HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by van Driel, P. W. Articles by Merkley, L. C. Search for Related Content PubMed PubMed Citation Articles by van Driel, P. W. Articles by Merkley, L. C. GeoRef GeoRef Citation Agricola Articles by van Driel, P. W. Articles by Merkley, L. C. Related Collections Wetlands and Aquatic Processes Bioremediation and Biodegradation Nitrogen

TECHNICAL REPORTS

Ground Water Quality

Upflow Reactors for Riparian Zone Denitrification

Department of Earth Sciences, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada
Upper Thames River Conservation Authority, 1424 Clarke Road, London, ON, Canada

^* Corresponding author (wroberts{at}uwaterloo.ca)

Received for publication January 24, 2005.

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS AND RECOMMENDATIONS REFERENCES

 
We used permeable reactive subsurface barriers consisting of a C source (wood particles), with very high hydraulic conductivities (0.1–1 cm s^–1), to provide high rates of riparian zone NO₃–N removal at two field sites in an agricultural area of southwestern Ontario. At one site, a 0.73-m³ reactor containing fine wood particles was monitored for a 20-mo period and achieved a 33% reduction in mean influent NO₃–N concentration of 11.5 mg L^–1 and a mean removal rate of 4.5 mg L^–1 d^–1 (0.7 g m^–2 d^–1). At the second site, four smaller reactors (0.21 m³ each), two containing fine wood particles and two containing coarse wood particles, were monitored for a 4-mo period and were successful in attenuating mean influent NO₃–N concentrations of 23.7 to 35.1 mg L^–1 by 41 to 63%. Mean reaction rates for the two coarse-particle reactors (3.2 and 7.8 mg L^–1 d^–1, or 1.5 and 3.4 g m^–2 d^–1) were not significantly different (p > 0.2) than the rates observed in the two fine-particle reactors (5.0 and 9.9 mg L^–1 d^–1, or 1.8–3.5 g m^–2 d^–1). A two-dimensional ground water flow model is used to illustrate how permeable reactive barriers such as these can be used to redirect ground water flow within riparian zones, potentially augmenting NO₃^– removal in this environment.

Abbreviations: HRT, hydraulic retention time • K, hydraulic conductivity • OC, organic carbon

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS AND RECOMMENDATIONS REFERENCES

 
NUTRIENTS applied to augment crop production have resulted in environmental consequences such as elevated ground water NO₃–N concentrations (Fleischer and Stibe, 1989; Canter, 1996; Panno and Kelly, 2004). Nitrate-N-enriched ground water discharging to surface water courses can lead to excessive algal growth and cause ecosystem deterioration (Canter, 1996). One means of protecting aquatic ecosystems from nutrient overloading is through the maintenance of naturally vegetated riparian buffer strips adjacent to surface water courses. Here, in addition to NO₃–N assimilation into plant biomass, the presence of sediments enriched in OC (organic C) can stimulate microbial denitrification (Ehrlich, 1990):

[1]

Studies of NO₃–N attenuation in riparian buffer strips in agricultural areas have been undertaken by a number of researchers (e.g., Young et al., 1980; Groffman et al., 1991; Simmons et al., 1992; Osborne and Kovacic, 1993; Hill, 1996; Schnabel et al., 1996; Gold et al., 1998; Mengis et al., 1999; Royer et al. (2004) among others). Several studies report higher rates of NO₃–N removal in shallow soil zone sediments with elevated OC contents compared with underlying C horizon material (Lowrance, 1992; Flite et al., 2001), or in patches of C horizon material containing locally higher amounts of OC as a result of the presence of decaying root material (Gold et al., 1998; Jacinthe et al., 1998). The linkage between NO₃^– consumption and the availability of OC from the biologically active soil zone has caused studies of riparian-zone NO₃^– attenuation to be focused generally at relatively shallow depths of <1 to 2 m. However, organic-rich soil zone sediments also contain clay minerals produced by weathering reactions, and as a result have hydraulic conductivity (K) values that are usually orders of magnitude less than that of typical sand and gravel aquifers that are most prone to NO₃^– contamination (e.g., 10^{– 3} to 10^–1 cm s^–1, Freeze and Cherry, 1979). Thus, although NO₃^– consumption rates may be high in shallow C-rich zones, at sites where thicker aquifers are present, it is possible that much of the ground water approaching stream discharge points does so via deeper, more permeable flow paths that may have less opportunity to interact with shallow C sources. Additionally, lower-K sediments associated with riparian soil zones and stream alluvium can have the effect of hydraulically isolating aquifer flow systems from some surface water courses. Only a few studies have addressed the fate of NO₃^– in riparian zones with thicker aquifer systems present, however (e.g., Bohlke and Denver, 1995; Hill, 1996; Tesoriero et al., 2000; Puckett et al., 2002).

