Preliminary Evaluation of a Permeable Reactive Barrier for Reducing Groundwater Nitrate Transport from a Large Onsite Wastewater System

Department of Health Education and Promotion, East Carolina University, 3408 Carol Belk Building, Greenville, NC 27858, USA North Carolina Department of Health and Human Services, 5605 Six Forks Road, Building 3, Raleigh, NC 27609, USA Department of Engineering, Institute for Coastal Science and Policy, East Carolina University, Slay 250, Greenville, NC 27858, USA Department of Geological Sciences, East Carolina University, 204 Graham, Greenville, NC 27858, USA Coastal Resources Management Program PhD Program, East Carolina University, USA


Onsite Wastewater Treatment Systems and Nitrogen
Onsite Wastewater treatment Systems (OWS) are used in locations where centralized sewer service is not available. Over 23% of residences in the United States and almost 50% of residences in North Carolina use OWS (USEPA, 2002;Pradhan et al., 2007). Wastewater Total Dissolved Nitrogen (TDN) entering the septic tank is mostly Organic Nitrogen (ON) and ammonium nitrogen (NH 4 + -N), with very low concentrations of nitrate nitrogen (NO 3 --N) (Lowe et al., 2007). The organic nitrogen is typically mineralized to NH 4 + -N in the septic tank, thus septic tank effluent is typically 75% or more NH 4 + -N (Cardona, 2006). Effluent rich in NH 4 + -N may be converted to NO 3 --N in aerated soil beneath the drainfield trenches via the nitrification process (Humphrey et al., 2010), thus drainfield trenches are often referred to as nitrification trenches. Nitrate is an anion and is susceptible to leaching through soils, resulting in elevated groundwater NO 3 -N concentrations beneath and down-gradient from OWS. For example, NO 3 -N concentrations exceeding 20 mg L −1 in groundwater beneath and/or adjacent to septic systems have been reported for the Coastal Plain of North Carolina (Buetow, 2002;Humphrey et al., 2010;O'Driscoll et al., 2014), a sandy aquifer in Ontario, Canada (Harman et al., 1996), a coastal barrier bar in Point Pelee, Ontario, Canada (Ptacek, 1998), in Rhode Island (Postma et al., 1992) and in the Coastal Plain of Virginia (Reay, 2004). Furthermore, studies by Harman et al. (1996), Robertson et al. (1991) and Ptacek (1998) in Canada, Buetow (2002), Humphrey et al. (2013) and O'Driscoll et al. (2014) in the Coastal Plain of North Carolina and Reay (2004) in Coastal Virginia, reported groundwater NO 3 --N concentrations exceeding 10 mg L −1 more than 15 m down-gradient from the systems. Water supplies with NO 3 --N concentrations greater than 10 mg L −1 may be hazardous to infants and pregnant women due to methemoglobinemia, or blue baby syndrome (Brady and Weil, 2003). Therefore, the US Environmental Protection Agency maximum contaminant level for NO 3 --N is set at 10 mg L −1 for drinking water sources/supplies. Surface water NO 3 --N concentrations of 1 mg L −1 or potentially less may lead to increased algal production, eutrophication, lowering of dissolved oxygen concentrations and potential fish kills (NOAA, 1996). Massive fish kills in Albemarle Pamlico Estuary, particularly near the mouths of the Neuse and Tar Rivers in North Carolina have been attributed to excess nitrogen and phosphorus loadings (Osmond et al., 2003). Therefore, elevated NO 3 --N concentrations in groundwater and surface waters are a major environmental issue.

