Cost effectiveness, nitrogen and phosphorus removal in field-based woodchip bioreactors treating agricultural drainage water

Abstract

Nitrogen (N) and phosphorus (P) losses to surface and coastal waters are still critically high across Europe and globally. Measures to mitigate and reduce these losses are presently implemented both at the cultivated land surface and at the edge-of-fields. Woodchip bioreactors represent a novel alternative for treating agricultural drainage water, and the present study based on two years of data from five Danish field-based bioreactors determined N removal rates varying from 1.49 to 5.37 g N m − 3 d − 1 and a mean across all bioreactors and years of 2.90 g N m − 3 d − 1 . The loss of phosphorus was relatively high the first year after bioreactor establishment with rates varying from 298.4 to 890.8 mg P m − 3 d − 1 , but in the second year the rates varied from 12.2 to 77.2 mg P m − 3 d − 1 . The investments and the costs of the bioreactors were larger than expected based on Danish standard investments. The cost efficiency analysis found the key issues to be the need for larger investments in the bioreactor itself combined with higher advisory costs. For the four woodchip bioreactors considered in the cost efficiency analysis, the N removal cost was around DKK 350 per kg N (€ 47 per kg N), which is ca. 50% higher than the standard costs defined by the Danish authorities. Based on the estimated costs of the four bioreactor facilities included in this analysis, a bioreactor is one of the most expensive nitrogen measures compared to other mitigation tools.

Full text

Page 1/19
Cost effectiveness, nitrogen and phosphorus removal in eld-based
woodchip bioreactors treating agricultural drainage water
Finn Plauborg (  nn.plauborg@agro.au.dk )
Aarhus University
Maja Hørning Skjødt
Aarhus University
Joachim Audet
Aarhus University
Carl Christian Hoffmann
Aarhus University
Brian H. Jacobsen
University of Copenhagen
Research Article
Keywords: Environmental measures, Constructed wetlands, N removal, Residence time, Investment, Unit costs
Posted Date: September 28th, 2022
DOI: https://doi.org/10.21203/rs.3.rs-2037342/v1
License:   This work is licensed under a Creative Commons Attribution 4.0 International License.  Read Full License

Page 2/19
Abstract
Nitrogen (N) and phosphorus (P) losses to surface and coastal waters are still critically high across Europe and globally. Measures to
mitigate and reduce these losses are presently implemented both at the cultivated land surface and at the edge-of-elds. Woodchip
bioreactors represent a novel alternative for treating agricultural drainage water, and the present study based on two years of data from ve
Danish eld-based bioreactors determined N removal rates varying from 1.49 to 5.37 g N m− 3 d− 1 and a mean across all bioreactors and
years of 2.90 g N m− 3 d− 1. The loss of phosphorus was relatively high the rst year after bioreactor establishment with rates varying from
298.4 to 890.8 mg P m− 3 d− 1, but in the second year the rates varied from 12.2 to 77.2 mg P m− 3 d− 1. The investments and the costs of the
bioreactors were larger than expected based on Danish standard investments. The cost eciency analysis found the key issues to be the
need for larger investments in the bioreactor itself combined with higher advisory costs. For the four woodchip bioreactors considered in the
cost eciency analysis, the N removal cost was around DKK 350 per kg N (€ 47 per kg N), which is ca. 50% higher than the standard costs
dened by the Danish authorities. Based on the estimated costs of the four bioreactor facilities included in this analysis, a bioreactor is one
of the most expensive nitrogen measures compared to other mitigation tools.
