both sites are summarised in Table 2. The concentrations of all the metals show a marked variation at each site. As stated earlier both the volume and pollutant load of urban runoff are heavily influenced by a range of factors, such as the weather, traffic density, road surface, etc., which can result in variations in concentrations and loadings of 1]2 orders of magnitude ŽEllis and Revitt, 1991.. The levels recorded at both sites fall within the range of other values reported for trace metals in urban runoff ŽMaltby et al., 1994; Mikkelsen et al., 1994; Mungur et al., 1995. and are comparable to concentrations of trace metals in runoff from suburban roads ŽHedley and Lockley, 1975.. 3.4. Metal remo¨ al efficiencies Results show that under normal dry weather conditions total removal efficiencies vary both between metals and between sites ŽTables 3 and 4.. The removal of total Ni, Cr, Cd and Zn from the water column is apparent at both sites. The variation of these removal efficiencies between sites could be due to differences in the quality of urban runoff entering the system, which is highly catchment specific. Nonetheless, these removal efficiencies are roughly comparable suggesting similar mechanisms may be responsible for heavy

217

metal removal at each site. Removal efficiencies reported in the literature also show variation. For example, removal efficiencies ranging from y29 to 82% for Zn and 27]94% for Pb are quoted for a wetland basin in a review of stormwater treatment ŽMaltby et al., 1994. and removal rates of 40% for Cr, 87.5% for Cu, 83.3% for Pb, 25% for Ni and 66.7% for Zn in a wetland receiving urban runoff. However, these latter efficiencies are calculated from concentrations rather than loads ŽSchiffer, 1989.. There is a significant difference between sites for the removal of Cu and Pb which are effectively reduced at Brentwood ŽTable 3. but show an overall increase at Dagenham ŽTable 4.. Although this increase appears substantial the removal efficiencies are calculated from low loads and therefore small variations are substantially magnified. However, these differences could also be attributed to differences in both the design and development of the sites. The Brentwood site is composed of both a subsurface flow system and a surface flow system, in comparison to the Dagenham wetland which is a surface flow system only. An important process in removing trace metals from the water column is the binding of trace metals to wetland substrates as insoluble compounds, particularly as sulphides. This process occurs under reducing conditions which dominate in a subsurface flow system, such as that at the Brentwood wetland and this may therefore be a factor in the increased removal efficiency at this site. Another important factor is that the Phragmites

Table 3 Mean inlet concentrations, mean inlet loads and total removal efficiencies for metals at the Brentwood wetland in dry weather conditions

Table 4 Mean inlet concentrations, mean inlet loads and total removal efficiencies for metals at the Dagenham wetland in dry weather conditions

Metals

Mean inlet concentration Ž m grl.

Mean inlet load Žmgrs.

Total removal efficiency Ž%.

Metals

Mean inlet concentration Ž m grl.

Mean inlet load Žmgrs.

Total removal efficiency Ž%.

Zn Cd Pb Cu Ni Cr

21.3 1.8 2.9 11.4 9.8 0.9

0.21 0.1 0.02 0.08 0.07 0.01

13 25 65 68 48 51

Zn Cd Pb Cu Ni Cr

65 3.3 7.7 11.3 70 3.7

1.17 0.06 0.14 0.28 1.52 0.08

13 53 y180 y171 52 43

218

L. Scholes et al. r The Science of the Total En¨ ironment 214 (1998) 211]219

beds at the Dagenham site have suffered various problems since the beginning of construction. Initial problems were due to heavy rain when planting was carried out, resulting in flooding of the beds and therefore many of the Phragmites did not become established. The following spring plants appeared to be recolonising across both Phragmites beds and it was hoped that this natural growth would mitigate the need for replanting. However, in early summer, further damage to Phragmites was caused by grazing animals. This has significantly reduced both the size of the wetland and development of the litter layer Žwhich is particularly important for the removal of Cu ŽKadlec and Knight, 1996.. and therefore reduces overall treatment performance. The reported data cover a monitoring period of 1 year during which one heavy rainfall event Ž1 May 1996. was sampled. A comparison of metal concentrations and loads during dry weather conditions and a storm event at Dagenham clearly demonstrates the increased pollutant levels associated with a storm event ŽTables 4 and 5.. Total removal efficiencies of all the metals, except for Ni, are greatest during a storm event. This supports studies reported in the literature which found removal efficiencies increased with increasing inlet concentration ŽKadlec and Knight, 1996; Mungur et al., in press.. Nickel concentrations were greater during normal dry weather conditions and as expected from the above, this corresponds to higher Ni removal efficiencies ŽTable 4.. However, it should also be emphasised that the sampling was carried out as part of a routine sample collection rather than the monitoring of the actual storm event and therefore these removal efficiencies do not reflect changes within a ‘plug-flow’ but are more a general indication of metal removal by a wetland during a storm event. 4. Conclusion The first year of a 2-year sampling programme indicates that a range of trace metals can be removed from the water column by wetland processes. However, the clear variation between some of the trace metal removal efficiencies reflects the poor understanding of how these