Onsite Wastewater Systems and Permeable Reactive Barriers
OWS are typically installed in aerobic soil environments that are favorable for nitrification (conversion of NH 4 to NO 3 ) to occur. Nitrate is mobile in groundwater and may discharge to surface waters contributing to increased nutrient loads and exacerbating eutrophic conditions (Bowen and Valiela, 2004). Denitrification (conversion of NO 3 to N 2 ) is a process for NO 3 removal from groundwater and generally requires anaerobic conditions, NO 3 , a carbon source and denitrifying microorganisms (Desimone and Howes, 1996). Denitrification in soil and/or groundwater beneath OWS is often limited by the lack of an available carbon source which functions as the electron donor (Robertson et al., 1991;Del Rosario et al., 2014). While septic tank effluent is typically enriched with carbon, the carbon is oxidized to CO 2 while NH 4 + -N is oxidized to NO 3 --N in aerobic soils beneath the trenches. So groundwater becomes enriched in NO 3 --N, but carbon escapes as a gas. Research in Canada and New Zealand has shown that permeable reactive barriers or PRBs (trenches excavated to depths below the water table and filled with woodchips, saw dust or other carbon sources ( Fig. 1)) can greatly reduce groundwater NO 3 concentrations in contaminant plumes via denitrification (Robertson and Cherry, 1995;Schipper and Vojvodic-Vukovic, 2001;Long et al., 2011). Furthermore, Schipper and Vojvodic-Vukovic (2001) demonstrated that the PRB treatment efficiency of NO 3 --N was 95% for after 5-years of operation. Long et al. (2011) reported a 92% NO 3 --N reduction efficiency for a 14 year old PRB installed to intercept groundwater impacted by spray irrigation from a dairy in New Zealand. Therefore, PRBs can be efficient treatment units for many years after installation.

Study Site
The study site was an elementary school in Martin County, NC that typically enrolls between 280 and 350 students. Wastewater generated at the school is managed via a large OWS. The OWS includes septic and pump tanks, a pressure manifold for effluent distribution and at least 9 drainfield trenches that are 64 min length. The drainfield trenches are located underneath a baseball field on the School's property. The OWS was installed in 2004. Three groundwater monitoring wells were installed near the OWS drainfield, with one well upgradient (Well 1) and two wells down-gradient (Well 2 and Well 3) from the OWS (Fig. 2).
Groundwater monitoring near the OWS began in 2005. Samples were typically collected 3 times per year from the wells and analyzed for NO 3 --N, fecal coli form and solids content for comparison to State water quality standards. Groundwater quality was initially met for all tested parameters, however the NO 3 --N concentrations in groundwater collected from Well 2 have steadily increased since monitoring was initiated in 2005 ( Fig. 3). Groundwater NO 3 --N standards are set at 10 mg L −1 and this threshold was first exceeded in Well 2 on the July 30, 2008 sampling event. The next two successive groundwater sampling events occurred in December 2008 (5.87 mg L −1 ) and April 2009 (6.46 mg L −1 ) and both were below 10 mg L −1 . However, the August 2009 sample from Well 2 (11.24 mg L −1 ) exceeded the standard and every sample collected from Well 2 since August 2009 has been above the 10 mg L −1 threshold. The objectives of this study were to install a permeable reactive barrier down-gradient from a large OWS and determine if the barrier was successful in reducing groundwater NO 3 --N and TDN concentrations.

Barrier Installation
A trench with approximate dimensions of 1.2 m wide x 8 m deep x 6 m long was installed with an excavator on May 7, 2014. The trench was excavated just up-slope of Well 2, the well with elevated NO 3 --N concentrations (Fig. 2). Once groundwater was encountered at approximately 5.8 m beneath the surface, several more bucket loads of soil were excavated and then woodchips were dumped into the trench (Fig. 4). The woodchips used for this barrier were a mixture of different size chips from different tree species (although mostly Pinus Taeda). The woodchips came from a local lumber mill. Soil from the trench excavation was placed over the woodchips to partially enclose and stabilize that section of the barrier and then the next section of the trench was excavated. This process was repeated until the trench and barrier extended across both up-gradient sides of Well 2, perpendicular to the groundwater flow direction. The trench was then filled with the excavated soil to the pre-trench elevation.

