The functioning of a wetland system used for polishing effluent from a sewage treatment plant

Ecological Engineering 25 (2005) 101–124

The functioning of a wetland system used for polishing effluent from a sewage treatment plant Sylvia Toet a,∗ , Richard S.P. Van Logtestijn a , Michiel Schreijer b , Ruud Kampf b , Jos T.A. Verhoeven a a

Department of Geobiology, Utrecht University, Sorbonnelaan 16, 3584 CA Utrecht, The Netherlands b Waterboard Hollands Noorderkwartier, P.O. Box 850, 1440 AW Purmerend, The Netherlands Received 14 October 2004; received in revised form 24 February 2005; accepted 10 March 2005

Abstract A surface-flow wetland system designed for polishing effluent from a sewage treatment plant (STP) on the island of Texel, The Netherlands, was studied between April 1996 and March 1997. The wetland system was composed of a sequence of several units with different water depth, hydraulic retention time and vegetation. The system had a relatively short hydraulic retention time of 2.4 days (hydraulic loading rate 25 cm day−1 ). The wetland system showed 92% removal of faecal coliforms (3.7 × 1010 cfu m−2 yr−1 ), a 26% reduction of nitrogen (126 g N m−2 yr−1 ) and less than 5% reduction of phosphorus (5 g P m−2 yr−1 ). The oxygen concentration, which was less than 3 mg l−1 in the STP effluent, showed a strong increase during passage through the wetland all year with a clear diurnal shift between 1 and 12 mg l−1 in summer. Turbidity of the surface water doubled, but the suspended solids changed from sewage sludge particles at the beginning of the system to microscopic biota characteristic for a wetland at the end. A presettling basin produced substantial reductions of faecal coliforms (11 × 1010 cfu m−2 yr−1 ) and also intercepted incidental peaks in organic N and P load. N removal was highest in the shallow front sections of the subsequent parallel ditches (240 g N m−2 yr−1 ), largely owing to denitrification. These ditch sections contained Phragmites australis or Typha latifolia. The increase of the oxygen dynamics predominantly occurred in the rear, deeper sections of the parallel ditches, due to the presence of submerged macrophytes, macro-algae and periphyton. The treatment of the wetland resulted in water with acceptable faecal coliform concentrations that closely resembled the quality of the receiving surface water. However, the removal of nutrients was insufficient to meet the criteria for good ecological quality, probably due to the short hydraulic retention time. © 2005 Elsevier B.V. All rights reserved. Keywords: Treatment wetland; Tertiary effluent; Mass budgets; Nitrogen; Phosphorus; Faecal coliforms; Turbidity; Oxygen

∗ Corresponding author at: Environment Department, University of York, Heslington, York Y010 5DD, UK. Tel.: +44 1904 434327; fax: +44 1904 432998. E-mail addresses: [email protected] (S. Toet), [email protected] (J.T.A. Verhoeven).

0925-8574/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.ecoleng.2005.03.004

102

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

1. Introduction The use of constructed wetlands for wastewater treatment has proven to be a low-cost, sustainable, and effective alternative for conventional wastewater treatment technologies (Moshiri, 1993; Kadlec and Knight, 1996; Vymazal et al., 1998). Removal efficiencies beyond 70% have generally been obtained in these wetlands for biochemical oxygen demand (BOD), chemical oxygen demand (COD), total suspended solids (TSS) and faecal coliforms. Nitrogen and P removal efficiencies have been more variable, because they are highly dependent on hydraulic and nutrient mass loading rates, and wetland design (Bastian and Hammer, 1993; Knight et al., 1993; Kadlec and Knight, 1996). Studies of treatment wetlands have dealt with a wide range of wastewater sources (Moshiri, 1993; Kadlec and Knight, 1996; Vymazal et al., 1998). Municipal wastewater with only a limited pre-treatment has been the most common type of water supplied to these systems. Much less experience exists on the treatment of tertiary effluent from sewage treatment plants (STP; Knight et al., 1987; Jackson, 1989; Kadlec and Knight, 1996; Sundblad, 1998; Lund, 1999). This effluent is characterised by relatively low COD and BOD contents and moderate N and P concentrations, which may still be too high for discharge to surface waters sensitive to eutrophication. This oxygen-poor water can probably best be polished in surface-flow wetlands (Kadlec et al., 2000). In such cases, the objectives for the water managers may not primarily be the reduction in nutrients, but rather further faecal coliform reduction and the buffering of concentration peaks in periods of very high discharge. Further, the aim is often to improve the ecological quality of the effluent to the point that it resembles that of the receiving surface waters as closely as possible. This implies that the load of sewage sludge particles should be further reduced and replaced by biota such as phytoplankton and detritus and that oxygen concentrations should show strong diurnal fluctuations with high peaks during the day. This study describes the performance of such a surface-flow wetland system supplied with tertiary STP effluent at a relatively high hydraulic loading rate of 25 cm day−1 , resulting in a hydraulic retention time of only 2.4 days. The wetland system was constructed

with the aim of improving the water quality of the STP effluent sufficiently to approach that of the receiving surface water on the island of Texel, The Netherlands. Considering the fact that the STP effluent had already been chemically dephosphatised, the rather short hydraulic retention time was expected to be sufficient to reach the goals set, i.e. a small additional COD, N and P removal, considerable faecal coliform reduction and oxygen enhancement. A pilot study with one small ditch had further supported these expectations (Schreijer et al., 1997). The wetland was composed of several serial units with different water depth, hydraulic retention time and vegetation. The STP effluent first flowed into a deep presettling basin for sedimentation of residual sludge and for interception of incidental peak loadings. The outflow from this basin was then divided over nine parallel ditches with emergent macrophytes (Phragmites australis or Typha latifolia) in the first half and submerged macrophytes and macro-algae in the second, deeper half of the ditches. The design of these ditches was aiming for further removal of faecal coliforms, COD, N and P (sections with emergent macrophytes) and for enhancing the oxygen content and biota composition of the water (sections with submerged macrophytes). Between April 1996 and March 1997, the performance of this wetland system was studied with respect to changes in turbidity, COD, faecal coliforms, nutrients and oxygen dynamics of the water flowing through the system. The main research questions were: (1) whether COD, faecal coliforms, N and P levels were sufficiently reduced by the system as a whole, (2) whether the diurnal oxygen dynamics were sufficiently enhanced to mimic the normal behaviour of surface water, (3) whether the various compartments (presettling basin, ditch sections with emergent macrophytes and submerged macrophytes) performed as expected by the designers, (4) what implications the results would have for the optimisation of the performance of this wetland. Estimates of pollutant removal efficiencies of the wetland system as a whole and of the different wetland compartments were based on pollutant mass budgets of the water inputs and outputs. Several processes involved in N and P removal were also measured directly and the total removal quantified by summation of these processes was compared to the removal calculated on the basis of the mass budgets.