Introduction
Agricultural production is a major source of diffuse pollution, which in Europe mostly is due to excessive emissions of nutrients (nitrogen
(N) and phosphorus (P)) and chemicals such as pesticides (EEA, 2018). Nutrient enrichment causes eutrophication, which in turn leads to
loss of aquatic biodiversity and reduction of sh stocks. Excessive nutrient enrichment can also endanger human health due to, for
instance, toxic algal blooms and can impair the use of water for drinking and bathing (EEA, 2018). Within the European Union (EU), member
states are implementing different kinds of measures to reduce diffuse nutrient pollution to comply with the EU Water Frame Directive (WFD,
2000). The measures at farm level include, among others, nutrient planning, fertiliser standards (e.g. timing), appropriate tillage, N-xing
and catch crops, buffer strips and crop rotation. Other tools focus on reducing the N loss from agricultural drainage water such as
constructed wetlands that capture and retain nutrient losses (EEA, 2018; Carstensen et al., 2020). In Denmark, more than 50% of the
country’s cultivated area is systematically tile drained (Olesen, 2009; Møller et al., 2018). The tile drainage systems act as a fast transport
route of dissolved chemicals, for instance nitrate (NO3), to recipient surface water bodies by bypassing the soil domain during the wet
season or after heavy rainfall events (Motarjemi et al., 2021). Diffusive losses of P with drainage water from agricultural areas to streams
and lakes have gained increased attention in Denmark as lakes are generally phosphorus sensitive (Andersen and Heckrath, 2020).
According to Dalgaard et al. (2014), in Denmark N losses to the aquatic and atmospheric environment have been signicantly reduced, but
complying with the WFD and Habitats Directives remains a major challenge that calls for new approaches, measures and technologies to
mitigate agricultural N losses and control N ows (Hoffmann et al., 2020). The establishment of free surface water constructed wetlands
and woodchip bioreactors was a central part of the so-called collective measures to be implemented from 2015–2021, and the plan was
that these would reduce N losses to the sea by 900 t N per year or 15% of the total expected reductions (MST, 2015). However, an
assessment made in 2020 showed that the expected effect at the end of 2021 would be only 332 t N per year due to slower implementation
than expected (Ministry of Environment and Food, 2020). In the draft version of the coming River Basin Management Plans it is expected
that constructed wetlands will generate a further reduction to the sea of 555 t N per year by 2027 (Ministry of Environment, 2021).
In a woodchip bioreactor drainage water is routed horizontally or vertically through a basin lled with woodchips before it reaches an outlet.
It is well established that woodchip bioreactors are ecient in removing N by denitrication and that the eciency increases with
increasing water temperature and hydraulic residence time (HRT) (Schipper et al., 2010; Christianson et al., 2012; Addy et al., 2016;
Hoffmann et al., 2019, Audet et al., 2021). As diffusive losses from agricultural areas of P in addition to N have detrimental effects on the
aquatic environment, the potential of P removal by woodchip bioreactors has gained increasing attention. For two 100 m3 woodchip
bioreactors, Carstensen et al. (2019) found that 67–85% of the annual loading of particulate P was retained, but in some years the
bioreactors acted as a sink and in other years as a source of phosphate. Thus, Gosch et al. (2020) found that a bioreactor (76.5 m3) turned
from acting as a phosphate source in the rst year of operation to a phosphate sink in the second and third year.
To gain wide acceptance as a relevant mitigation measure, it is important to determine the cost-effectiveness of woodchip bioreactors as
this will have to be evaluated against that of other mitigation measures used in Denmark and around the world (Eriksen et al., 2020,
Hoffmann et al., 2020). In particular, it is relevant to analyse to what extent the reduction costs of bioreactors per kg N and P compare with
the estimated standard costs used by the Ministry of Environment in Denmark.
Although the eciency of N removal by woodchip bioreactors is relatively well established, Addy et al. (2016) concluded in a meta-analysis
that more eld-based studies of the performance of woodchip bioreactors are needed to determine removal rates in different landscapes, at
different nitrate loadings and under different climate conditions. Hence, the main objectives of the present study were to i) present results

Page 3/19
on N and P removal of ve eld-based woodchip bioreactor facilities that receive agricultural drainage water and are located in different
geo-regions in Denmark and ii) to present the results of a cost-effectiveness analysis for each reactor facility.