Groundwater and Onsite System Monitoring
Depth to groundwater was determined at each well using a Solinst TLC meter. Each well was purged using a disposable bailer and groundwater was transferred from the bailer to Nalgene samples bottles and field meter cups. The sample bottles were labeled and placed in a cooler with ice for transport to the lab. An YSI 556 multimeter was used to determine groundwater pH, Electrical Conductivity (EC), Dissolved Oxygen (DO), temperature and Oxidation Reduction Potential (ORP) in the field. The YSI was calibrated one day prior to each field sampling event.
Water samples were collected from the Wells 1-3 on May 6, 2014, one day prior to the barrier installation. The samples were sent to Environment 1 laboratory in Greenville, NC for NO 3 --N analysis. This was a routine sampling event for the school and Environment 1 is the laboratory they have used historically. After the barrier installation, groundwater samples from Wells 1-3 were collected on 5 occasions including May 22, June 5, 17, 24, 2014 and September 3, 2014. The samples were filtered and analyzed for NO 3 --N, NH 4 + -N, TDN, Dissolved Organic Carbon (DOC) and Cl at the East Carolina University Environmental Research Laboratory during May and June and at Environment 1 on September 3, 2014. The NO 3 --N, NH 4 + -N and Cl analysis were performed using a Smart Chem 200 at East Carolina University. The TDN and DOC analysis were performed using a Shimadzu TOC-VCPN/TNM-1 and the combustion catalytic oxidation method and chemiluminescence detector. Pump tank effluent samples were collected and analyzed for the same constituents on these same dates; except for June 17 (lock would not open on pump tank). Water samples from an unnamed tributary of the Roanoke River approximately 170 m down-gradient from the OWS were also evaluated on site for the environmental parameters including pH, DO, EC, ORP and temperature using the YSI 556 meter and for NO 3 --N using an YSI Pro-Plus. The tributary will be referred to as "creek" from this point forward in the manuscript.

Groundwater Nitrogen Trends and Permeable Barrier Performance
Groundwater NO 3 --N data from Wells 1-3 from 2005 to May 6, 2014 were provided by the NC Department of Health and Human Services. Well data were plotted on graphs and regression and correlation analysis were used to determine if there were significant trends in NO 3 --N concentrations from 2005 to May 6, 2014 (before barrier installation). Groundwater NO 3 --N concentrations postbarrier were compared to the pre-barrier concentration trends and to the May 6, 2014 data (one day prior to barrier installation) to determine if the barrier was influencing groundwater NO 3 --N concentrations. Groundwater Cl concentrations were evaluated in relation to NO 3 --N concentrations to determine if potential changes in NO 3 --N were related to dilution. If both Cl and NO 3 --N decrease, then dilution is a potential dominant mechanism (Desimone and Howes, 2006;Del Rosario et al., 2014). If NO 3 --N concentrations decrease but Cl concentrations remain stable or increase, then denitrification may be occurring (Desimone and Howes, 2006). Groundwater DOC concentrations in the Wells were graphed to determine if noticeable changes occurred in DOC concentrations in Well 2 after the barrier was installed and in comparison to Well 1 and Well 3. If groundwater moves through the woodchip barrier, then DOC concentrations may increase in Well 2 (down-gradient of barrier) relative to Well 1 and Well 3 (not down-gradient of the barrier). The Well 2 NO 3 --N, Cl and DOC data were compared to Well 1 and Well 3 to help determine if potential changes in groundwater chemistry after the barrier was installed were related to the barrier (changes in Well 2 only) or some other natural processes (all wells show no changes or similar changes).

