Diffuse nitrate (NO3) contamination from intense agriculture adversely impacts freshwater ecosystems, and can also result in nitrate concentrations exceeding limits set in drinking water regulation, when receiving surface waters are used for drinking water production. Implementation of near-natural mitigation zones such as reactive swales or wetlands have been proven to be promising measures to reduce nitrate loads in agricultural drainage waters. However, the behavior of these systems at low temperatures and its dependence on system design has not been well known until now. In this study, the behavior of a full-scale (length: 45 m) reactive swale treating drainage water from an agricultural watershed in Brittany (France), with high nitrate concentrations in the receiving river, was monitored for one season (6 months). As flow in this full-size field system is usually restricted to winter and spring months (December–May), it usually operates at low water temperatures of 5–10 °C. Tracer tests revealed shorter than designed retention times due to high inflows and preferential flow in the swale. Results show a correlation between residence time and nitrate reduction with low removal (<10%) for short residence times (<0.1 day), increasing to >25% at residence times >10 h (0.4 day). Performance was compared to results of two technical-scale reactive swales (length: 8 m) operated for 1.5 years with two different residence times (0.4 and 2.5 days), situated at a test site of the German Federal Environmental Agency in Berlin (Germany). Similar nitrate reduction was observed for comparable temperature and residence time, showing that up-scaling is a suitable approach to transferring knowledge gathered from technical-scale experiments to field conditions. For the design of new mitigation systems, one recommendation is to investigate carefully the expected inflow volumes in advance to ensure a sufficient residence time for effective nitrate reduction at low temperatures.

INTRODUCTION

Agriculture is widely regarded as the most important diffuse source of nutrients in surface waters (Di & Cameron 2002; EPA 2002; EEA 2005). The wash-out of fertilizer and pesticides via surface run-off or subsurface leaching into drainage systems which discharge into surface waters presents an increasing risk for drinking water production and biodiversity in rivers and lakes. Despite efforts to reduce this pollution by changed agricultural practices (i.e. by limiting fertilization and pesticide application), high concentrations of nitrate (NO3) and pesticides in rural rivers remain a concern (e.g. Bernot et al. 2006; Alexander et al. 2008). For example, waterworks had to be closed in Brittany due to nitrate concentrations in the Ic River exceeding 50 mg-NO3/L−1 (Rouault et al. 2012). Agricultural areas, which are artificially drained by surface drainage systems or subsurface tile drains, are particular pollution hotspots, increasing the amount and rate of transfer of NO3 and pesticides from agricultural areas to surface waters (Randall & Mulla 2001; DWA 2008).

In addition to changing agricultural practices, implementation of near-natural mitigation zones such as reactive swales or wetlands have been proven to be promising measures to reduce nitrate loads in agricultural drainage waters (Vymazal 2007; Kadlec & Wallace 2009; Périllon & Matzinger 2010). According to a comprehensive literature review on NO3 removal (Périllon & Matzinger 2010), the addition of organic carbon sources enhances the effectiveness of mitigation zones considerably. Reactive swales with organic material emerged to be one of the most efficient designs for denitrification at short hydraulic retention times (Schipper et al. 2010).

Laboratory column experiments under saturated conditions have shown that usage of straw and bark mulch (mixture) as organic material is very efficient for retention of nitrate at short hydraulic retention times (HRT) of 0.17 days (4 h) and 0.34 days (8 h) at constant flow and room temperature (Krause Camilo et al. 2013). In the same project, two technical-scale reactive swales filled with the same organic material were run outside at varying seasonal temperatures (summer and winter) at two hydraulic retention times of 0.4 and 2.5 days to investigate nitrate and pesticide retention at short HRT and under realistic conditions.

However, the effect of low temperature on nitrate removal in full-size reactive swales as well as the long-term behavior of such systems needs to be better understood. Therefore, the objective of this study was to evaluate the nitrate removal of a full-size reactive swale with flow occurring at low temperatures (winter and spring) in its third year after construction and to compare results with technical-scale swales. With this study it is also possible to indicate if up-scaling is a suitable approach to transferring knowledge gathered from technical-scale experiments to field conditions.

METHODS

Full-size reactive swale in Brittany

The river Ic catchment (92 km2 catchment situated in Brittany, France) was chosen as the study watershed because nitrate concentrations in the river Ic frequently exceed the European threshold of 50 mg-NO3/L in the river Ic due to intensive agriculture in the catchment. Seven locations were investigated for their suitability, of which one site was chosen to build a reactive swale (see Figure 1 left, catchment size: 8.5 ha).