Materials And Methods
Study sites and weather conditions
This study includes results from ve new bioreactor facilities located in Denmark (Fig.1). It was a prerequisite from the Agricultural Agency,
which funded the project, that the ve facilities should be established in different geo-regions in order to assess differences in the nal
construction design and performance at sites with different yearly precipitation amounts and patterns (Table S1); hence, each facility
received different hydraulic loads. Further, a prerequisite for the approval of the construction of the facility was that NO3− concentrations
above 5 mg N L− 1 were measured several times during the discharge season. Four facilities (Gyldenholm, Egsmarken, Dundelum and
Serupgård) were built during the rst half of 2018 and came into operation in autumn 2018, while Hofmansgave came into operation in
January 2019. The facilities at Gyldenholm and Hofmansgave consisted of three bioreactors (i.e. 3 subunits), while the other facilities only
had one bioreactor. The dimensions and design of each bioreactor are specied in Table1, and photos of the sites can be found in the
Supplemental information, Figs. S1-S5.The bioreactors had a horizontal ow regime except for the Dundelum reactor, which had a vertical
(top-down) ow design. The horizontal ow regime was controlled by xed levels in inlet and outlet wells (Fig. S6) or a dynamic build-up of
the hydraulic gradient with an increasing inlet rate because the diameter of the outlet pipe is smaller than the diameter of the inlet pipe
(Figs. S7-S8, Table1). The volume of the bioreactors varied from 271 to 652 m3, reecting differences in the drainage discharge from the
catchment areas. The bioreactors were not dimensioned to receive all the drainage water from the catchment, and at high ow a fraction of
the water was therefore bypassed by a manually set damper or pump at the inlet wells to prevent overloading of the bioreactor and ensure
an HRT of minimum 10. All bioreactors except Hofmansgave were equipped with a bentonite geo-membrane (John Hunderup Import &
Export, Denmark) to prevent exchange of water between the surrounding soil matrix and the bioreactor. At Hofmansgave, the membrane
was made of polypropylene (Junifol, Millag, Denmark) because the groundwater was saline. All bioreactors were lled with a layer of 1.2–
1.8 m lter matrix consisting of 100% willow woodchips (Ny Vraa I/S, DK, chip sizes 0.4–6 cm). This woodchip layer was dened as the
active wet lter matrix of the bioreactors. Subsequently, the wet lter matrix layer was topped with an additional unsaturated woodchip
layer of 0.3–0.5 m to allow methane oxidation (Carstensen et al., 2019). A sedimentation pond was located in front of the bioreactors at all
the facilities to act as a buffer at peak ow events and trap a fraction of the sediment transported in the tile drainage water towards the
bioreactor. All the bioreactors had a re-oxygenation system next to their outlets, consisting of a change in elevation to produce a vertical fall
to ensure that the water discharging from the bioreactors was not anoxic when it reached the recipient waters (often low order streams or
ditches).
The ve facilities were studied during two hydrological years (1 August 2018 to 31 July 2019 and 1 August 2019 to 31 July 2020), later
referred to as the hydrological years 2018–2019 and 2019–2020, respectively. Although the mean air temperature in the two hydrological
years was relatively similar, varying from 9.1 to 10°C, the “annual” precipitation differed markedly, being 657–1135 mm in 2018–2019 and
569–859 mm in 2019–2020 (Table S1).

Page 4/19
Table 1
Characteristics of the bioreactors at the different study sites.
Bioreactors Coordinates Dimensions Volume Flow
design* Total
drained
catchment
Relative
amount of
inlet
drainage
water treated
in the
bioreactor**
Volume
sedimentation
pond
Latitude
(°) Longitude
(°) W×L×D (m) m3Ha % m3
Gyldenholm A 55.331 11.457 16.2x21.1x1.4 477 Horizontal-
a120 10 4.7
B 55.3308 11.4571 16.2x21.1x1.4 477 Horizontal-
a17
C 55.3306 11.4571 16.2x21.1x1.4 477 Horizontal-
a30
Hofmansgave A 55.5387 10.4889 22.3x15.0x1.8 602 Horizontal-
a129 20 111
B 55.5386 10.4888 22.3x15.0x1.8 602 Horizontal-
a15
C 55.5385 10.4886 22.3x15.0x1.8 602 Horizontal-
a15
Egsmarken  55.234 10.5162 13.2x33.2x1.2 526 Horizontal-
b57 20 135
Dundelum  55.1551 9.4859 13.2x41.2x1.2 652 Vertical 45–60 33 81
Serupgård  56.3765 9.5984 5.8x36.0x1.3 271 Horizontal-
b80 13 76
* -a. Fixed hydraulic gradient by inlet and outlet well (Fig. S6). -b. Hydraulic gradient evolving dynamically (Figs. S7–S8).