Table 5 Inlet concentrations, inlet loads and removal efficiencies for metals at the Dagenham wetland during a storm event Metals

Inlet concentration Ž m grl.

Inlet load Žmgrs.

Removal efficiency Ž%.

Zn Cd Pb Cu Ni Cr

632 9.6 31.9 35 31 5.4

15.8 0.2 0.8 0.9 0.8 0.1

100 100 79 92 17 99

processes operate and the interaction of such processes with other variables. Although both wetlands are treating runoff from similar urban environments, this runoff by definition is highly variable, affecting both treatment of the wastewaters and development of the receiving system. This variability also leads to difficulties in the design of constructed wetlands treating urban runoff in that, in contrast to the treatment of domestic sewage, there is no ‘characteristic’ effluent or flow on which to base the design. This study also demonstrates the value of full-scale field systems in highlighting both the potential benefits and problems associated with the development and adaptation of an existing technology to a new application. References Cooper PF, Job GD, Green MB, Shutes RBE. Reed beds and constructed wetlands for wastewater treatment. Medmenham, UK: Water Research Centre Publications, 1996. EC Dangerous Substances Directive 76r464rEEC, 1976. Ellis JB, Revitt DM. Drainage From roads: control and treatment of highway runoff in report NRA 43804rMID.012, commissioned by Thames NRA, Technical Services Administration, 1991. Hedley G, Lockley JC. Quality of water discharged from an urban motorway. J Water Pollut Control 1975;74:659]674. Kadlec RH, Knight RL. Treatment wetlands. Boca Raton: Lewis, 1996. Livingston EH. Use of wetlands for urban storm water management. In: Hammer DA, editor. Constructed wetlands for wastewater treatment } municipal. Michigan: Industrial and Agricultural Lewis, 1989:253]264. Merritt A. Wetlands, industry and wildlife } A manual of principles and practices. The Wildfowl and Wetlands Trust. Newcastle upon Tyne: Hindson Print Limited, 1994.

L. Scholes et al. r The Science of the Total En¨ ironment 214 (1998) 211]219 Maltby L, Forrow D, Boxall A, Calow P, Clifford C. The effects of motorway runoff on freshwater ecosystems: 1. Field study. Environ Toxicol Chem 1994;14:1079]1092. Mikkelsen PS, Weyer G, Berry C, et al. Pollution from urban stormwater infiltration. Water Sci Technol 1994;29:293]302. Mungur AS, Shutes RBE, Revitt DM, House MA. An assessment of highway runoff treatment by a natural wetland. Water Sci Technol 1995;32:169]175. Mungur AS, Shutes RBE, Revitt DM, House MA, Fallon C. A constructed wetland for the treatment of urban runoff. Proceedings of the International Conference on the Remediation and Management of Degraded Lands, Hong Kong, 2]5 December 1996, Žin press..

219

Salomons W, Forstner U. Metals in the hydrocycle. Berlin: Springer-Verlag, 1984. Schiffer DM. Water-quality variability in a central Florida wetland receiving highway runoff. In: Davis FE, editor. Bethseda, MD: Water: Laws and Management, 1989. UNCED, Agenda 21. United Nations Conference on the Environment and Development. Switzerland: Conches, 1992. Zhang TT, Ellis JB, Revitt DM, Shutes RBE. Metal uptake and associated pollution control by Typha latifolia in urban wetlands. In: Cooper PF, Findlater BC, editors. Constructed wetlands in water pollution control. Oxford: Pergammon Press, 1990:451]459.