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

103

Table 1 Dimensions (summed values for nine parallel ditches) and mean annual hydraulic retention time (HRT) and hydraulic loading rate (HLR) of the wetland sections Wetland section Presettling basin Front parallel ditch sections Rear parallel ditch sections Discharge ditch Total wetland

Volume (m3 )

HRT (days)

HLR (cm day−1 )

3483 4338 4464 839

4603 801 1818 427

1.4 0.3 0.6 0.1

94 73 71 377

13124

7650

2.4

25

Area (m2 )

2. Methods 2.1. Site description This study was carried out in a treatment wetland system on the island of Texel in The Netherlands (53◦ 05 N, 4◦ 47 E). The design of the wetland was based on the results of a pilot-scale wetland system that had received part of the STP effluent (Schreijer et al., 1997). The full-scale system was constructed in 1994 to polish on average 3400 m3 day−1 of effluent from the STP ‘Everstekoog’ (up to 45,000 p.e.). The activated sludge plant treated mostly domestic sewage and stormwater. The wastewater received tertiary treatment including chemical P removal with FeSO4 . The STP effluent was discharged to the surface water of the

island where it served to diminish the shortage of freshwater for agricultural purposes. The small volume of receiving water bodies resulted in a high impact of the STP effluent on the surface water quality and quantity. Low oxygen concentrations in the surface water were observed at a distance of 1.5–2 km from the discharge point of the STP (Schreijer et al., 1997). The conservation of valuable flora and fauna on the island therefore required further improvement of the STP effluent. The surface-flow wetland system had a total water surface of 1.3 ha (Table 1). The STP effluent first entered a presettling pond, was then divided over nine parallel ditches, after which it was collected in a discharge ditch and supplied to the surface water of the island (Fig. 1). The first half (i.e. near the ditch inlet) of eight ditches was 0.2 m deep and contained Phragmites

Fig. 1. Overview of the wetland system on the island of Texel, the Netherlands. The vegetation types in the front and rear section of the parallel ditches are indicated.

104

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

australis or Typha latifolia, while the second half was 0.4 m deep and contained submerged aquatic macrophytes (mainly Elodea nuttallii, Ceratophyllum demersum and Potamogeton spp.). One ditch without vascular macrophytes served as a control. Dimensions and mean annual hydraulic retention times and hydraulic loading rates of the wetland compartments are shown in Table 1. 2.2. Water budgets Water budgets were compiled for the presettling basin, the front and rear sections of the ditches and the discharge ditch between 1 April 1996 and 31 March 1997. The inputs to the wetland compartments were STP effluent and precipitation, and the outputs were surface water outflow, evapotranspiration and groundwater recharge. The budget over the entire year was based on monthly sums or averages of these inputs and outputs. Net changes in storage in the wetland compartments approached zero on a monthly and annual basis. 2.2.1. Water inflow and outflow The STP effluent was pumped into the presettling basin through a pipe. Daily inflow rates to the presettling basin were calculated from measurements with a velocity meter and the diameter of the pipe. At the other side of the basin, the outflow was divided over the nine parallel ditches by rectangular weirs and supplied to the ditches through culverts. Weirs and culverts also connected the ditches with the discharge ditch, and the discharge ditch with the surface water of the island. The height of the water column above the weirs was monitored continuously with two pressure gauges in the presettling basin. A clear diurnal pattern of the height measurements was observed, which depended on the diurnal supply of sewage to the STP. The pressure gauge measurements were stored in data loggers. The results for the two pressure gauges were averaged, because only small differences were observed. The averages of the height measurements of the water column above the weirs and the width of the rectangular weirs were used to calculate the inflow rate to each of the ditches (Haan et al., 1994). The more accurate daily flow measurements of the velocity meter in the pipe supplying STP effluent to the presettling basin in com-

bination with the other water fluxes in this basin were used to correct, if necessary, the daily inflow rates to the ditches. The outflow rates of the parallel ditch sections and the discharge ditch were calculated as the difference between the inputs and outputs of the water budgets, thus no error term of the water budgets was revealed. No pressure gauge measurements were obtained from November through March, because the pressure gauges had to be removed from the ditches due to long, severe periods of frost. STP effluent was still supplied to the system during these cold periods, when it flowed underneath the ice cover in the wetland compartments. The distribution of water over the nine ditches showed only relatively small fluctuations in time between April and October. Daily inflow rates to the ditches in November through March were therefore inferred from velocity meter data to the wetland system during this period and from the mean distribution over the nine ditches for the rest of the year. 2.2.2. Precipitation and evapotranspiration Daily measurements of precipitation were obtained from the Royal Netherlands Meteorological Institute (KNMI) weather station Den Burg, 7 km southeast of the wetland system. Daily Makkink’s reference crop evapotranspiration Er (Makkink, 1957, 1960) from the KNMI weather station De Kooy, 3 km southwest of the wetland system in combination with the regression equation from Koerselman and Beltman (1988) were used to determine the evapotranspiration in ditch sections with Phragmites australis or Typha latifolia. Evaporation from free water surfaces (E0 ) was calculated by multiplying Er with 1.25 (R. Jilderda, KNMI, personal communication). In the ditch sections with emergent macrophytes, open water evaporation was assumed between the mowing of the shoots in December until the emergence of shoots in April. Evapotranspiration in Typha stands was corrected for the presence of unvegetated areas during spring and early summer. No evapotranspiration was assumed during frost periods with ice cover on the wetland compartments. 2.2.3. Groundwater recharge Piezometers with a filter at 1.5 m sediment depth were installed at the midpoint along three ditches (numbers 1, 5 and 9), in the presettling basin and

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

in the discharge ditch. Hydraulic head was measured each month. Groundwater flows were calculated using Darcy’s law (Freeze and Cherry, 1979). Groundwater recharge in the ditches without piezometers was estimated by the mean value of the other three ditches. The hydraulic conductivity (k) of the saturated sediment was determined with the piezometer method (Luthin and Kirkham, 1949; Boersma, 1965; Youngs, 1968). All parallel ditches had fine sandy-silty sediments with k = 15.8 mm day−1 , whereas k was slightly different in the presettling basin and discharge ditch (23.4 mm day−1 ). 2.3. Water quality Turbidity, COD, faecal coliforms, NH4 -N, NO3 -N, Kjeldahl N, PO4 -P and total P concentrations (NEN standards, Dutch Standardization Institute, Delft, The Netherlands) of the through-flowing water were determined biweekly by taking grab water samples at the inlets and outlets of all the wetland compartments. Flow-proportional samples were collected with an automated sampler at the inlet and outlet of the wetland system as a whole. Biochemical oxygen demand (BOD5 ) and TSS were also measured, but were frequently below the detection limit of the methods used (5 and 3 mg l−1 , respectively). Nitrite was measured regularly at the inlet and at several locations within the wetland, but it was always less than 0.25 mg l−1 . Monthly removal efficiencies of pollutants were determined on mass budgets of the water inputs and outputs (REmass ). Pollutant mass input and output rates were calculated for each wetland compartment as the monthly averages of the pollutant concentration at the inlet and outlet (Cin and Cout , respectively) times the water inflow or outflow (Qin or Qout ) during the corresponding month. The mass input to the wetland compartments also included atmospheric deposition (D, see nutrient budgets).
 (Cin × Qin + D) − Cout × Qout REmass = Cin × Qin + D ×100 (%) The monthly pollutant concentrations, mass input and output rates, and removal efficiencies were averaged for the replicate front ditch sections with Phragmites australis and for the adjacent rear ditch sections, as

105

well as for the replicate front ditch sections with Typha latifolia and adjacent rear sections. Summation of the monthly pollutant mass input and output rates yielded the annual values, which were used to calculate the annual removal efficiencies of pollutants. Oxygen concentration of the through-flowing water was determined continuously with dissolved oxygen meters (WTW Trioxmatic 161T) at four locations between 15 April and 18 December 1996. Three of the locations were in ditch 5, i.e., at the input, the end of the front section with Phragmites australis, and the end of the rear section. The fourth location was at the end of the wetland system as a whole. Photosynthetically active radiation (PAR) was measured continuously with a LI-COR quantum meter (LI-190SB) in the air of a non-shaded part of the wetland during the same period. 2.4. Nutrient budgets Annual N and P mass budgets were compiled for the presettling basin, the front ditch sections, the rear ditch sections and the entire wetland system. Data for the replicate front ditch sections with emergent macrophytes and their rear ditch sections were averaged. Nutrient inputs to the wetland and its compartments were STP effluent input and atmospheric deposition, and the outputs were surface water output, groundwater recharge, denitrification, harvest of emergent macrophytes and accumulation in the sediment. The error term was calculated as the difference between the inputs and outputs. 2.4.1. Atmospheric deposition Bulk deposition was collected with two funnels (inner diameter = 33.5 cm) and stored in two jars at 4 ◦ C. Samples were collected biweekly and analysed for NH4 -N, NO3 -N and PO4 -P concentrations (NEN standards, Dutch Standardization Institute, Delft, The Netherlands). Bulk deposition of these compounds on the wetland compartments was calculated by multiplication of the total volume of precipitation on a compartment (data from KNMI weather station Den Burg) and the average concentrations. Total atmospheric NH4 -N and NO3 -N inputs on areas with emergent macrophytes were obtained by multiplying the bulk deposition (g N per m2 ditch area) by 2.7 during the growing season