Groundwater Nitrogen Trends Prior to Barrier Installation
Groundwater NO 3 --N concentrations from Well 2 showed a steady increase since the initiation of routine sampling in 2005. Regression analysis indicated a strong relationship between Well 2 groundwater NO 3 --N concentrations (annual average) and years after the start of the sampling (R 2 = 0.89) (Fig. 5). The NO 3 --N concentration in groundwater from Well 2 has increased an average of 1.8 mg L −1 each year since 2006 and the yearly average NO 3 --N concentrations exceeded 20 mg L −1 in 2012 and 2013 (Fig. 3). There was a strong, positive correlation between years since the sampling was initiated and groundwater NO 3 --N concentrations at well 2 (r = 0.947; p = 0.000). In contrast, the groundwater NO 3 --N concentrations from Well 1 were not strongly associated with years after sampling started (R 2 = 0.0089) and there was not a significant correlation between NO 3 --N concentrations and time after sampling started (r = 0.094; p = 0.796) (Fig. 5). Well 1 is located up-gradient from the OWS (Fig. 2). Groundwater NO 3 --N concentrations from Well 3 showed a moderate relationship with years after sampling (R 2 = 0.47) (Fig. 5). There was a moderate (r = 0.685) but significant (p = 0.029) correlation between the NO 3 --N concentrations and years after sampling was initiated. These data indicate that groundwater downgradient from the OWS was becoming more enriched with NO 3 --N over time, especially near Well 2.

Groundwater Nitrogen Trends after Barrier Installation
The groundwater samples collected from Well 2 on May 6, 2014, the day prior to the PRB installation, had a NO 3 --N concentration of 19.04 mg L −1 . Samples collected on May 22, June 5, 17, 24 and September 3, 2014 after the installation each had NO 3 --N concentrations lower (range of 12.2 to 15.95 mg L −1 ) than pre-barrier installation concentrations. On the last sampling date of September 3, groundwater NO 3 --N concentrations from Well 2 were 6.84 mg L −1 or 36% lower than on May 6, 2014 the day prior to the PRB installation and the lowest measured concentrations since 2006. Groundwater NO 3 --N concentrations for Well 1 and Well 3 were also lower after the installation of the barrier, but the differences were not as large (Well 1 difference: 0.13 mg L −1 ; Well 3 difference: 0.29 mg L −1 ) (Fig. 6). When comparing groundwater Cl concentrations in the Wells after the barrier installation there was a slight increase for Well 2 (+1.97 mg L −1 ) and Well 3 (+2.79 mg L −1 ) and a slight decrease for Well 1 (0.95 mg L −1 ) (Fig. 6). Had the NO 3 --N concentration reductions for Well 2 been due to dilution alone, then Cl concentrations should have also been reduced during that time frame. A reduction in groundwater NO 3 --N concentration coupled with no change or a slight increase in groundwater Cl concentration may indicate that the NO 3 --N reductions were not because of dilution, but some other mechanism such as denitrification. Wastewater typically has elevated concentrations of Cl relative to most groundwater (Humphrey et al., 2013;Del Rosario et al., 2014). At the study site, the average Cl concentration of wastewater was 158 mg L −1 and groundwater up-gradient from the OWS had average Cl concentrations of less than 5 mg L −1 (Table 1). Therefore, if the wastewater plume mixes with more groundwater (or infiltrating rainwater), then both NO 3 --N and Cl concentrations would be expected to decline in the plume because of dilution. This did not occur at Well 2. Groundwater DOC concentrations from Well 2 increased from 16.2 mg L −1 on May 22 to 27.48 mg L −1 on June 24 (Fig. 7), possibly because of DOC leaching from the woodchip barrier. The DOC concentrations were stable over this same time frame for Well 1 and initially dropped for Well 3, but were then stable (Fig.  7). While Well 2 and Well 3 were both down-gradient from the OWS, Well 2 was located 26.8 m north of Well 3 (Fig. 2). The predominant direction of groundwater flow was northwest (Fig. 2), thus it is unlikely that the initial drop in DOC concentrations observed at Well 3 were related to the increases in DOC concentrations at Well 2.
Groundwater samples collected from Well 2 showed a decrease in NO 3 --N, increase in DOC and slight increase in Cl after the barrier was installed (Fig. 7). These data suggests that the barrier may be influencing groundwater quality as intended. Had the drop in NO 3 --N been due to dilution, we would have expected to see a decline in Cl concentrations in Well 2 (and possibly Well 3) samples, but that did not happen. Also, the environmental conditions were favorable for denitrification to occur in the barrier with average ORP, DO and temperature values of -4 mV, 2.7 mg L −1 and 19.8°C, respectively in groundwater samples collected from Well 2 ( Table 2). In reviewing the long term data and trends, there was a noticeable drop in average NO 3 --N concentrations after the installation of the barrier (Fig.  8). Funding for this project would only allow for a few months of monitoring after the installation of the barrier and thus the results are preliminary. However a field study by Schipper and Vojvodic-Vukovic (2001) and a field and lab study by Robertson et al. (2008) both reported that PRBs could reduce groundwater NO 3 --N concentrations within a few months after installation and PRBs could continue to be effective for many years. Therefore, it is possible that the PRB installed in Martin County may help lower groundwater NO 3 --N concentrations for decades.