The use of artificially constructed permeable reactive barriers augmented with OC solids is an emerging technology for remediation of NO₃–N in ground water. Wood particles have been identified as an effective material for use in such barriers because of low cost, high permeability, and reaction rates that are suitable for long-term, multiyear treatment. Barriers or containerized reactors using particulate wood media have been used to treat NO₃–N from septic tank effluent (Robertson and Cherry, 1995; Robertson et al., 2000, 2005a), agricultural tile drainage (Blowes et al., 1994; Robertson et al., 1994, 2000; van Driel, 2004), and shallow ground water contaminated by agricultural practices (Schipper and Vojvodi-Vukovi, 1998; Robertson et al., 2005b). Robertson et al. (2005b) used a wood particle medium that was coarse grained (1–50 mm diam.) and had very high K (1 cm s^–1) in a reactive layer that caused convergence of ground water flow into the layer and allowed the capture of ground water from aquifer zones located below the bottom of the layer.

We conducted field trials in which wood particle reactors were installed at two sites where high-NO₃^– ground water from agricultural activity was migrating through streamside riparian zones. The Caddy-Bott Creek site has a thick (>4 m), areally extensive aquifer; at the Woodstock site, a thick aquifer (>10 m) is also present and the trial was run for a longer period, which included winter season monitoring. Our objective was to demonstrate the feasibility of using artificial permeable reactive barriers to passively augment NO₃^– removal in riparian zones. A numerical ground water flow model was also used to illustrate how high-permeability media can be used to modify flow paths within a riparian zone.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS AND RECOMMENDATIONS REFERENCES

 
Site Descriptions
The Woodstock site is in southwestern Ontario near the city of Woodstock, at a location where a ground water spring discharges at the headwaters of a stream. The site is in a local woodlot that is surrounded by agricultural terrain dominated by corn (Zea mays L.) and soybean [Glycine max (L.) Merr.] production. An unconfined, glacial outwash sand and gravel aquifer is present, which has elevated NO₃–N (5–15 mg L^–1) throughout its 10- to 15-m saturated thickness (Robertson et al., 2005b). The aquifer serves as a local source of water supply.

The Caddy-Bott Creek site is flat and is located beside a perennial stream near the city of London, also in southwestern Ontario. The creek has a catchment of 4 km², which is intensively cultivated, predominantly for corn and soybean production. An unconfined, glacial outwash sand and gravel aquifer of variable thickness underlies much of the catchment (Sado and Vagners, 1972), including the study site. Aerial photograph evidence indicates that the stream was ditched and straightened before 1955 to facilitate drainage, as is often the case in this area of Ontario, and that the creek's watershed has been under intensive cultivation since that time. Between 1994 and 1996, the creek was re-dredged and a 15-m-wide buffer strip was established along both sides of the creek (Fig. 1; Ontario Stewardship Centre, 2004). The buffer strips were revegetated naturally with grasses or were planted with trees. At the time of this study (2003), 40 cm of recent stream alluvium consisting of soft to moderately firm, organic rich, sandy silt sediment had accumulated in the creek bed during the 8-yr period following re-dredging of the creek. High well yields in two multilevel piezometer bundles installed in the up-gradient farm field (e.g., CD4, Fig. 1) indicate that the aquifer extends to at least the 4-m depth at the study site and contains ground water with NO₃–N concentrations >10 mg L^–1 and as high as 67 mg L^–1 (van Driel, 2004). The up-gradient field receives both chemical fertilizer and swine manure applications.

View larger version (71K): [in this window] [in a new window]   Fig. 1. Caddy-Bott Creek site, showing water table elevation (31 Oct. 2003) and location of the five upflow reactor test cells installed through the riparian sediments at the edge of the creek. Water level datum is the creek elevation at CD6.

View larger version (51K): [in this window] [in a new window]   Fig. 2. Hydraulic head values and ground water flow directions associated with the upflow reactors at: (a) the Woodstock site and (b) the Caddy-Bott Creek site. Head values (m) were measured 4 May 2003 at the Woodstock site and 31 Oct. 2003 at the Caddy-Bott Creek site.