Figure 1

Full-size reactive swale constructed in Brittany, France (left) and technical-scale swale at a test site of the UBA in Berlin, Germany (right).

Figure 1

Full-size reactive swale constructed in Brittany, France (left) and technical-scale swale at a test site of the UBA in Berlin, Germany (right).

The unplanted reactive swale (length: 45 m, width: 2 m, depth: 1.3 m, slope: 2%) was constructed in a former drainage ditch beside an agricultural field. The specific length of the swale in relation to catchment size was 5.3 m/ha. The swale was filled with filter materials (sand/gravel) to ensure hydraulic efficiency and organic substrates as reactive materials (wood chips) to provide a carbon source for denitrifying bacteria. Layer design (cross section) and filter materials can be seen in Figure 2. The swale was run under saturated conditions (water level within the top gravel layer above the organic substrate) for the whole period of the investigation to promote anoxic conditions within the swale.

Figure 2

Cross section and filling materials of full-size reactive swale in Brittany.

Figure 2

Cross section and filling materials of full-size reactive swale in Brittany.

Monitoring was conducted from December 2012 until May 2013 (third year of operation). Inflow and outflow were determined using an ultrasonic water-level probe (Ijinus M0111501) and manually measured rating curves (inflow) or ‘V’-notch weirs (outflow). Automatically measured flows were validated by regular (usually weekly) manual flow measurements using a 10 L bucket and a stopwatch. A piezometer (Ponsel Ponsts 32) was installed at the lower end of the reactive swale to monitor water level within the swale during operation. Hydraulic retention time (HRT) was determined at average flow conditions (6–8 m3/h) in February 2013 and at low flow conditions (∼1 m3/h) in May 2013 with tracer tests (addition of salt at inflow with subsequent measurement of electric conductivity at outflow).

Water samples were regularly (every 1–2 weeks) taken as grab samples at inflow and outflow of the site. General water quality parameters (temperature, oxygen content, pH, conductivity, redox conditions) were measured on-site using a YSI multiprobe meter. Cooled samples were sent the same day to an accredited laboratory and analysed for nitrate (NO3), nitrite (NO2), ammonium (NH4), phosphate (PO4) and dissolved organic carbon (DOC).

Technical-scale reactive swale

Results from the field-scale system described above were compared with results from technical-scale swales operated from 2010 until 2012 at a test site of the German Federal Environmental Agency (UBA) in Berlin. To evaluate the impact of temperature variation and HRT on removal, two parallel technical-scale swales (length: 8 m, width: 2 m, depth: 0.9 m) were each manually filled to a height of 0.8 m with a mixture of straw (550 kg) and bark mulch (1,050 kg) and operated at a water level of 0.4 m at HRTs of about 0.4 and 2.5 days, respectively (Figure 1 right). The swales were covered with a foil to prevent influence of precipitation.

Highly polluted drainage water was simulated by spiking bank-filtered pond water (: <1 mg L−1, phosphate (): <1 mg L−1, sulphate (): 180–260 mg L−1, ammonium (): <0.1 mg L−1, nitrite (): <0.1 mg L−1, DOC: 2–4 mg L−1) with nutrients (100 mg NO3 L−1, 5 mg- L−1). Samples were taken weekly.

RESULTS AND DISCUSSION

The water quality of the inflowing water at the full-size reactive swale in Brittany was characterized by high and stable nitrate concentrations, around 55 mg-NO3 L−1 (12.4 mg-N L−1), and low concentrations of other nitrogen forms (e.g. ammonium, nitrite), DOC and phosphate (see Table 1 and Figure 3). Inflow water temperature varied between 6 and 11 °C (Figure 3), as the flow was usually restricted to winter and spring months (December–May). While temperatures were below 9 °C during most of the monitoring, they increased in springtime after mid-April. As low temperatures negatively affect microbial processes, the temperature range observed at both sites was likely to have limited denitrification performance.