** The bioreactor units and subunits did not treat all drainage water from the catchment but only the relative share indicated in the
column.
Monitoring methods
Flow and hydraulic residence time
To measure the water ow, the bioreactors were equipped with electromagnetic owmeters (Waterux 3070, Krohne, Germany) connected to
a data logger (Campbell CR1000, Campbell Sci., Logan, USA) that recorded mean ow velocity every 10 min. The owmeters were placed at
the inlet and outlets of the bioreactors at Hofmansgave and Gyldenholm but only at the outlet at Dundelum, Egsmarken and Serupgård.
The hydraulic residence time (HRT) in the present study was calculated as:
(Eq.1)
where is the water-saturated volume of the bioreactor (in m3), is the porosity of the bioreactor, and is the discharge of water in the
bioreactor (in m3) over a selected period of time (i.e., the whole period with active ow or per day in the present study). The water level was
kept constant as the outlet pipes were xed at a certain height and since the hydraulic conductivity in woodchips is very high, only peak
ow events might have resulted in a temporary, short increase (hours) of the water level. Porosity was set to 60% in agreement with Bruun
et al. (2016a, b).
Nutrients, oxygen and water temperature
We used a mass balance approach to determine the retention or release of solutes in the bioreactors. Flow proportional water samples (100
mL) were taken automatically at all bioreactors by use of a CR1000 datalogger and ISCO water samplers (Teledyne ISCO, Inc., NE, USA)
installed at the inlet and outlet. At Hofmansgave and Gyldenholm, the datalogger triggered the ISCO samplers to sample when the
accumulated water ow reached 720 m3 of water at the main inlet and for every 240 m3 of water from each of the outlets from the three
HRT
=
VBR
×
θ
/
Q
VBR θ Q

Page 5/19
subunits. The ISCO samplers kept the sampled water refrigerated at 4°C. At Egsmarken, Dundelum and Serupgård, the datalogger triggered
the ISCO sampler (ISCO 6712 Full-size Portable Sampler, Teledyne ISCO, Inc., NE, USA) to take a sample for every 120 m3 of ow from the
bioreactor. Samples were collected every second week and transported in an icebox to the laboratory. Upon arrival, the water samples were
kept at 4°C until analysis for total nitrogen (TN) and total phosphorus (TP). All TN samples were analysed using a Technicon AutoAnalyzer
according to DS 221 (1975) and TP by the photometric method according to DS 292 (1985). In this study, nutrient removal is expressed as
the quantity of nutrient removed per m3 of lter matrix per year (g N m− 3 yr− 1) and per day (g N m− 3 d− 1). The latter was obtained when
normalised with the number of days with ow (Table2).
Cost analysis
The purpose of the economic analyses was to estimate the required investment and the yearly costs of a bioreactor. Based on this and the
effect of the bioreactors, the cost eciency with respect to mainly nitrogen removal could be calculated and compared to other reduction
measures (Eriksen et al., 2020; Jacobsen and Gachango, 2013).
The economic calculations were based on the actual investments and expected running costs involved in the project. The estimated
lifetime of the investment was 20 years and the discount rate used was 4% following the socioeconomic guidelines from the Danish
Ministry of Finance (Finansministeriet, 2019). The data on investments, based on actual agreements made with sub-contractors, and
running costs were provided by Aarhus University (pers. communication, Finn Plauborg, Aarhus University). The investment costs covered
the construction of the facility, including, for instance, soil removal and pipe laying. In some cases, additional wells were established.
The lost income concerned the area taken out of production, here often around 0.3–0.4% of the catchment area, including a sedimentation
pond and a free area for moving the machinery used to maintain the bioreactors such as addition of supplementary woodchips over time.
The area taken out for free surface water constructed wetlands was around 1%. The loss of income was set to 253 € per ha, following the
national average (Eriksen et al., 2020). Should the project be stopped after 20 years, money is needed to re-establish the area, but this was
not included in the calculations. Also, if the project is continued after 20 years, pond redesign may be required.