106

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

based on standard ratios between bulk deposition and total atmospheric N deposition in the region (Bobbink et al., 1992; Erisman, 1992; Ridder et al., 1984). Dry deposition of ortho-phosphate was assumed to be negligible (R. Bobbink, personal communication). 2.4.2. Groundwater recharge Monthly retention rates of N and P due to groundwater recharge in the wetland compartments were estimated from the means of the monthly input and output nutrient concentrations of each wetland compartment and the amount of infiltrated STP effluent during the corresponding month. 2.4.3. Denitrification Since no denitrification data were available for the time period when this study was conducted, denitrification rates were estimated from measurements made in the same wetland system 1 year later. These measurements took place in a ditch dominated by Phragmites australis and Elodea nuttallii between May 1997 and February 1998 (Toet et al., 2003). The results showed that the periphyton on plant shoots was always the major system component in which denitrification took place. Denitrification in Phragmites shoots was linearly related to the chlorophyll-a content of the periphyton. Periphyton chlorophyll-a contents were measured during the time period of this study at five sampling dates in two ditch sections with Phragmites australis and two with Typha latifolia. Denitrification rates of periphyton on Phragmites and Typha shoots were estimated on the basis of these data by using regression (linear, quadratic or cubic, p < 0.05) and interpolation. The relatively high cover of Elodea nuttallii in four rear ditch sections in 1996 was similar to that in the rear ditch section used for denitrification measurements in 1997. The same denitrification rate (per ditch area) was therefore assumed in these rear ditch sections in 1997. The four other rear ditch sections contained very little Elodea nuttallii or other submerged macrophytes in 1996. The denitrification rate of the sediment was shown to be very low when the nitrate concentrations in the through-flowing water were below 4 mg l−1 but to reach high values in periods with nitrate concentrations higher than that value (Toet et al., 2003). This was taken

into account in the calculations of the total denitrification of the sediment. Denitrification rates estimated in this way by extrapolating many data to other spatial and temporal scales only give an order of magnitude of the process in the wetland system. 2.4.4. Harvest of emergent macrophytes Shoots of Phragmites australis and Typha latifolia were cut 0.15 m above the water level on 9 December 1996. The amounts of N and P removed by harvesting were determined. Shoot density was measured in four 1 m × 1 m permanent plots, equally spaced over the length of each ditch section. A sample of 12 shoots was collected near each plot. The samples were dried for 2 days at 80 ◦ C and at least 3 days at 70 ◦ C to determine dry weight. Ground samples were analysed for N and P with a salicylic acid thiosulphate modification of the Kjeldahl method (Bremner and Mulvaney, 1982). Nitrogen and P concentrations of the diluted digests were determined colorimetrically on a continuous flow analyser (Skalar SA-40) by using the indophenol blue method with salicylate and the ammonium molybdate method, respectively. 2.4.5. Accumulation in the sediment Annual storage of N and P in the sediment was determined in the presettling basin, ditches with Phragmites australis and Typha latifolia, the control ditch and the discharge ditch between March 1996 and March 1997. At the beginning and end of this period, the sediment layers at 0–20 and 20–50 cm depth, and the top layer of sludge and litter (1–6 cm) were sampled at eight sampling locations distributed over the full length of the ditch sections. Two rows of three samples were collected in the presettling basin. At each sampling location, four subsamples were pooled and the thickness of the sludge layer was measured (mean of four measurements). Sludge samples were dried for 2 days at 70 ◦ C. Dried and ground samples were analysed for N and P with a salicylic acid thiosulphate modification of the Kjeldahl method (see emergent macrophytes). Only data on the sludge layer will be shown, because accumulation of N and P was only observed in this layer and not in the sediment layer (0–50 cm) underneath.

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

107

Table 2 Annual water budgets of the wetland system and its compartments between April 1996 and March 1997 Water fluxes (m3 section−1 yr−1 )

Total system

Presettling basin

Parallel ditches

Per parallel ditch section Front sections

Rear sections

Phragmites

Typha

Control

Behind Phragmites

Behind Typha

Control

Inputs STP effluent inflow (×1000) Precipitation

1190 8181

1190 2169

1162 5489

137 296

135 297

131 294

137 304

135 310

131 307

Outputs STP effluent outflow (×1000) Groundwater recharge Evapotranspiration

1155 34929 8634

1162 27662 2389

1156 6344 5662

137 299 264

135 311 285

131 148 328

137 367 339

135 385 345

131 189 343

3. Results 3.1. Water budgets The STP effluent inflows and surface water outflows were by far the largest water inputs and outputs in the annual water budgets of the wetland system (Table 2). The other annual water fluxes to and from these wetland compartments were in most cases smaller than 0.5% of the STP effluent inflows. Only groundwater recharge in the presettling basin resulted in an output of 2.3%. Seasonal dynamics of

the STP effluent inflow rate to the wetland, precipitation and air temperature showed that inflow rates were high in summer and in periods of high rainfall (Fig. 2). Differences of inputs and outputs among the front and rear ditch sections with emergent, submerged or no vascular macrophyte vegetation were generally minor (Table 2). Groundwater recharge was lower in the shallow, front sections than in the deeper, rear sections. Evapotranspiration was about 25% higher from ditch sections with open water than from sections with emergent macrophytes.

Fig. 2. Monthly averages of the wetland inflow rate of STP effluent (grey bars), precipitation (white bars) and air temperature (curve).

Means (±S.E.) of the three replicate parallel ditches with Phragmites australis or Typha latifolia in the front sections are shown. Values as g m−2 yr−1 . Units of faecal coliforms are cfu 109 m−2 yr−1 . Negative values mean that the section produces rather than removes the pollutant. % of total: share of presettling basin and parallel ditches in the total removal if both removal values are positive.

−451 ± 125 2.3 ± 1.4 46.1 ± 27.0 21.0 ± 7.3 13.3 ± 20.2 11.7 ± 9.9 13.8 ± 3.1 12.7 ± 2.2 1.2 ± 1.0 −297 ± 97 1.3 ± 1.0 27.4 ± 15.0 −6.0 ± 11.3 34.0 ± 19.4 −1.0 ± 8.7 −0.3 ± 5.2 3.7 ± 3.3 −3.9 ± 2.7 147 5.6 261 14.3 219 26.1 0.7 0.6 0.04 −17.3 ± 135.8 20.1 ± 1.6 213 ± 6 19.8 ± 5.1 148 ± 40 45.3 ± 43.4 −26.3 ± 1.5 −23.6 ± 1.9 −2.7 ± 2.9 2563 39.0 489 64.1 237 189 70.8 59.8 11.1 COD Faecal coliforms Total N NH4 -N NO3 -N Organic N Total P PO4 -P Organic P

9.2 36.5 126 28.1 17.5 80.2 5.1 6.7 −1.6

−84.0 11.1 130 21.9 93.1 14.7 −4.6 −4.1 −0.5

− 22 63 + + 14 − − − − 78 37 − − 86 + + +

Typha

−319 108 194 −2.6 −35.6 232 28.0 16.1 11.9

393 ± 128 23.7 ± 0.6 264 ± 22 70.7 ± 7.5 170 ± 34 22.4 ± 2.5 −9.9 ± 1.9 −13.8 ± 0.6 4.0 ± 1.7