Table 1

Water quality parameters in inflow and outflow of reactive swale in Brittany

 NO3 [mg/L] NH4 [mg/L] NO2 [mg/L] DOC [mg/L] PO4 [mg/L] Conductance [μS/cm] O2 [%] Temperature [°C] pH 
Inflow (mean) ± SD 54.2 ± 4.7 0.06 ± 0.06 0.11 ± 0.09 2.4 ± 0.7 0.16 ± 0.07 364 ± 29 81 ± 14 9.3 ± 1.7 7.3 ± 0.3 
Outflow (mean) ± SD 47.6 ± 12.0 0.08 ± 0.12 0.15 ± 0.17 2.9 ± 0.7 0.10 ± 0.05 354 ± 7 72 ± 18 9.0 ± 2.0 7.2 ± 0.3 
 NO3 [mg/L] NH4 [mg/L] NO2 [mg/L] DOC [mg/L] PO4 [mg/L] Conductance [μS/cm] O2 [%] Temperature [°C] pH 
Inflow (mean) ± SD 54.2 ± 4.7 0.06 ± 0.06 0.11 ± 0.09 2.4 ± 0.7 0.16 ± 0.07 364 ± 29 81 ± 14 9.3 ± 1.7 7.3 ± 0.3 
Outflow (mean) ± SD 47.6 ± 12.0 0.08 ± 0.12 0.15 ± 0.17 2.9 ± 0.7 0.10 ± 0.05 354 ± 7 72 ± 18 9.0 ± 2.0 7.2 ± 0.3 
Figure 3

Nitrate concentrations and water temperature in inflow to reactive swale during monitoring season 2013.

Figure 3

Nitrate concentrations and water temperature in inflow to reactive swale during monitoring season 2013.

Inflow varied mainly between 5 and 15 m3/h during most of the season (mid-December–mid-April) with higher peaks after intense rain events (especially on 12 March, with a total rain­fall of 37 mm) reaching up to 55 m3/h (Figure 4). The average flow (determined from weekly averages) for the whole flow season 2013 (December 2012–May 2013) was 9.3 m3/h. These values were much higher than inflows assumed for design purposes of ∼1 m3/h, resulting in short residence times (see below). Manual measurements of inflow and outflow were conducted regularly during the period of moni­toring to confirm continuous flow measurements (Figure 4). Furthermore, it can be seen that outflow equals inflow, indi­ca­ting that no losses or additional inflows to the system are likely (Figure 4). Results of the piezometer measurements show that the ditch was filled (saturated conditions) during the whole flow season (data not shown).

Figure 4

Inflow and outflow of reactive swale in Brittany during monitoring season 2013. Hydraulic retention time on secondary axis. Upper graph shows daily rainfall at climate station in Tremuson (∼5 km from sites).

Figure 4

Inflow and outflow of reactive swale in Brittany during monitoring season 2013. Hydraulic retention time on secondary axis. Upper graph shows daily rainfall at climate station in Tremuson (∼5 km from sites).

Hydraulic retention times (HRT) determined by tracer tests resulted in values of 1.6 h at average flow conditions in February and 7.1 h at low flow of 1 m3/h in May. The resulting hydraulic efficiency, considering a saturated volume determined from a filling experiment of 12.9 m3, was 67%, indicating some inefficiency in hydraulic retention (preferential flow). Both HRT values were used to derive a relationship between inflow and HRT (regression to power function) to determine HRT of other inflows as shown in Figure 4. It can be seen that HRT was low during most of the monitoring season (until April) with values <2 h. When flow decreased below 1 m3/h (after mid-April), HRT increased >0.4 day (Figure 4). Average HRT for the whole flow season 2013 was just 1.1 h (0.044 day) which was much lower than HRT assumed for design (minimum HRT of 0.5 day), mainly due to higher than assumed inflows and to some extent because of preferential flow within the system.

Nitrate removal was low (<5%) at nitrate outflow concentrations around 52 mg/L for most of the season, until inflow dropped in April with subsequent increase of hydraulic retention time (nitrate removal >25% at residence times >10 h = 0.4 day) and increasing water temperatures (see Figure 3). At the end of the flow season (end of May), at very low inflows and HRT >2 days, nitrate removal increased above 50% (Figure 5 left), demonstrating that systems are generally capable of reducing nitrate concentrations if designed properly or run under more favorable conditions (e.g. lower inflow, higher T). However, nitrate loads that were reduced at very low inflows and resulting high HRT were small in comparison to inflow loads during average inflow (Figure 6). Other water quality parameters (e.g. pH, O2, PO4) in the outflow of the full-size swale are shown in Table 1 together with inflow values. Only small changes can be noticed between inflow and outflow; most noticeable (beside nitrate reduction) is a reduction of phosphate, however at a low level of concentration below 0.2 mg/L.