The Caddy-Bott Creek reactors were installed into the riparian buffer strip at the edge of the creek on 24 and 25 June 2003. Five pits, 0.6 by 0.5 m in surface area, were dug manually to a depth of 70 cm through the low-K soil zone and into the underlying aquifer, which was encountered at a depth of 40 cm. These were each backfilled with 0.21 m³ of wood particle media (Fig. 2b), two with a fine particle medium (1–5 mm diam.) and three with a coarse particle medium (1–50 mm diam.). Different medium sizes were used to provide a comparison of reaction rates and hydraulic performance in reactors with different permeability characteristics. It was hoped that the use of the more permeable coarse medium would help to alleviate flow bypass problems that have been noted in previous studies using the fine wood particle medium (Schipper et al., 2004). A higher porosity value of 0.7 was assumed for the coarse medium, based also on an in situ tracer test performed by van Driel (2004). The pore volume of the fine particle reactors were thus estimated at 0.10 m³ (0.21 m³ x 0.47) and the coarse particle reactors at 0.15 m³ (0.21 m³ x 0.7). The reactors were unlined and drainage pipes were again installed into their upper surfaces (Fig. 2b) so that water levels could be maintained lower than the adjacent water table. The drainage pipes consisted of 8-cm-diam. perforated PVC (polyvinyl chloride) pipe that was capped at both ends and was fitted with a drainage tube of 0.6-cm-o.d. polyethylene tubing. The smaller diameter tubing facilitated the measurement of the lower flow rates that occurred in these reactors. Piezometers of 1.3-cm-diam. PVC pipe with 5-cm-long screened tips were installed into the bottom of each reactor to allow monitoring of the influent ground water composition (Fig. 2b). Additionally, water table wells and multiple piezometer bundles similar to those at the Woodstock site were installed near the creek and in the adjacent farm fields (Fig. 1) to establish the ground water flow direction and aquifer water quality parameters.

Monitoring
The Woodstock reactor was monitored weekly to monthly during the 20-mo period from April 2002 to December 2003 for NO₃–N concentrations, flow rate, and temperature. Influent water was drawn from the piezometer tip located 20 cm below the bottom of the reactive medium (Fig. 2a), while outflow water was sampled from the drainage pipe. Samples were filtered (0.45 µm) immediately on collection, stored in 20-mL polyethylene containers, and kept refrigerated until analysis, which occurred within 2 wk of collection. Except during Month 7 (13 Nov.–12 Dec. 2002) when semiquantitative NO₃–N values were obtained using a field test kit (Kit K-6902, CHEMetrics, Calverton, VA), NO₃–N analyses were performed at the Soil and Nutrient Laboratory, University of Guelph, ON. These were done using a colorimetric method (Cd reduction) with an autoanalyzer (Model TRAACS 800, Technicon Instruments Corp., Tarrytown, NY), using standards that provided a detection limit of 0.01 mg L^–1. Water temperature was measured using a digital temperature meter (Model WD-35607-10, Oakton Instruments, Vernon Hills, IL). Flow rate was measured from the outlet pipe using a calibrated container and stopwatch. A more detailed sampling of the Woodstock reactor was undertaken during Month 13 of operation (4 May 2003), at which time depth profiles of temperature, pH, Eh, NO₃–N, and alkalinity were obtained through the reactor. The Eh and pH were measured in the field using a digital pH/mV meter (Model 20, Barnant Co., Barrington, IL). The pH probe was calibrated after every five to seven samples using buffers of pH 4 and 7 and Eh readings were checked against Zobell's solution (Nordstrom, 1977). Alkalinity was measured in the field by titration of filtered samples with 1.6 M H₂SO₄ to pH 4.6. During Months 11 to 14 (March–June 2003), the flow rate through the reactor was deliberately varied in the range of 50 to 700 mL min^–1 by manipulating the elevation of the drainage pipe. This was done to assess the effect of variations in hydraulic retention time (HRT) on the amount of NO₃–N removed. This flow rate range equates to HRT of 0.3 to 5 d. Nineteen sampling events were undertaken during the period of flow manipulation, which generated most of the data from which reaction rates were determined. On all occasions > 0.5 pore volumes passed through the reactor at the modified flow rate before sampling, and on 16 of 19 occasions, >1.5 pore volumes passed through before sampling.

The Caddy-Bott Creek reactors were sampled five to eight times for NO₃–N content, flow rate, and temperature between 4 and 31 Oct. 2003, 4 mo after their installation, when a period of relatively stable flow was achieved. Reactor 5 did not flow adequately and was excluded from the trial. Samples were collected and flow rates were measured by attaching a plastic bag to the drainage tubes, otherwise measurement techniques and laboratory analyses were the same as at the Woodstock site. A standard t-test, assuming normal distribution, was used to assess the significance of reaction rate differences between the fine- and coarse-particle reactors.

Sediment samples were collected at both sites for analysis of grain size distribution and OC contents. These were obtained from grab samples collected during excavation of the reactor pits and, at the Caddy-Bott Creek site, also from a continuous, 0.7-m-long, undisturbed sediment core retrieved from the creek bed at the CD 6 location (Fig. 1). The core was obtained using a 5-cm-diam. aluminum core tube advanced into the sediment by percussion and was sectioned into 5-cm-long increments for grain size and OC analyses. The OC was analyzed at the University of Guelph Soil and Nutrient Laboratory by measuring mass loss on ignition (Heiri et al., 2001), which provided a detection limit of 0.01% dry wt. Grain size distribution was determined by dry sieving the Caddy-Bott Creek samples and by hydrometer analysis (Lambe, 1951) of the Woodstock samples, which were generally finer grained. Hydraulic conductivity was estimated from the grain size data using the Hazen method (Freeze and Cherry, 1979, p. 350). Most of the grain size analyses (11) were undertaken at the Caddy-Bott Creek site to provide K values for the numerical flow model, whereas fewer (two) were completed at the Woodstock site.