Behind Phragmites Phragmites

Control

Rear sections Front sections % of total Mass rate % of total Mass rate

Per parallel ditch section Parallel ditches Presettling basin Total system removal rate Wetland mass input Pollutant (all values per m2 per year)

Table 3 Annual pollutant mass inputs to the wetland and pollutant mass removal rates in the wetland sections

3.2.1. Entire wetland, contribution of presettling basin and parallel ditches The turbidity of the STP effluent generally increased during passage through the wetland, particularly between November and May (Fig. 3a, Table 4). The increases of turbidity in winter and spring were much more prominent in the parallel ditches than in the presettling basin. On an annual basis, the turbidity of the through-flowing water was more than doubled at the end of the wetland (Table 4). The low annual mean COD only slightly changed in the wetland (Tables 3 and 4). Faecal coliforms were strongly reduced in the wetland system throughout the year, and removal efficiencies were higher in spring and summer (>99%) than in autumn and winter (85–90%, Fig. 3b). The presettling basin was the major wetland compartment in this respect in spring and summer, whereas the parallel ditches were at least equally important in autumn and winter (Fig. 3b). On an annual basis, 73 and 19% of the faecal coliforms in the wetland input were removed in the presettling basin and parallel ditches, respectively (Tables 3 and 4). The annual mass input of total N (approximately 15% ammonium, 45% nitrate, and 40% organic N), was reduced by 26% in the system (Table 3). Total N removal was on average 1.3 times higher during the spring-summer period due to a relatively large seasonal difference in nitrate removal (Fig. 3c and e). The parallel ditches accounted for almost twice the annual total N removal compared to the presettling basin (Tables 3 and 4, Fig. 3c). Ammonium and nitrate removal rates in the presettling basin varied throughout the year, leading to a small annual net release of ammonium and nitrate (Fig. 3d and e, Tables 3 and 4). The annual total N removal in the parallel ditches was mostly due to reductions in inorganic N. Annual mass input and mass removal rate in the ditches were four times higher for nitrate than for ammonium. Annual total P removal in the system was small and mainly took place in the presettling basin (10% of the mass input, Fig. 3f, Tables 3 and 4). Both orthophosphate and organic P contributed to this removal, even though ortho-phosphate was always the major P species in the wetland input (annual mean of 84%, Tables 3 and 4). A small annual net release of total

Behind Typha

3.2. Water quality improvement

−182 5.9 44.1 −16.0 65.1 −5.1 −13.1 −16.4 3.4

S. Toet et al. / Ecological Engineering 25 (2005) 101–124 Control

108

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

109

Fig. 3. Monthly averages of the wetland mass input (curve) and the mass removal rate of pollutants in the presettling basin (grey bars) and the parallel ditches (white bars) between April 1997 and March 1997. The turbidity was expressed on a concentration basis (ntu). (a) Turbidity, (b) faecal coliforms, (c) total N, (d) ammonium, (e) nitrate, and (f) total P.

110

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

Fig. 3. (Continued ).

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

111

Table 4 Mean annual pollutant concentrations of the wetland input and of outputs of the wetland (sections) Pollutant

Wetland input

Presettling basin output

Parallel ditches output

Wetland output

Standard for effluent/water

COD Turbidity Faecal coliforms Total N NH4 -N NO3 -N Organic N Total P PO4 -P Organic P

28.4 2.05 45.9 5.15 0.69 2.56 1.90 0.79 0.67 0.12

30.2 3.41 11.0 4.90 0.72 2.76 1.42 0.74 0.63 0.10

32.3 7.97 3.3 3.90 0.53 1.91 1.47 0.76 0.67 0.11

30.3 5.97 2.8 4.14 0.42 2.30 1.42 0.78 0.61 0.15

125/– –/– –/2 10/2.2 1/– –/– –/– 1/0.15 –/– –/–

The unit of the pollutant concentrations is mg l−1 , except for turbidity (ntu) and faecal coliforms (103 cfu 100 ml−1 ). Standards indicated are the current standards for STP effluent and surface water in The Netherlands.

P was observed in the parallel ditches (3% compared to total input, Tables 3 and 4). Total P removal in the parallel ditches fluctuated during the year, with distinct periods of net release between May and August, and November–December (Fig. 3f). These total P releases largely consisted of ortho-phosphate and, to a smaller extent, of organic P. 3.2.2. Comparison of parallel ditch sections Turbidity increased less during passage through Phragmites stands than through Typha stands and the front section of the control ditch (Fig. 4a). Annual relative increases in turbidity in the front ditch sections with Typha, no vascular macrophytes and Phragmites were 127, 83 and 21%, respectively. In the ditches with Phragmites, turbidity mainly increased in the rear ditch sections, while in the other ditch types turbidity in the rear sections showed only small net annual changes (Fig. 4a, Table 3). Only minor annual changes in COD were observed in any of the ditch types (Table 3). Faecal coliform removal was usually considerably higher in the front than in the rear sections of the ditches with Phragmites or Typha throughout the year. The opposite regularly occurred in the control ditch, with similar faecal coliform removal rates in the front and rear ditch section on an annual basis (Fig. 4b, Tables 3 and 4). The removal of faecal coliforms was at least three times higher in the ditch sections with Phragmites and Typha than in the front section of the control ditch. The removal of faecal coliforms was often higher in the Phragmites stands than in the Typha stands, resulting in an almost 10% higher

annual removal efficiency in the Phragmites stands (Fig. 4b). No clear and consistent differences in total N removal among the types of both front and rear ditch sections were observed in the course of the year (Fig. 4c). On an annual basis, total N removal was somewhat lower in the Typha stands than in the other two types of front ditch sections (Table 3). Removal rates of all N species were generally higher in the front than in the rear ditch sections. Ammonium removal was often higher in the Phragmites stands than in the other two types of front ditch sections (Fig. 4d), resulting in a three times higher value on an annual basis (34% of the ditch mass input; Table 3). Lower removal rates or net releases of ammonium were found in the rear ditch sections. Annual ammonium removal was observed in the ditches with Phragmites and Typha stands (approximately 30 and 20%, respectively), but no annual change in ammonium mass was observed in the control ditch. Nitrate removal was frequently higher in the front section of the control ditch than in the other two types of both front and rear ditch sections (Fig. 4e), so that annual nitrate removal in the entire control ditch approached 40% of the input, at least 10% more than in the other ditches (Table 3). There was no consistent effect of ditch type on P removal (Fig. 4f). The limited, net annual P release observed for the three types of ditches, however, came about in different ways (Table 3). In the front sections of ditches with Phragmites and Typha, a considerable annual ortho-phosphate release occurred, which was partly compensated for by ortho-phosphate

112

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

Fig. 4. Monthly averages of the ditch mass input (curve) and the mass removal rates (bars) of pollutants in the front (left graphs) and rear sections (right graphs) of the three types of parallel ditches between April 1997 and March 1997. The turbidity was expressed on a concentration basis (ntu). (a) Turbidity, (b) faecal coliforms, (c) total N, (d) ammonium, (e) nitrate and (f) total P. Dark grey bars represent ditches with Phragmites australis in the front sections, light grey bars those with Typha latifolia and the white bars the control ditch (without vascular macrophytes). The rear ditch sections contain submerged macrophytes and macro-algae.

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

Fig. 4. (Continued )

113

114

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

Fig. 4. (Continued )

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

Fig. 4. (Continued ).