Onsite System and Barrier Nitrogen Concentration Reduction Efficiency
There was an average TDN (NO 3 --N + NH 4 + -N + Organic N) concentration reduction of 78% from pump tank effluent (65.7±3.6 mg L −1 ) to groundwater at Well 2 (14.2±1.2 mg L −1 ) after the barrier was installed (Fig.  9). Most of the groundwater TDN was NO 3 --N (Table  1) and thus there was opportunity for further reduction of groundwater NO 3 --N concentrations via denitrification, plant uptake and/or dilution in the riparian zone and creek bed over 152 m down-gradient from the OWS (Buetow, 2002;Osmond et al., 2002).
Creek readings were performed on June 17, 2004 and June 24, 2004 with an YSI ProPlus meter and the NO 3 --N concentration on both occasions was less than 1 mg L −1 (0.65±0.35 mg L −1 ).

Conclusion
Groundwater NO 3 --N concentrations steadily increased by 1.8 mg L −1 each year since 2006 near Well 2 at the study site prior to the installation of the barrier. Groundwater samples (5) from Well 2 collected 2 weeks to 4 months after the installation were on average 5 mg L −1 lower than samples collected the day prior to the installation. Groundwater DOC increased by more than 11 mg L −1 and Cl concentrations increased almost 2 mg L −1 over the same period. These data suggest that the NO 3 --N reductions were not due to dilution alone. It should be noted that while groundwater NO 3 --N concentrations have exceeded the 10 mg L −1 drinking water standard at Well 2, community water has been extended to the area near the study site. The public water supply is routinely tested for contaminants to protect public health. Also, most water supply wells are not installed in the surficial aquifer, but deeper, confined aquifers that may not be very well connected to the surficial unit (Stone et al., 1995;Humphrey et al., 2013). Therefore, elevated NO 3 --N concentrations in the surficial aquifer may not have a significant influence on groundwater in deeper aquifers (used for water supply). There is great potential for denitrification in the riparian area and creek bed to limit OWS plume NO 3 --N loading to the creek. Because the creek is located over 152 m from the OWS, there is also opportunity for dilution and dispersion to lower NO 3 --N concentrations as groundwater approaches the stream. The creek had less than 1 mg/L NO 3 --N on both dates it was tested.
Groundwater residence times from the OWS to the stream are not known at this time. With additional funding a more comprehensive investigation could be completed. Installing at least one monitoring well for sample collection and analysis up-gradient and at least one well within the barrier would be beneficial for quantifying the NO 3 --N treatment efficiency of the barrier. The NO 3 --N concentrations and NO 3 --N to Cl ratios up-gradient, within the barrier and down-gradient (Well 2) over time could be analyzed to provide a more definitive answer as to the effectiveness of the barrier. Cost-effectiveness of these systems may be related to the water table and plume depth. In the North Carolina Coastal Plain, shallow (<5 m depth) water tables are common and recent groundwater depth models by Fan et al. (2013) may help to screen areas  where water table depth is suitable for PRBs. There is a lack of field based research comparing the efficiency of barriers in remediating NO 3 --N and other contaminants in different geologic settings and with various mixtures of reactive materials. More work is suggested in these research areas.