Figure 5

Nitrate removal vs. HRT (left) and temperature (right) in full-size reactive swale.

Figure 5

Nitrate removal vs. HRT (left) and temperature (right) in full-size reactive swale.

Figure 6

Load reduction in relation to HRT in full-size reactive swale in Brittany.

Figure 6

Load reduction in relation to HRT in full-size reactive swale in Brittany.

As two favorable processes for denitrification occur simultaneously during springtime (increasing water temperature up to 11 °C due to seasonal changes and increasing hydraulic retention times caused by decreasing inflows), we investigated whether HRT or temperature was the driving parameter for the increasing nitrate removal after mid-April. In Figure 5 (left) nitrate removal is displayed as a function of HRT showing increasing nitrate removal at increasing HRT in a linear relationship with high correlation coefficient (R2 = 0.89). In comparison, Figure 5 (right) shows nitrate removal plotted against water temperature. It can be seen that the general trend is an increase in nitrate removal with increasing temperature. However, the correlation is much weaker compared to HRT (R2 = 0.33, e.g. very different retentions of 10% and 60% for the same temperature of 9.5 °C), suggesting that HRT is more relevant than temperature within the observed temperature range. Linear multi-parameter regression analysis (retention = a·HRT + b·Temp) showed a three times higher factor for HRT (a = 6.5, b = 2.3), also confirming that HRT is more relevant. However, as temperature is a driving parameter for microbial activity, further increase of temperature above 11 °C (e.g. to 20 °C or more) is likely to positively affect nitrate removal (see also temperature dependence in technical-scale results at UBA in Figure 7), potentially exceeding the effect caused by HRT increase.

Figure 7

Comparison of selected results for nitrate removal in reactive swale in Brittany with results of two technical-scale swales run at HRTs of 0.4 and 2.5 days (aged substrate in second year of operation in autumn/winter 2011) at UBA.

Figure 7

Comparison of selected results for nitrate removal in reactive swale in Brittany with results of two technical-scale swales run at HRTs of 0.4 and 2.5 days (aged substrate in second year of operation in autumn/winter 2011) at UBA.

Results from the full-size system described above were compared to results from technical-scale swales operated from 2010 until 2012 at a test site of the UBA in Berlin (Krause Camilo et al. 2014). While hydraulic retention times in the infiltration swale in Brittany were lower for most of the season (see Figure 4), two points at the end of flow season with HRTs of 0.4 and 2.3 days could be compared with nitrate removals in the technical swales. Figure 7 shows nitrate removal relative to temperature for both technical swales. The two values obtained from the infiltration ditch in Brittany are comparable to technical swale results (when comparing similar temperatures) despite different designs and substrates (see the triangles in Figure 7). This indicates that up-scaling is a suitable approach to transferring knowledge gathered from technical-scale experiments to field conditions. However, as for most of the flow season, HRT in the full-size reactive swale was much less than 0.4 days. Therefore, no nitrate removal at other temperatures are available for comparison.

Daily load reductions of nitrate were calculated from flow data and relative nitrate removal (Figure 6). Reduced nitrate loads were mostly below 0.5 kg NO3/d (=1.7 g NO3-N/m3/d, with Vswale = 66 m3), which is low compared to the results of technical swales at UBA, which have resulted in retained nitrate loads of 8 g NO3-N/m3/d at 9 °C and HRT of 0.4 days (Krause Camilo et al. 2014). The higher efficiency of technical swales is likely due to differences in layer design as the technical swales were entirely filled with reactive material (bark mulch/straw mixture) allowing a more efficient use of reactor volume for denitrification, whereas the reactive material of the infiltration ditch in Brittany comprised an organic substrate layer of 15 cm above 70 cm of sand and gravel.

CONCLUSIONS

The full-size reactive swale constructed in a former drainage ditch as a special form of near-natural mitigation system, receiving inflow from nearby agricultural fields, was successfully operated and monitored for 6 months in its third year after construction. As inflow to the full-size reactive swale was much higher than inflow considered in the design for most of the flow season (and therefore HRT was much lower than the minimum design retention time of 0.5 day), nitrate removal was lower than expected. However, when HRT increased at the end of the flow season due to decreasing inflow volumes, nitrate removal increased considerably, resulting in nitrate removals of 20–60% at HRT between 0.4 and 2.5 days, despite low water temperatures around 10 °C. This demonstrates that nitrate removal in reactive swales at low water temperatures is possible. However, as treated water volumes during times of higher HRT (low inflow) are low, reduced nitrate loads in the field system are low as well.