115

116

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

removal in the rear sections of ditches with Typha. The releases of ortho-phosphate mainly occurred during the growing season (seasonal ortho-phosphate data not shown). In the control ditch, there was a net annual ortho-phosphate release to the water column of the rear section, whereas organic P removal was relatively low. This resulted in a small annual total P release in the entire control ditch (7% of the total P mass input). 3.2.3. Oxygen An example of the general changes in oxygen concentration of the STP effluent during passage of the parallel ditches is shown for ditch number 5 between 15 and 20 August 1996 (Fig. 5). The oxygen concentration of the wetland input of approximately 4.5 mg l−1 decreased in the presettling basin to 3 mg l−1 and even further to about 1.5 mg l−1 in the front ditch section with Phragmites. The oxygen concentration showed a large increase in diurnal fluctuations in the rear ditch section with submerged macrophytes and macro-algae, with daily minima and maxima about 1 and 11 mg l−1 , respectively. The front ditch sections with emergent macrophytes only showed such an increase in winter and early spring when the plants were too short to shade the surface water. The height of the diurnal maximum oxygen concentration was positively related to the respective level of solar radiation.

Oxygen levels at the end of the wetland system were highest in early spring, when maximum values were often higher than the detection limit of 20 mg l−1 (Fig. 6). Relatively small differences were observed between minimum and maximum oxygen concentrations during this period. Oxygen concentrations decreased from the end of May. Maximum, mean and minimum daily values reached the lowest values in July–August (10, 5 and 1 mg l−1 , respectively). Minimum oxygen concentrations were often close to zero during this period. The amplitudes of the daily oxygen concentration were relatively large in September and October, while minimum and maximum oxygen concentrations became quite similar again in November, with mean values of around 8 mg l−1 . 3.3. Nutrient budgets 3.3.1. Nitrogen budget The N budget of the wetland calculated on the basis of mass fluxes (Table 5) shows that the STP effluent input to the wetland was by far the largest N source. Overall, the wetland system removed 26% of the incoming N. Independent estimates of N removal processes (e.g., denitrification, mowing of plants, net storage in the sediment) in the wetland only accounted

Fig. 5. The change in oxygen concentration (hourly averages) of the STP effluent along with the passage of ditch 5 between 15 and 20 August 1996. The input of ditch 5 (measured in front of the outlet of the presettling basin, light grey), the output of the front section with Phragmites australis (dark grey) and the output of the rear section with submerged macrophytes and macro-algae (black) are depicted. The photosynthetically active radiation (PAR) is also shown for this period (dotted).

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

117

Fig. 6. Minimum (dotted), mean (bold) and maximum (normal) oxygen concentration (daily averages) of the STP effluent at the wetland outlet between 15 April and 18 December 1996.

for 35% of the N removal estimated on the basis of mass fluxes (Table 5). The largest N removal was observed in the parallel ditches (18% of the wetland mass input) and most of it occurred in the front sections. The process measurements only accounted for 16% of the N removal

here. Denitrification was the most important N removing process in the vegetated front ditch sections (at least 72% of the net N removal observed by the process measurements), while the rate of this process was almost three times higher in the Phragmites than in the Typha stands. A relatively large net release of N from the sedi-

Table 5 Annual N budgets of the wetland (sections) between April 1996 and March 1997 N mass fluxes (kg N section−1 yr−1 )

Total wetland

Presettling basin

Parallel ditches

Parallel ditches Front sections

Rear sections

Phragmites

Typha

Control

Behind Phragmites

Behind Typha

Control

Inputs Surface water input Atmospheric deposition

6393 25.1

6393 5.2

5721 18.7

667 1.5

658 1.3

638 0.7

543 0.7

558 0.7

515 0.7

Outputs Surface water output Groundwater recharge Denitrification Harvest of macrophytes Accumulation in sediment Error

4770 178 108 23.0 161 1178

5721 148 20.9

543 1.3 14.1 2.4 1.7 106

558 1.4 4.8 3.4 −4.0 95.8

515 0.7 1.6

531 1.4 0.8

536 1.5 1.0

494 0.7 0.5

50.4 457

4598 26.2 83.9 23.0 82.6 926

12.8 108

7.5 3.4

7.7 12.6

17.8 2.7

Removal efficiencies (%) By mass budgets By process studies Wetland section area (m2 )

25.7 7.3 13130

10.6 3.4 3483

19.9 3.8 8808

18.8 2.9 476

15.4 0.8 477

19.3 2.4 472

2.4 1.8 488

4.1 1.8 497

4.2 3.7 493

Plant names in table refer to Phragmites australis and Typha latifolia. Removal efficiencies were based on N budgets of the water inputs and outputs (‘mass budgets’) and on summation of the independently measured N removal processes (‘process studies’).

118

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

Table 6 Annual P budgets of the wetland (sections) between April 1996 and March 1997 P mass fluxes (kg P section−1 yr−1 )

Total wetland

Presettling basin

Parallel ditches

Parallel ditches Front sections Phragmites

Inputs Surface water input Atmospheric deposition Outputs Surface water output Groundwater recharge Harvest of macrophytes Accumulation in sediment Error

930 0.18

930 0.05

832 0.12

863

832

873

26.7

21.4

3.9

98.5 0.01

103

Rear sections Typha

97.0 0.01

110

4.7

0.2

0.1

3.9

0.2

0.8

Control

Behind Phragmites

Behind Typha

Control

93.7 0.01

103 0.01

110 0.01

93.4 0.01

93.4

103

103

99.9

0.1

0.3

0.3

0.1

60.0

55.4

−0.4

0.1

−2.8

0.7

1.0

1.0

1.8

−23.5

20.6

−48.8

−5.2

−10.6

−0.5

−1.4

5.5

−8.4

Removal efficiencies (%) By mass budgets 7.2 By process studies 9.7 Wetland section 13130 area (m2 )

10.5 8.3 3483

−4.4 1.0 8808

−4.8 0.5 476

−12.9 −2.0 477

0.4 0.9 472

−0.1 1.2 488

6.3 1.3 497

−6.9 2.1 493

Plant names in table refer to Phragmites australis and Typha latifolia. Removal efficiencies were based on P budgets of the water inputs and outputs (‘mass budgets’) and on summation of the independently measured P removal processes (‘process studies’).

ment in the Typha stands led to a somewhat lower total N removal than in the other two types of front ditch sections (Table 5). Accumulation in the sediment was the most important N removal process in the front section of the control ditch (80% of the removal observed by process measurements). In the rear ditch sections, N removal rates based on the mass budgets of the water inputs and outputs were lower and more in agreement with those quantified for the removal processes (i.e., process estimates accounted for 45–88% of N removal as calculated from mass fluxes). Nitrogen accumulation in the sediment was the largest removal process in these ditch sections. No differences in N removal were observed among the three ditch types when the front and rear sections were combined. The overall error of the N budget was 18% of the wetland input. 3.3.2. Phosphorus budget Just as for N, the STP effluent input was by far the most important input for P (Table 6). The an-

nual removal of P (7% of the wetland mass input) largely occurred in the presettling basin, due mostly to accumulation in the sediment and to a lesser extent to groundwater recharge (Table 6). Almost 80% of the P removal in the presettling basin estimated by mass fluxes was explained by these process measurements. Although the estimates based on mass fluxes in individual parallel ditches resulted in small net release or small removal of P (less than 5% of the mass P input), independent process measurements indicated net P removal occurred in all ditches together through harvest of emergent vegetation (6% of total P removal).