Ground Water Flow Model
A two-dimensional finite-element ground water flow model (FLONET version 5.0, Molson and Frind, 2000), which solves for steady-state hydraulic head, stream functions, and ground water velocity distribution, was used to simulate ground water flow in the cross-sectional plane transverse to Caddy-Bott Creek and approximately parallel to the direction of ground water flow indicated on Fig. 1. The modeled domain was 10 m long and 1 m in depth below the water table, and focused on the local flow system immediately adjacent to the creek. It utilized a 2121-node grid (101 nodes in the horizontal direction and 21 in the vertical direction), with 4000 uniform triangular elements. The objective of the modeling exercise was to illustrate how ground water underflow can occur when low-K stream alluvium is present and how flow paths change when these low-K sediments are breached. Field data from the site, including water level measurements and K values determined from the grain size data, were used to establish the boundary conditions and the approximate relative K distribution for the modeled domain.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS AND RECOMMENDATIONS REFERENCES

 
Woodstock Site
Ground Water Flow
At the Woodstock site, the ground water spring discharged actively during the winter, spring, and autumn months, but flow ceased during the summer when lower water table conditions occurred. During periods of active discharge, flow also occurred through the reactor (Fig. 2a). The reactor was not installed into typical aquifer sediments, but rather into cobbley gravel that had silty marl present as interstitial material. The K of the marl material, based on a single grain size analysis from the 1.05-m depth, was estimated at 10^–4 cm s^–1 (van Driel, 2004), which is three orders of magnitude less than the fine wood particle medium used in the reactor (0.12 cm s^–1, Table 1). During active flow periods (53% of the 20-mo monitoring period), flow through the reactor ranged from 0.05 to 1.05 L min^–1 (Fig. 3). During monitoring episodes when quantitative sampling for NO₃–N was undertaken, the mean flow rate was 0.35 ± 0.16 L min^–1 (Table 2), which equates to a one-dimensional flux rate of 69 cm d^–1.

View larger version (20K): [in this window] [in a new window]   Fig. 6. Comparison of grain size curves for the aquifer and shallow riparian sediments (soil zone and stream alluvium) at the Caddy-Bott Creek site. The alluvium is from core CD6 and the aquifer samples are from core CD6 and the reactor pit excavations (see Table 1).

View larger version (28K): [in this window] [in a new window]   Fig. 3. Flow rates and NO3–N attenuation in the Woodstock reactor during 20 mo of operation, beginning 25 Apr. 2002.

View larger version (10K): [in this window] [in a new window]   Fig. 4. Nitrate-N attenuation vs. hydraulic retention time in the Woodstock reactor. Data from the first 2 mo of operation, semi-quantitative analyses from Month 7, and monitoring events when effluent NO3–N was < 1 mg L–1 or flow rate was < 50 mL min–1 are excluded.

View larger version (22K): [in this window] [in a new window]   Fig. 5. Depth profiles of (a) NO3 and Eh, and (b) alkalinity and pH in the Woodstock reactor during Month 13 of operation (4 May 2003, flow rate 0.3 L min–1, temperature 8°C).

Figure 7 shows simulated ground water flow lines traversing a riparian zone–aquifer system with physical dimensions, hydraulic head values, and permeability characteristics similar to that of the Caddy-Bott Creek site. To allow for the possibility of additional secondary permeability from macropores, the stream alluvium was assigned a K value only 100 x less than the aquifer sediments for these simulations. With the low-K alluvium intact (Fig. 7a), the aquifer flow system is partially isolated from the stream and only 50% of the ground water flowing in the upper 1 m of the aquifer discharges to the creek. With the high-K reactor in place, however, flow to the creek increases by a factor of 2.5, and 80% of the flow in the upper aquifer then discharges to the creek, primarily via the flow conduit provided by the high-K reactor (Fig. 7b). The exact amount of the diversion into the creek will depend on a number of factors, including where the reactor is positioned. For example, positioning at the down-gradient edge of the creek would probably result in additional diversion into the creek. Although these flow conditions were not verified in the field because of the difficulty in quantifying natural seepage through the stream alluvium, these simulations nonetheless illustrate the degree to which flow paths can change when lower K riparian sediments are breached.