The running costs included additional biomaterial (woodchips), supposing addition of 0.5 m woodchips on top of the bioreactor in years 6,
12 and 18 to compensate for loss of material (following the approach adopted in Eriksen et al., 2020). This does not mean that the
woodchips might not be fully used in year 20, but in some cases woodchip addition will be required every ve years. For some locations, the
running costs of the bioreactor also included ushing the inlet pipes every second year for sediments and algae blooms. In low-lying areas,
a pump may be needed, involving electricity costs. With respect to advisory costs, Aarhus University used both catchment ocers and
private consultants. It is dicult to establish whether Aarhus University have had more need for private consultants than most farmers, but
it is not unlikely as these sites were public projects.
In our analyses, the costs and the cost eciency were compared to the standard costs found in the guidelines from the Agricultural Agency
regarding support for new bioreactors (Agricultural Agency, 2020). These standard costs are based on 30 project estimates for 11
bioreactor locations (Agricultural Agency, 2019), resulting in the amounts shown in Supplementary Information Table S2, and they are used
to estimate the investment support that should be given to a constructed wetland project.
The standard costs also form the basis for the calculations in the Danish overview of nitrogen measures (the N catalogue), which
compares a range of possible measures to reduce nitrogen losses to the sea (Eriksen et al., 2020).
Results And Discussion
Performance of the bioreactors during hydrological years 2018 to 2020
Hydraulic residence time, water temperature and nutrient concentrations
The different bioreactors have different volumes and treat different relative amounts of inowing tile drainage water from the catchment
(Table 1).
Table2
Hydraulic residence time (HRT), water temperature and N and P concentrations (inlet) at the bioreactors.

Page 6/19
Bioreactors Monitoring
period HRT** Dayswith
ow Drainage water
temperature Inlet TN concentration Inlet TP concentration
  mean
[min] (h)  Yearly mean [min–
max] (°C) Yearly mean [min–max]
(mg L–1)
Yearly mean [min–max]
(mg L–1)
Gyldenholm A 2018–
2019* 43 [3] 36 6.3 [4.6–7.9] 20.6 [8.9–26.9]
 2019–2020 49 [3] 138 7.9 [6.2–12.3] 14.6 [12.7–17.1] 0.057 [0.025–0.21]
Gyldenholm B 2018–
2019* 125 [3] 30 6.5 [4.9–8] 20.2 [8.9–26.9]
 2019–2020 62 [2] 121 8.2 [6.5–12.7] 14.9 [12.7–17.1] 0.048 [0.025–0.21]
Gyldenholm C 2018–
2019* 96 [7] 138 7.2 [4.9–11.8] 21.6 [8.9–26.9]
 2019–2020 19 [6] 191 8.2 [6.5–12.3] 13.8 [12.7–17.1] 0.055 [0.025–0.21]
Hofmansgave
A2018–
2019** 92 [6] 69 7.8 [3.8–13.7] 11.2 [6–16.1] 0.045 [0.025–0.18]
 2019–
2020** 66 [8] 63 7.7 [4.5–13.7] 11.5 [7.6–12.7] 0.064 [0.04–0.11]
Hofmansgave
B2018–
2019** 112 [5] 46 8.7 [3.9–15.6] 11.3 [6–16.1] 0.046 [0.025–0.18]
 2019–
2020** 99 [6] 48 7.8 [4.5–13.7] 12.2 [7.6–12.7] 0.067 [0.04–0.11]
Hofmansgave
C2018–
2019** 99 [5] 49 7.8 [4–13.2] 10.3 [6–16.1] 0.045 [0.025–0.18]
 2019–
2020** 44 [6] 56 7.8 [4.5–13.7] 11.3 [7.6–12.7] 0.030 [0.04–0.11]
Egsmarken 2018–2019 209 [12] 107 6.4 [4.3–9.9] 12.2 [8.3–24.6] 0.070 [0.029–0.15]
 2019–2020 217 [4] 217 7.7 [5.5–12.7] 13.8 [9.9–20.7] 0.099 [0.026–0.31]
Dundelum 2018–2019 85 [4] 129 6.9 [4.6–11.2] 8.9 [7.1–13.4] 0.268 [0.065–1.4]
 2019–2020 33 [7] 202 6.4 [2.4–11.6] 10.0 [5.3–16.7] 0.245 [0.098–0.82]
Serupgård 2018–2019 143 [5] 129 6.1 [4.8–8.5] 28.2 [13.7–46.4] 0.269 [0.19–1.4]
 2019–2020 58 [5] 188 6.2 [4–10.8] 14.5 [5.6–21.5] 0.156 [0.083–0.39]
Mean  92 [5] 109 7.3 [4.7–11.9] 14.1 [8.6–20.1] 0.112 [0.045–0.347]
*Sampling of TP failed
**At Hofmansgave the facility was only in operation from February to April in both monitoring periods due to technical problems.