4. Discussion This study was carried out to test the performance of a constructed wetland for polishing STP effluent. In the discussion, we will first critically evaluate our research approach and methodology. Then, we will discuss the performance of the wetland system in terms

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

of pollution control and surface water quality enhancement. The role of the various components of the wetland will also be discussed. Finally, we will reflect on the design of this wetland system and its management and give recommendations for optimisations in the future. 4.1. Approach and methodologies Our study aimed for a detailed picture of the water quality improvement by the wetland system and for a basic understanding of the role of the various components. Mass budget calculations for COD, suspended solids, N, P and faecal coliforms were based on a study of the water budget of the system combined with twoweekly measurements of component concentrations at the various sites in the wetland system where the water flowed from one compartment to the next. The flow rates measured were combined with data on the dimensions of the various compartments and resulted in quite accurate estimates of the surface water flow through the system. The surface water flows were several orders of magnitude larger than the other flow rates in the water budget, i.e. precipitation, evapotranspiration and groundwater recharge. Changes in storage were negligible on a monthly or annual basis. Little change in the distribution of the surface water over the nine ditches were observed in time. We did not separately measure the surface water outflow rate at the end of the system and, therefore, cannot calculate an error term in the budget, which has been advised by many authors to avoid sources of undefined bias (Winter, 1981; LaBaugh, 1986; Koerselman, 1989; Gan et al., 1990; Owen, 1995). However, in our case the very minor contribution of water flows other than surface water input and output, and the accurate measurements of the surface water flows did result in a reliable picture of the hydraulics of the system. Surface water samples for measuring water quality parameters were collected simultaneously at the inlet and outlet of the wetland compartments every 2 weeks in the morning, whereas fluctuations in flow rate and pollutant concentrations occurred during the day (data not shown). Flow-proportional samplers were used for continuously collecting the wetland input (inflow of the presettling basin) and wetland output (outflow of the discharge ditch) during the total study period, so that at least our input–ouput budgets for the system as a

119

whole are reliable. Daily variation in pollutant concentrations were occasionally relatively large in the STP effluent entering the presettling basin, but were much smaller at the outlets of the presettling basin and the ditches (Schreijer et al., 2000). Seasonal variations in pollutant concentrations and flow rates were also observed. To reduce such sources of error, mass budgets were compiled on a monthly basis, which was a much longer period than the hydraulic retention time of the wetland and its compartments. To enhance our understanding of the role of the various compartments of the system, we also carried out process studies of the N and P dynamics in the vegetation and the sediment. Removal of N and P though harvesting of the emergent macrophytes in the ditch front sections, as well as accumulation or release of N and P in the sediment of presettling basin and ditch sections were estimated with standard techniques. For denitrification, we used information collected in the same wetland system 1 year after the present study. We assumed that it would be reasonable to extrapolate correlations of the denitrification rates to characteristic features of the system, such as seasonal changes in temperature and periphyton biomass, to the year of the present budget study. However, the mass budget for N has shown that this approach has failed. Only 28% of the total N removal calculated from the mass budget was accounted for by the measurements of the various processes. Measurements of the processes other than denitrification, i.e., harvest of macrophytes, groundwater recharge and accumulation in the sediment, were reasonably accurate for the system as a whole, as was also shown by the close correspondence of the removal efficiencies for P calculated on the basis of mass budgets and process studies. The large N removal unaccounted for by our process measurements could not have been caused by ammonia volatilisation, as pH and temperature has been continuously too low for that process to be quantitatively important. By exclusion of other possibilities and considering that all other measurements have been sufficiently accurate, we conclude that the contribution of denitrification has been seriously underestimated by our extrapolations. As we found out ourselves (Toet et al., 2003) and is also confirmed in many other studies (Cl´ement et al., 2003; Hefting et al., 2004; Rutherford and Nguyen, 2004), denitrification is strongly variable in space and time and, therefore, difficult to es-

120

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

timate for larger spatial and temporal scales. Hence, we assume that the major proportion of the N removal in our system (more than 70%) can be attributed to denitrification. Even if we cannot be certain that denitrification was our main N removal process, our data on total N removal, as well as N removal through processes other than denitrification, are reliable. 4.2. Performance of the system as a whole and of its compartments Table 4 clearly shows that the system as a whole showed a good (94%) reduction of faecal coliforms, although standards for surface water were not totally met. The first-order, zero-background rate constant (ka ) of faecal coliform reduction in the entire wetland of 0.704 m day−1 was high compared to those of other surface-flow systems summarised in Kadlec and Knight (1996, 0.007–0.591 m day−1 ). It was, however, less efficient during autumn and winter than during spring and summer, resulting in wetland output densities usually larger than 2 × 103 cfu 100 ml−1 during the cold half-year. Total N was reduced by 26%, with the highest relative reduction of ammonium-N. Total P showed only a small reduction (5%) during passage through the wetland on an annual basis. Occasional peaks in total N and P concentrations were, however, buffered almost entirely by the system (data not shown). Total N as well as P concentrations at the output of the wetland were well below the standards for STP effluent, but distinctly higher than the standards set for surface water by the Dutch Government (Table 4, Anonymus, 1999). A twice higher N removal might have been expected on the basis of the correlation between mass removal rate and mass loading rates found in a large data base of treatment wetlands (Kadlec and Knight, 1996). The apparent low efficiency of our system for N may be due to the high hydraulic loading rate (25 cm day−1 ) in combination with a relatively low N input concentration of 5.15 mg l−1 . None of the surface or subsurface-flow wetlands in the data base mentioned in Kadlec and Knight (1996) had hydraulic loading rates in this high range. Two subsurface-flow systems in Denmark (Schierup et al., 1990) and two in the USA with hydraulic loading rates between 21 and 26 cm day−1 showed a range of N removal effi-

ciencies close to that of the Dutch surface-flow system (21.5–35.5%). COD showed a slight increase and turbidity more than doubled during passage of the water through the wetland system, as has also been found in other surfaceflow wetlands with low COD in the input (Kadlec, 1995; Knight et al., 1993). However, turbidity remained quite low and did not impair light availability for aquatic plants in the rear ditch sections. Microscopic inspection showed that the particle composition changed drastically from a dominance of sewage sludge particles rich in bacteria at the entrance of the wetland to a dominance of phytoplankton and aquatic plantderived detritus at the end. The particle composition at the discharge point of the wetland strongly resembled that of the receiving surface water: sewage sludge particles were absent but phytoplankton as well as plant detritus made up almost 100% of the total. Comparable changes in the composition of suspended solids were also found in some other surface-flow systems (Bode et al., 1998; Hey et al., 1994). Oxygen dynamics of the water also changed strongly and indicated a similar transition from typical STP effluent with low oxygen concentrations without diurnal or seasonal fluctuations towards ‘healthy’ surface water with diurnal oxygen fluctuations during the growing season involving high peaks during the day. The various components of the system behaved very differently in their contributions to the total water quality improvement. The presettling basin was very effective in the removal of the faecal coliforms. The ka value in this basin (1.27 m day−1 ) was almost two times higher than in the other compartments of this system, and was also much higher than found in other wetland systems treating STP effluent (Wittgren et al., 1996). Removal was also mostly higher in the front than in the rear sections of the parallel ditches. Pathogenic microorganisms may be reduced by natural die-off in the water column, but also by several other factors including sedimentation, filtration, desintegration through sunlight, adsorption to organic matter, predation, competition for limiting nutrients and toxins from microorganisms or plants (Gersberg et al., 1989; Kadlec et al., 2000). Daphnia spp. will have contributed substantially to the predation of faecal coliforms in our system, because high densities were observed in the presettling basin during the growing season (data not shown). They appeared to mainly feed on bacteria associated with