View larger version (28K): [in this window] [in a new window]   Fig. 7. Simulation of ground water flow lines associated with (a) the riparian zone of a small creek with an unconfined aquifer present and (b) modified flow lines with a high-hydraulic-conductivity (K) wood medium reactor present connecting the creek and the underlying aquifer sediments. Left and right boundary conditions are constant head (1.0- and 0.975-m head, respectively); top and bottom boundaries are zero flux, except for the creek, which has a constant head of 0.98 m. Each stream tube (area between adjacent flow lines) carries an equal amount of ground water flux. The model reflects field conditions observed at the Caddy-Bott Creek site.

Nitrate-Nitrogen Treatment
At Caddy-Bott Creek, monitoring results are reported for Month 4 after installation (4–31 Oct. 2003, Table 2), when a period of relatively continuous reactor operation was achieved under stable water table conditions. Before this, monitoring was undertaken during the first month of operation (July 2003), but results during startup were probably not representative of long-term reaction rates, and are thus not included in Table 2. The reactors did not operate consistently during August and September 2003, when lower water table conditions occurred. Subsequent to monitoring in October, the reactors were inundated during several precipitation events and the experiment was terminated.

Mean NO₃–N values entering the four reactors ranged from 23.7 ± 12.8 to 35.1 ± 5.1 mg L^–1. Mean flow rates through the reactors ranged from 18 ± 3 to 69 ± 34 mL min^–1, which equates to one-dimensional flux rates of 9 to 33 cm d^–1 (Table 2). Using effective porosity values determined from in situ tracer tests in similar wood particle media (0.47 for the fine medium and 0.7 for the coarse medium, van Driel, 2004), these flow rates equate to mean hydraulic retention times in the four reactors ranging from 1.4 ± 0.6 to 5.6 ± 1.1 d. Mean NO₃–N removal in the four reactors ranged from 12.1 ± 5.0 to 20.6 ± 7.6 mg L^–1 (Table 2). Reaction rates in the two coarse-particle reactors (means of 3.2 and 7.8 mg L^–1 d^–1, or 1.5 or 3.4 g m^–2 d^–1) were not significantly different (p > 0.2) than the rates observed in the two fine particle reactors (means of 9.9 and 5.0 mg L^–1 d^–1, or 3.5 and 1.8 g m^–2 d^–1, Table 2). It should be noted that there was difficulty in achieving consistent flow through the Caddy-Bott Creek reactors because of the development of flow restrictions in the discharge tubing as a result of gas bubble accumulation and bio-fouling. The tubing was purged at the end of each sampling event, and slowly declining flow rates appear to have ensued thereafter. Thus, flow rates measured immediately before each sampling event, which were used for NO₃–N consumption rate calculations, were probably minimum values that may have underestimated average flow during the previous several days. If this was the case, rate estimates would be conservative. The higher reaction rates observed in three of the four Caddy-Bott reactors, compared with the Woodstock reactor (3.2–9.9 vs. 4.5 mg N L^–1 d^–1, Table 2), is presumably at least partly the result of the higher average ground water temperatures encountered at the Caddy-Bott site (13 vs. 9°C, Table 2).

Comparison of Reaction Rates
Generally similar reaction rates were measured in all five of the reactors tested in this study (3.2–9.9 mg L^–1 d^–1, Table 2), including the two that used the coarse wood particle medium (3.2 and 7.8 mg L^–1 d^–1). These rates are in the same range as those reported previously for other wood particle reactors treating NO₃–N contamination. Schipper and Vojvodic-Vukovic (1998) monitored a reactive wall treating NO₃–N in ground water and estimated removal rates of up to 3.6 mg L^–1d^–1 during the first year of operation and observed similarly high values during longer term monitoring (Schipper and Vojvodic-Vukovic (2001). Robertson et al. (2000) measured removal rates of 3 to 22 mg L^–1d^–1 during a 6-yr period in a reactor treating agricultural tile drainage, while van Driel (2004) reported rates of 2 to 20 mg L^–1 d^–1 in two reactors also treating agricultural tile drainage. Robertson et al. (2005a) monitored four reactors treating nitrified septic tank effluent during a 3- to 5-yr period and reported reaction rates in the range of 7 to >10 mg L^–1 d^–1. Several of the previous studies also included the coarse wood particle medium (Robertson et al., 2000; van Driel, 2004) and relatively high reaction rates were also noted. It was suggested that the high rates resulted from the dual-porosity characteristics of the medium, which allowed reaction rims to penetrate several millimeters into the particles, rather than reaction sites being restricted to the particle surfaces.