***Mean HRT was calculated for the whole monitoring period with ow (see column “days with ow”), while minimum HRT was based on
the daily HRT.
As a result, the yearly mean and min HRT – calculated from daily values – varied markedly between the bioreactors with a yearly mean HRT
of 19 h at bioreactor Gyldenholm C to 217 h at Egsmarken and with a yearly min HRT varying between 2 and 12 h (Table 2). At very low
inow rates, the yearly max HRT increased to unrealistic high values, and these are therefore omitted from Table 2. It should be noted that
the number of days with inow of tile drainage water to the bioreactors differed greatly between the ve sites, from 30 to 217 days (Table
2). 
At Gyldenholm, the number of days where the ow-through bioreactor C was in operation was higher than for reactors A and B as bioreactor
C had a smaller inlet pipe diameter and a lower inlet pipe level than A and B (Plauborg et al. 2021). The reason for this was to avoid a too
high HRT as this would promote release of hydrogen sulphide and methane (Carstensen et al. 2019). Also, the water temperature of the

Page 7/19
reactors differed due to differences in the number of ow days and between years. For all bioreactors, the average inlet yearly mean TN
concentration ranged between 9.6 and 24.7 mg N L-1, while the yearly mean TP varied from 0.035 to 0.54 P mg L-1 (Table 2).
Total nitrogen
In this and the following section, we present a synthesis of nutrient removal results. All the bioreactors removed TN during the study period,
average removal being 30.5 % of the TN load (Table3). TN removal rates in the treated water (i.e. not bypassed) varied largely among the
bioreactors (6-55%) and between the two hydrological years (Table 3). However, the hydraulic load of the bioreactors also showed large
variations (17-229 m3 m-3 bioreactor yr-1), and it should be noted that the hydrological year 2018-2019 was relatively dry, which also
impacted the TN load at the bioreactor inlets, except for Gyldenholm. We hypothesised that the dry period had lowered the removal rates in
2018-2019 and, as expected, there was a clear increase in TN load with an increasing hydraulic load to the bioreactors (Fig. 2, top). At the
same time there was an increase in TN removal rates (in g N m-3 d-1) at increasing hydraulic load to the bioreactors (and thus TN load) (Fig.
2, middle). However, a negative relationship was observed between TN load and TN removal eciency (as percent TN removal) (Fig. 2,
bottom). Hence, higher TN removal (in g N m-3 d-1) was achieved at lower eciency (percent TN). Thus, both the percent eciency and the
absolute TN removal should be considered when evaluating the annual, seasonal and daily performance of different bioreactors.
Table3
Water ow, total nitrogen (TN) load and TN removal eciency and rates per cubic metre active bioreactor.