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

the activated sludge particles and not on microalgae, which were only present in low densities (Kampf et al., 1998). Additional faecal coliform removal occurred in the front sections of the parallel ditches. Contact with the plant shoots probably stimulated interception, adsorption and sedimentation of the bacteria. The parallel ditches had a main function in reducing N and enhancing oxygen dynamics in the through-flowing water. The bulk of the N removal in the total system (60%) occurred in the front sections of the ditches with the emergent stands of Phragmites australis or Typha latifolia (Tables 3 and 4). Presumably, this N removal has been mostly caused by denitrification in the periphyton attached to the shoot bases of the macrophytes (Toet et al., 2003). The conditions in the front ditch sections were particularly suitable for denitrification with respect to high nitrate loading rates, large substrate area for periphyton and associated microbial communities and low oxygen concentrations during the growing season. Harvesting the macrophytes in December contributed only 2–3% to the total N removal in the ditches, whereas sedimentation and adsorption contributed 7%. It did not matter much whether the front ditch sections were covered by Phragmites australis or by Typha latifolia. The only important differences were that ditch sections with Phragmites had higher ammonium as well as COD removal than ditch sections with Typha, which was probably related to higher amounts of plant surfaces available for interception of pollutants and support of microbial activity. The enhancement of the oxygen dynamics of the through-flowing water entirely occurred in the rear ditch sections, where submerged vegetation, macroalgae and periphyton were the major primary producers. The mean hydraulic retention time of 0.6 days in these sections was apparently sufficient to increase oxygen concentrations substantially in daylight conditions. The system as a whole did not have much effect on the total P concentrations of the through-flowing water (only a 2% decrease). This is also true for the role of the various compartments (Tables 4 and 6). The STP effluent had been chemically dephosphatised before entering the wetland system. There was some P retention in the presettling basin but the parallel ditches showed on average a small net release of P, in spite of the harvesting of the emergent macrophytes, which removed about 2% of the input to the parallel ditches,

121

and some minor removal through sedimentation and sediment adsorption. The low P removal was not unexpected. The low concentrations in the STP effluent and the high hydraulic loading rate of the system had been recognised on beforehand as factors limiting further P removal. This study showed that it is has actually been close to zero in this system. Still, peaks in P loading have been buffered, and output P concentrations have remained low even in these occasional events. 4.3. Reflection on design and management The current wetland system was designed as a buffer between a conventional STP with tertiary P treatment and a small system of fresh surface waters consisting of canals and ponds on the barrier island of Texel, the Netherlands. The total volume of fresh water on this island is rather limited, and the STP effluent from up to 40,000 people constitutes a quantitatively major freshwater input. The wetland system can be seen as an intermediary between STP effluent and surface water. The objective of water managers in the area was to use the wetland system as one of their tools to further improve the quality of the surface water on the island. In order to be effective for this purpose, the wetland system had to (1) substantially decrease the bacteriological pollution; (2) further reduce the nutrient load (total N and P) and buffer peak N and P concentrations; (3) enhance the oxygen concentration of the STP effluent and (4) enhance the biological quality of the effluent, i.e., reduction of sewage particles and increase of planktonic freshwater biota. The system has performed these functions reasonably well, with the exception of the reduction of the N and P load. The concentrations of faecal coliforms have been reduced to below the standard for bathing water during the growing season (see Table 4), which is a remarkably good result. Concentrations often more than ten times higher than this standard still occurred in autumn and winter, which shows that the system did not entirely meet the targets of the designers for faecal coliforms. An increase of the hydraulic retention time would enhance the performance of the system in this respect. It can be expected that the concentrations would meet this standard continuously after a reduction of the hydraulic loading rate of at least 50%.

122

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

Oxygen concentrations and the quality of the suspended particles were very much in line with those in the receiving surface water at the end of the system. These aspects also met the expectations of the designers. The system also buffered N and P discharge peaks effectively. The design of the system in terms of the size and sequence of the compartments composition has been appropriate for these aspects of water quality. The system performed less than expected for N removal. The removal was still substantial (26% of the input), but lower than anticipated from correlations between N loading and N removal in large data sets (Kadlec and Knight, 1996; Vymazal et al., 1998). The removal mostly occurred in the front sections of the ditches with emergent vegetation. The hydraulic retention time of the water in these shallow ditch sections was only 0.3 days. Further, N concentrations in the input were low, which leads to relatively slow diffusion into biologically active tissues. For P, there was hardly any removal by the system. Here, again, the combination of high hydraulic loading and low input P concentrations was most likely the main cause. The N and P concentrations in the output were still continuously higher than the standards for surface water, which will not be acceptable in future water management requirements. As the amounts of N and P removed by mowing the emergent plants is very small compared to the N and P mass throughflow, a more intensive mowing will not enhance the removal substantially. Annual mowing should still be continued to ensure the vitality of these plant beds on the longer term (Van der Toorn and Mook, 1982; Gran´eli, 1989; Van der Putten et al., 1997). A possible way to enhance the performance would be to increase the hydraulic retention time of the water, in particularly in the presettling basin (P) and the parallel ditch front sections (N). Further chemical treatment of the effluent for P removal could also be an option. Obviously, the current system has been shown to be robust in its performance for water quality improvement for faecal coliforms, COD, oxygen and biota composition. The compartments used and their sequential design have been paramount for this success. The removal of N and P, and faecal coliforms during autumn and winter could be optimised further by increasing the hydraulic retention time by enlarging the wetland system, especially by using a larger presettling basin and

larger areas with shallow water and emergent vegetation (Kadlec and Knight, 1996).

Acknowledgements We thank the personnel of the STP ‘Everstekoog’ for their hospitality and help, and the laboratory of the waterboard for the water sampling and analyses. Ren´e van den Berg, Mignonne Fakkel-Slothouwer, Alex Heikens, Anita Noordzij, Ester Olij and Marco Witte are kindly thanked for all their assistance. Dennis Whigham is gratefully acknowledged for his valuable comments on an earlier draft of this manuscript. This study was financially supported by the Waterboard Hollands Noorderkwartier, the Province of North Holland and the Ministry of Transport, Public Works and Water Management (REGIWA project nr A2-592-EUT93), the Netherlands Agency for Energy and Environment (NOVEM project nr 351240/1110), the Foundation for Applied Water Research (STOWA report nr 2000-10) and the Institute for Inland Water Management and Waste Water Treatment (RIZA report nr 2000.006).

References Anonymus, 1999. Vierde Nota Waterhuishouding. Staatsdrukkerij Den Haag. Bastian, R.K., Hammer, D.A., 1993. The use of constructed wetlands for wastewater treatment and recycling. In: Moshiri, G.A. (Ed.), Constructed Wetlands for Water Quality Improvement. CRC Press, Boca Raton, pp. 59–68, Chapter 5. Bobbink, R., Heil, G.W., Raessen, M.B.A.G., 1992. Atmospheric deposition and canopy exchange processes in heathland ecosystems. Environ. Poll. 75, 29–37. Bode, H., Gr¨unebaum, T., Imhoff, K.R., Nusch, E.A., Tatano, F., 1998. Polishing ponds as tertiary treatment of municipal waste water: design, construction, ecological aspects, efficiency. Eur. Wat. Manage. 1 (6), 51–56. Boersma, L., 1965. Field measurement of hydraulic conductivity below a water table. Agronomy 9, 222–233. Bremner, J.M., Mulvaney, C.S., 1982. Nitrogen-total. In: Page, A.L., Miller, R.H., Keeney, D.R. (Eds.), Agronomy 9: Methods of Soil Analysis. Part 2: Chemical and Microbiological Properties, 2nd ed. Soil Science Society of America, Madison, USA, pp. 595–624. Cl´ement, J.C., Holmes, R.M., Peterson, B.J., Pinay, G., 2003. Isotopic investigation of denitrification in a riparian ecosystem in western France. J. Appl. Ecol. 40, 1035–1048. Erisman, J.W., 1992. Atmospheric deposition of acidifying compounds in the Netherlands. Ph.D. Dissertation. Utrecht University, Utrecht.