Rates of NO₃–N removal in natural riparian zones have been estimated using a variety of techniques, including laboratory microcosms measuring denitrifying enzyme activity or N₂O evolution (Flite et al., 2001, <0.08–42 mg L^–1 d^–1; Royer et al. (2004), up to 0.043 mg L^–1 d^–1), laboratory measurements on intact sediment cores (Schnabel et al., 1996, 0.02–0.12 mg L^–1 d^–1; Jacinthe et al., 1998, 0.2 mg L^–1 d^–1), in situ microcosms (Mengis et al., 1999, 0.01 mg L^–1 d^–1), in situ tracer tests (Nelson et al., 1995, 0.13–0.39 mg L^–1 d^–1; Tesoriero et al., 2000, 0.53 mg L^–1 d^–1) and field mass balance estimates (Peterjohn and Correll, 1984, 0.035 mg L^–1 d^–1; Jorden et al., 1993, 0.031–0.072 mg L^–1 d^–1; Tesoriero et al., 2000, 0.0038–0.01 mg L^–1 d^–1; Flite et al., 2001, 0.42–2.9 mg L^–1 d^–1). Only a few of these estimates for natural riparian zones approach the rates measured in our reactors. Flite et al. (2001) reported NO₃–N consumption as high as 42 mg L^–1 d^–1 in laboratory microcosms using soil A horizon material impacted by NO₃^– from domestic wastewater, and at the same site estimated consumption of 0.4 to 2.9 mg L^–1 d^–1 based on field mass balance considerations. Most other reported rates are much lower than the direct in situ values obtained in our study. It is important to point out, however, that although natural rates may be lower, the large areal extent of many riparian zones can still provide for substantial mass removal. Where unit conversions were required for the above rate comparisons, the riparian zone sediments were assumed to be 1.0 m in depth, with soil particle density of 2.65 g cm^–1 and porosity of 0.35, unless specified otherwise.

Long-Term Operation
If it is assumed that all of the observed NO₃–N loss is the result of denitrification (Eq. [1]), the approximate lifespan of these reactors can be estimated based on their contained C mass and rate of NO₃–N removal (e.g., Robertson et al., 2000). Using the mean flow and removal rates from Table 2, NO₃–N consumption in the Woodstock reactor, when operating, was 1.9 g d^–1, which would require 2.0 g C d^–1. If the reactor operates 53% of the time (193 d yr^–1), annual C consumption would be 400 g. The Woodstock reactor contains 90 kg of OC (120 kg C m^–3 x 0.73 m³), thus about 0.4% would be consumed annually. Likewise, the mean flow and removal rates in the Caddy-Bott Creek reactors indicate NO₃–N consumption of 0.47 to 1.22 g d^–1, which would require 0.50 to 1.2 g C d^–1. If these reactors also operated 53% of the time, annual C consumption would be 97 to 230 g C yr^–1. Each reactor contains 25 kg C (120 kg C m^–3 x 0.21 m³), thus 0.4 to 0.9% would be consumed annually. These calculations are presented for discussion purposes only, as it is uncertain what fraction of the OC will ultimately be suitably labile to support denitrification and other reactions, such as reduction of dissolved O₂ and SO₄^2–, will also consume C. Nonetheless, the potential for long-term operation is illustrated. Robertson et al. (2000) documented up to 7 yr of operation of four wood-medium reactors treating NO₃^– from septic systems and agricultural tile drainage without noting an obvious decline in treatment effectiveness during that period. Schipper and Vojvodic-Vukovic (2001) documented 5 yr of operation of a wood particle wall treating ground water NO₃^– and observed consistent removal during that period. Successful long-term operation of reactors such as these will more likely depend on hydraulic factors related to clogging from gas bubble accumulation, bio-fouling, and inundation. Future demonstrations should further address longer term hydraulic performance.

	CONCLUSIONS AND RECOMMENDATIONS

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS AND RECOMMENDATIONS REFERENCES

 
This study has demonstrated the feasibility of using wood-particle reactors to passively augment NO₃^– removal in riparian zones. Reactive media with K values an order of magnitude or more greater than associated riparian sediments can have a substantial ability to capture ground water flow within riparian environments. At the ground water temperatures encountered in this study (9–13°C), the reactors achieved mean NO₃–N removal rates of 3.2 to 9.9 mg L^–1d^–1 (0.7–3.5 g m^–2 d^–1), which are higher than the removal rates reported for most natural riparian zones. Similarly high reaction rates observed in the coarse-particle reactors suggests that the coarse medium can be used to achieve improved flow convergence without sacrificing treatment performance. The challenge of achieving consistent and controllable flow through such reactors should be lessened in larger installations with higher flow rates. At some sites, it may be preferable to site the reactors farther away from the stream edge to minimize flow variations and inundation problems associated with stream fluctuations. The use of variable-height discharge tubing appears to be an effective technique for assisting with flow control. The reactors contain sufficient OC to potentially sustain denitrification for a number of years.