Furthermore, nitrate removal values observed in the full-size swale are comparable to results of technical-scale reactive swales filled with straw and bark mulch at similar temperatures and substrate ages (>1 year). Specific nitrate load reductions for the technical-scale swales were higher, though, likely due to a higher fraction of organic substrate compared to the design of the full-size swale, which only contained a 15 cm layer of organic substrate (see Figure 2). Therefore, for the design of new systems a higher percentage of organic substrate (e.g. bark/straw mixture) should be considered for effective nitrate removal at low temperatures and low HRTs. In addition, for the design of new mitigation systems the authors recommend investigating expected inflow volumes carefully in advance to ensure an appropriate residence time for effective nitrate reduction at low temperatures.

REFERENCES

REFERENCES
Alexander
R. B.
Smith
R. A.
Schwarz
G. E.
Boyer
E. W.
Nolan
J. V.
Brakebill
J. W.
2008
Differences in phosphorus and nitrogen delivery to the Gulf of Mexico from the Mississippi river basin
.
Environmental Science and Technology
42
(
3
),
822
830
.
Bernot
M. J.
Tank
J. L.
Royer
T. V.
David
M. B.
2006
Nutrient uptake in streams draining agricultural catchments of the midwestern United States
.
Freshwater Biology
51
(
3
),
499
509
.
Di
H. J.
Cameron
K. C.
2002
Nitrate leaching in temperate agroecosystems: sources, factors and mitigating strategies
.
Nutrient Cycling in Agroecosystems
64
(
3
),
237
256
.
DWA (Deutsche Vereinigung für Wasserwirtschaft, Abwasser und Abfall e.V.)
2008
Dränung. Nährstoffausträge, Flächenerfassung und Management
(Drainage: Nutrient export, catchment areas and management). DWA
,
Hennef
,
Germany
.
EPA (Environmental Protection Agency)
2002
National Water Quality Inventory – 2000
.
Report to Congress
,
Office of Water
,
Washington, DC
,
USA
.
European Environment Agency
2005
The European Environment – State and Outlook 2005
.
EEA
,
Copenhagen
,
Denmark
.
Kadlec
R. H.
Wallace
S. D.
2009
Treatment Wetlands
,
2nd edn
,
CRC Press
,
Boca Raton, FL
.
Krause Camilo
B.
Matzinger
A.
Litz
N.
Tedesco
L. P.
Wessolek
G.
2013
Concurrent nitrate and atrazine retention in bioreactors of straw and bark mulch at short hydraulic residence times
.
Ecological Engineering
55
,
101
113
.
Krause Camilo
B.
Heise
S.
Litz
N.
2014
Report on Efficient Design of Mitigation Zones for Pesticide Removal
.
KompetenzZentrum Wasser Berlin gGmbH
,
Berlin
(accessible from first author)
.
Périllon
C.
Matzinger
A.
2010
Identification of Existing Mitigation Systems That Can Attenuate Nitrates during High Flow Events from Drained, Agricultural Fields
.
KompetenzZentrum Wasser Berlin gGmbH
,
Berlin
.
Rouault
P.
Krause Camilo
B.
Soyeux
E.
Pagotto
C.
Perillon
C.
Wicke
D.
Breuer
L.
Litz
N.
2012
Mitigation systems to attenuate diffuse agricultural pollution: location and design choice
. In:
Remediation Technologies and Their Integration in Water Management: Proceedings of the 1st European Symposium, September, 25–26, 2012, Barcelona, Spain
,
Bastiaens
L.
(ed.),
VITO
,
Mol, Belgium
, pp.
25
26
.
Schipper
L. A.
Robertson
W. D.
Gold
A. J.
Jaynes
D. B.
Cameron
S. C.
2010
Denitrifying bioreactors – an approach for reducing nitrate loads to receiving waters
.
Ecological Engineering
36
,
1532
1543
.
Vymazal
J.
2007
Removal of nutrients in various types of constructed wetlands
.
Science of the Total Environment
380
,
48
65
.