S. Toet et al. / Ecological Engineering 25 (2005) 101–124 Freeze, R.A., Cherry, J.A., 1979. Groundwater. Prentice-Hall, Englewood Cliffs, NJ. Gan, K.C., McMahon, T.A., O’Neill, I.C., 1990. Errors in estimated streamflow parameters and storages for ungauged catchments. Wat. Res. Bull. 26 (3), 443–450. Gersberg, R.M., Gearheart, R.A., Ives, M., 1989. Pathogen removal in constructed wetlands. In: Hammer, D.A. (Ed.), Constructed Wetlands for Wastewater Treatment: Municipal, Industrial and Agricultural. Lewis Publishers, Inc., Chelsea, MI, pp. 431–445. Gran´eli, W., 1989. Influence of standing litter on shoot production in reed, Phragmites australis (Cav.) Trin. ex Steudel. Aquat. Bot. 35, 99–109. Haan, C.T., Barfield, B.J., Hayes, J.C., 1994. Design Hydrology and Sedimentology for Small Catchments. Academic Press, San Diego. Hefting, M., Clement, J.C., Dowrick, D., Cosandey, A.C., Bernal, S., Cimpian, C., Tatur, A., Burt, T.P., Pinay, G., 2004. Water table elevation controls on soil nitrogen cycling in riparian wetlands along a European climatic gradient. Biogeochemistry 67, 113–134. Hey, D.L., Kenimer, A.L., Barrett, K.R., 1994. Water quality improvement by four experimental wetlands. Ecol. Eng. 3, 381–397. Jackson, J., 1989. Man-made wetlands for wastewater treatment: two case studies. In: Hammer, D.A. (Ed.), Constructed Wetlands for Wastewater Treatment: Municipal, Industrial and Agricultural. Lewis Publishers, Chelsea, MI, pp. 574–580. Kadlec, R.H., 1995. Overview: surface flow constructed wetlands. Wat. Sci. Technol. 32 (3), 1–12. Kadlec, R.H., Knight, R.L., 1996. Treatment Wetlands. CRC Press, Boca Raton, 893 pp. Kadlec, R.H., Knight, R.L., Vymazal, J., Brix, H., Cooper, P., Haberl, R., 2000. Constructed Wetlands for Pollution Control: Processes, Performance, Design and Operation. IWA Publishing, London, 156 pp. Kampf, R., Schreijer, M., Toet, S., Verhoeven, J.T.A., Jak, R.G., Groot, M., 1998. From sewage to (re)usable surface water, the use of a full-scale constructed wetland to improve the quality of effluent from an oxidation ditch in The Netherlands. In: Jha, R.K. et al. (Ed.), Environment and Agriculture at the Crossroad of the New Millenium, Proceedings of the International Conference on Environment and Agriculture, November 1–3. Ecological Society, Kathmandu, pp. 443–455. Knight, R.L., McKim, T.W., Kohl, H.R., 1987. Performance of a natural wetland treatment system for wastewater management. J. Wat. Poll. Contr. Fed. 59 (8), 746–754. Knight, R.L., Ruble, R.W., Kadlec, R.H., Reed, S., 1993. Wetlands for wastewater treatment: performance database. In: Moshiri, G.A. (Ed.), Constructed Wetlands for Water Quality Improvement. Lewis Publishers, Boca Raton, pp. 35–58. Koerselman, W., 1989. Groundwater and surface water hydrology of a small groundwater-fed fen. Wetlands Ecol. Manage. 1, 31– 43. Koerselman, W., Beltman, B., 1988. Two models for the calculation of the evapotranspiration term in the water budget of peatlands from routine weather data. In: Proceedings of the International Symposium on Hydrology of Wetlands in Semi-arid and Arid Regions, Seville (Spain), May 1988, Seville, Spain.

123

LaBaugh, J.W., 1986. Wetland ecosystem studies from a hydrologic perspective. Wat. Res. Bull. 22 (1), 1–10. Lund, L.J., 1999. Nitrogen balance in a pond system receiving tertiary effluent. J. Envir. Qual. 28, 1258–1263. Luthin, J.N., Kirkham, D., 1949. A piezometer method for measuring permeability of soil in situ below a water table. Soil Sci. 68, 349–358. Makkink, G.F., 1957. Testing the Penman formula by means of lysimeters. J. Inst. Wat. Eng. 11, 277–288. Makkink, G.F., 1960. Evaporation from vegetations in relation to Penman’s formula. TNO Comm. Hydrol. Res. 4, 90–115 (in Dutch). Moshiri, G.A., 1993. Constructed Wetlands for Water Quality Improvement. Lewis Publishers, Boca Raton, 632 pp. Owen, C.R., 1995. Water budget and flow patterns in an urban wetland. J. Hydrol. 169, 171–187. Ridder, T.B., Baard, J.H., Buishand, T.A., 1984. De invloed van monstermethoden en analysetechnieken op gemeten chemische concentraties in regenwater, T.R.- 55. KNMI, De Bilt (in Dutch). Rutherford, J.C., Nguyen, M.L., 2004. Nitrate removal in riparian wetlands: interactions between surface flow and soils. J. Environ. Qual. 33, 1133–1143. Schierup, H.H., Brix, H., Lorenzen, B., 1990. Spildevandsrensning i rodzoneanlœg. Botanical Institute, Aarhus University, Aarhus, Denmark. Schreijer, M., Kampf, R., Toet, S., Verhoeven, J., 1997. The use of constructed wetlands to upgrade treated sewage effluents before discharge to natural surface water in Texel Island, The Netherlands—Pilot study. Wat. Sci. Technol. 35, 231– 237. Schreijer, M., Kampf, R., Toet, S., Verhoeven, J.T.A., 2000. Nabehandeling van RWZI-effluent tot bruikbaar oppervlaktewater in een moerassysteem—Resultaten van een 4-jarig demonstratieproject op praktijkschaal op RWZI Everstekoog, Texel, 1995–1998. Hoogheemraadschap van Uitwaterende Sluizen in Hollands Noorderkwartier, Edam, 95 pp. (in Dutch). Sundblad, K., 1998. Sweden. In: Vymazal, J., Brix, H., Cooper, P.F., Green, M.B., Haberl, R. (Eds.), Constructed Wetlands for Wastewater Treatment in Europe. Backhuys Publishers, Leiden, pp. 251–259. Toet, S., Huibers, L.H.F.A., Van Logtestijn, R.S.P., Verhoeven, J.T.A., 2003. Denitrification in the periphyton associated with plant shoots and in the sediment of a wetland system supplied with sewage treatment plant effluent. Hydrobiologia 501, 29– 44. Van der Putten, W.H., Peters, B.A.M., Van den Berg, M.S., 1997. Effects of litter on substrate conditions and growth of emergent macrophytes. New Phytol. 135, 527–537. Van der Toorn, J., Mook, J.H., 1982. The influence of environmental factors and management on stands of Phragmites australis. I. Effects of burning, frost and insect damage on shoot density and shoot size. J. Appl. Ecol. 19, 477– 499. Vymazal, J., Brix, H., Cooper, P.F., Green, M.B., Haberl, R., 1998. Constructed Wetlands for Wastewater Treatment in Europe. Backhuys Publishers, Leiden, 366 pp.

124

S. Toet et al. / Ecological Engineering 25 (2005) 101–124

Winter, T.C., 1981. Uncertainties in estimating the water balance of lakes. Wat. Res. Bull. 17 (1), 82–115. Wittgren, H.B., Stenstr¨om, T.-A., Sundblad, K., 1996. Removal of indicator microorganisms in surface-flow treatment wetlands. In: Proceedings of the Fifth International Conference on Wetland Systems for Water Pollution Control, Vienna, September 15–19,

Vol. 1. Institute for Water Provision, Water Quality and Waste Management, Universit¨at f¨ur Bodenkultur, Vienna, I/19, pp. 1–8. Youngs, E.G., 1968. Shape factors for Kirkham’s piezometer method for determining the hydraulic conductivity of soil in situ for soils overlying an impermeable floor or infinitely permeable stratum. Soil Sci. 106 (3), 235–237.