Simplicity of construction and the low cost of the reactive medium could allow reactors such as these to be considered for larger scale, cost-effective NO₃^– control at some sites, particularly where natural denitrification is inadequate. At sites where natural denitrification already occurs, however, the use of engineered systems should be considered carefully. Although natural removal rates may be lower, the larger size of many riparian zones, compared with any reactive barrier likely to be constructed, already provides for considerable mass removal in many cases. This raises the possibility that total mass removal could actually decrease after installation of a reactive barrier if substantial ground water is diverted into the barrier and much shorter overall hydraulic retention times in the denitrifying environment ensue. Additional studies should further examine the importance of natural denitrification in riparian zones where thicker aquifers are present.

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS AND RECOMMENDATIONS REFERENCES

 

• Blowes, D.W., W.D. Robertson, C.J. Ptacek, and C. Merkley. 1994. Removal of agricultural nitrate from tile-drainage effluent water using in-line bioreactors. J. Contam. Hydrol. 15:207–221.
• Bohlke, J.K., and J.M. Denver. 1995. Combined use of groundwater dating, chemical, and isotopic analyses to resolve the history and fate of nitrate contamination in two agricultural watersheds, Atlantic Coastal Plain, Maryland. Water Resour. Res. 31:2319–2339.
• Canter, L.W. 1996. Nitrates in groundwater. Lewis Publ., New York.
• Ehrlich, H.L. 1990. Geomicrobiology. 2nd ed. Marcel Dekker, New York.
• Fleischer, S., and L. Stibe. 1989. Agriculture kills marine fish in the 1980's. Who is responsible for fish kills in the year 2000? Ambio 18: 3347–3350.
• Flite, O.P., III, R.D. Shannon, R.R. Schnabel, and R.R. Parizek. 2001. Nitrate removal in a riparian wetland of the Appalachian Valley and Ridge physiographic province. J. Environ. Qual. 30:254–261.[Abstract/Free Full Text]
• Freeze, R.A., and J.A. Cherry. 1979. Groundwater. Prentice Hall, Englewood Cliffs, NJ.
• Gold, A.J., P.A. Jacinthe, P.M. Groffman, W.R. Wright, and R.H. Puffer. 1998. Patchiness in groundwater nitrate removal in a riparian forest. J. Environ. Qual. 27:146–155.
• Groffman, P.M., G. Howard, A.J. Gold, and W.M. Nelson. 1991. Microbial NO₃ processing in shallow ground water in a riparian forest. J. Environ. Qual. 25:1309–1316.
• Heiri, O., A.F. Lotter, and G. Lemcke. 2001. Loss on ignition as a method for estimating organic and carbonate content of sediments: Reproducibility and comparability of results. J. Paleolimnol. 25:101–110.
• Hill, A.R. 1996. Nitrate removal in stream riparian zones. J. Environ. Qual. 25:743–755.[Abstract/Free Full Text]
• Jacinthe, P.A., P.M. Groffman, A.J. Gold, and A. Mosier. 1998. Patchiness in microbial nitrogen transformations in groundwater in a riparian forest. J. Environ. Qual. 27:156–164.[ISI]
• Jorden, T.E., D.L. Correll, and D.E. Weller. 1993. Nutrient interception by a riparian forest receiving inputs from adjacent cropland. J. Environ. Qual. 22:467–473.[Abstract/Free Full Text]
• Keeney, D. 1986. Sources of nitrate to groundwater. Crit. Rev. Environ. Control 16:257–304.
• Lambe, T.W. 1951. Soil testing for engineers. John Wiley & Sons, New York.
• Lowrance, R.R. 1992. Groundwater NO₃ and denitrification in a coastal riparian forest. J. Environ. Qual. 21:401–405.[ISI]
• Mengis, M., S.L. Schiff, M. Harris, M.C. English, R. Aravena, R.J. Elgood, and A. MacLean. 1999. Multiple geochemical and isotopic approaches for assessing ground water NO₃^– elimination in a riparian zone. Ground Water 37:448–457.[CrossRef][ISI]
• Mitsch, W.J., and J.G. Gosselink. 2000. Wetlands. 3rd ed. J. Wiley & Sons, Toronto, ON, Canada.
• Molson, J., and E.O. Frind. 2000. FLONET; User guide version 5.0. Two-dimensional steady-state flownet model. Dep. of Earth Sciences, Univ. of Waterloo, Waterloo, ON, Canada.
• Nelson, W.M., A.J. Gold, and P.M. Groffman. 1995. Spatial and temporal variation in groundwater nitrate removal in a riparian forest. J. Environ. Qual. 24:691–699.
• Nordstrom, D.K. 1977. Thermodynamic redox equilibria of Zobell's solution. Geochim. Cosmochim. Acta 41:1835–1841.
• Ontario Stewardship Centre. 2004. Agriculture and wildlife: Caddy Bott agricultural drain for wildlife habitat enhancement. Project 22-13. Ontario Stewardship Centre, ON, Canada. Available at www. stewardshipcentre.on.ca/allDemoProjects/demoDisplay.asp?request_no=206 (verified 26 Oct. 2005).
• Osborne, L.L., and D.A. Kovacic. 1993. Riparian vegetated buffer strips in water-quality restoration and stream management. Freshwater Biol. 29:243–258.
• Panno, S.V., and W.R. Kelly. 2004. Nitrate and herbicide loading in two groundwater basins of Illinois' sinkhole plain. J. Hydrol. (Amsterdam) 290:229–242.
• Peterjohn, W.T., and D.L. Correll. 1984. Nutrient dynamics in an agricultural watershed: Observations on the role of a riparian forest. Ecology 65:1466–1475.[CrossRef][ISI]
• Puckett, L.J., T.K. Cowdery, P.B. McMahon, L.H. Tornes, and J.D. Stoner. 2002. Using chemical, hydrologic, and age dating analysis to delineate redox processes and flow paths in the riparian zone of a glacial outwash aquifer–stream system. Water Resour. Res. 38:9-1–9-20.
• Robertson, W.D., D.W. Blowes, and C.J. Ptacek. 1994. Treatment of contaminated agricultural run-off. U.S. Patent 5 330 651. Date issued: 19 July.
• Robertson, W.D., D.W. Blowes, C.J. Ptacek, and J.A. Cherry. 2000. Long-term performance of in situ reactive barriers for nitrate remediation. Ground Water 38:689–695.
• Robertson, W.D., and J.A. Cherry. 1995. In situ denitrification of septic-system nitrate using reactive porous media barriers: Field trials. Ground Water 33:99–111.[CrossRef][ISI]
• Robertson, W.D., G.I. Ford, and P.S. Lombardo. 2005a. Wood-based filter for nitrate removal in septic systems. Trans. ASAE 48:121–128.
• Robertson, W.D., N. Yeung, P.W. van Driel, and P.S. Lombardo. 2005b. High permeability layers for remediation of ground water: Go wide, not deep. Ground Water 43:574–581.[CrossRef][ISI][Medline]
• Royer, T.V., J.L. Tank, and M.B. David. 2004. Transport and fate of nitrate in headwater agricultural streams in Illinois. J. Environ. Qual. 33:1296–1304.[Abstract/Free Full Text]
• Sado, E.V., and U.J. Vagners. 1972. Surficial geology of the St. Mary's area. Map sheet 40P/3. Ontario Ministry of Natural Resources, Toronto, ON, Canada.
• Schipper, L.A., G.F. Barkle, J.C. Hadfield, M. Vojvodic-Vukovic, and C.P. Burgess. 2004. Hydraulic constraints on the performance of a groundwater denitrification wall for nitrate removal from shallow groundwater. J. Contam. Hydrol. 69:263–279.[Medline]
• Schipper, L.A., and M. Vojvodic-Vukovic. 1998. Nitrate removal from groundwater using a denitrification wall amended with sawdust: Field trial. J. Environ. Qual. 27:664–668.
• Schipper, L.A., and M. Vojvodic-Vukovic. 2001. Five years of nitrate removal, denitrification and carbon dynamics in a denitrification wall. Water Res. 35:3473–3477.[Medline]
• Schnabel, R.R., L.F. Cornish, W.L. Stout, and J.A. Shaffer. 1996. Denitrification in a grassed and a wooded valley and ridge, riparian ecotone. J. Environ. Qual. 25:1230–1235.[Abstract/Free Full Text]
• Simmons, R.C., A.J. Gold, and P.M. Groffman. 1992. Nitrate dynamics in riparian forests: Groundwater studies. J. Environ. Qual. 21:659–665.[Abstract/Free Full Text]
• Tesoriero, A.J., H. Liebscher, and S.E. Cox. 2000. Mechanisms and rate of denitrification in an agricultural watershed: Electron and mass balance along groundwater flow paths. Water Resour. Res. 36:1545–1559.[CrossRef]
• van Driel, P.W. 2004. Enhancing denitrification of agricultural tile drainage and discharging groundwater in riparian zones using wood media reactors. MS thesis. Dep. of Earth Sci., Univ. of Waterloo, Waterloo, ON, Canada.
• Young, R.A., T. Huntrods, and W. Anderson. 1980. Effectiveness of vegetated buffer strips in controlling pollution from feedlot runoff. J. Environ. Qual. 9:483–487.[Abstract/Free Full Text]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by van Driel, P. W. Articles by Merkley, L. C. Search for Related Content PubMed PubMed Citation Articles by van Driel, P. W. Articles by Merkley, L. C. GeoRef GeoRef Citation Agricola Articles by van Driel, P. W. Articles by Merkley, L. C. Related Collections Wetlands and Aquatic Processes Bioremediation and Biodegradation Nitrogen