Page 8/19
Bioreactors Hydrological
year Water
ow Yearly TN
load inlet TN removal
eciency Yearly TN
removal rate Daily TN
load inlet1
Daily TN
removal rate2
  m3 m-
3 yr-1
g N m-3 yr-1 %g N m-3 yr-1 g N m-3 d-1 g N
m-3 d-1
Gyldenholm A 2018-2019 86 1776 5.5 97.00 49.33 2.69
 2019-2020 117 1706 23.8 406.00 12.36 2.94
Gyldenholm B 2018-2019 55 1111 8.8 97.50 37.03 3.25
 2019-2020 229 3405 19.1 649.50 28.14 5.37
Gyldenholm
C 2018-2019 139 3012 20.5 617.00 21.83 4.47
 2019-2020 229 3161 22.2 700.20 16.55 3.67
Hofmansgave
A2018-2019333 368 30.2 111.00 5.33 1.61
2019-2020350 579 26.6 154.00 9.19 2.44
Hofmansgave
B2018-2019321 238 29.8 71.00 5.17 1.54
2019-2020327 329 41.9 138.00 6.85 2.88
Hofmansgave
C2018-2019317 175 54.6 95.50 3.57 1.95
2019-2020344 496 38.8 192.50 8.86 3.44
Egsmarken 2018-2019 40 487 43.9 214.00 3.78 1.66
 2019-2020 89 1224 32.9 402.50 6.06 1.99
Dundelum 2018-2019 44 392 49.0 192.00 3.04 1.49
2019-2020 170 1702 46.0 782.50 9.05 4.16
Serupgård 2018-2019 49 1383 31.4 434.00 12.93 4.06
 2019-2020 140 2025 23.4 473.50 9.33 2.18
Mean All years 88 1325 30.5 327.33 13.88 2.90
1) Calculated as the yearly load rate divided by number of days with water ow, cf. Table 2.
2)Calculated as the yearly removal rate divided by the number of days with water ow, cf. Table 2.
3) The Hofmansgave facility was only in operation from February to April in both monitoring periods due to technical problems.
The bioreactors were designed to enable control of the inlet rate, aiming at a minimum HRT of 10 h. For the reactors at Serupgård,
Dundelum, Egsmarken and Hofmansgave, control involved manual change of a shutter and at Gyldenholm by modication of the pumping
rate. However, this control showed as evidenced by the yearly mean minimum HRT of 5 (Table 2). Even though the HRT were as low as 2 h
in some periods, the N removal in the eld-based bioreactors equalled that recorded for woodchip bioreactors in Denmark, USA and New
Zealand (Table 4). 
Table 4
Nitrogen removal rates measured in eld studies. The rst row is total nitrogen (TN), all other results show nitrate (NO3--N) removal.

Page 9/19
Reference Composition lter matrix Nitrogen removal rate
g N m-3 d-1
Monitoring
period
This project, all bioreactors, See the section Material and methods 1.49–5.37* 2 years
Hoffmann et al. (2019), horizontal ow Woodchips + mussel shells 50:50 2.03–2.10 2 years
Hoffmann et al. (2019), horizontal ow Woodchips + mussel shells 75:25 2.20–2.22 2 years
Hoffmann et al. (2019), vertical upwards ow Woodchips + mussel shells 50:50 1.67–2.14 2 years
Hoffmann et al. (2019), vertical upwards ow Woodchips + mussel shells 75:25 1.77–2.15 2 years
Hoffmann et al. (2019), vertical downwards ow Woodchips + mussel shells 50:50 2.16–2.21 2 years
Hoffmann et al. (2019), vertical downwards ow Woodchips + mussel shells 75:25 1.16–1.79 2 years
Christianson et al. (2012) Woodchips (60%) + gravel 0.38–3.78 7 years
Christianson et al. (2012) Woodchips (100%) 0.86–1.56 2 years
Christianson et al. (2012) Mixed hardwood chips 0.41–7.76 3 years
Christianson et al. (2012) Woodchips (100%) 0.42–5.02 2 years
Elgood et al. (2010)** Pine woodchips (100%) 0.69 1.1 years
Jaynes et al. (2008) Oak woodchips 0.62 5 years
Schipper et al. (2005) Pine sawdust (30%) + soil 1.4 9 days
Schipper et al. (2010) Coarse sawdust + woodchips 50:50 1.4 1 year
Schipper et al. (2010) Sawdust + woodchips 50:50 5–10 2 years
Schipper et al. (2010) Sawdust + woodchips 50:50 0–11 1.4 year
Schmidt & Clark (2012) Pine sawdust (50%) + quartz sand 4.9–5.5 1.8 years
van Driel et al. (2006) ** Fine and coarse wood particles 2.66 2.2 years
van Driel et al. (2006) ** Fine and coarse wood particles 0.76 1.7 years
Woli et al. (2010) Woodchips mixed 6.4 2-3 years
* Only TN was measured, but in general NO3--N constitutes about 90-95% of TN in Danish drainage water.
** Transformed to volumetric units
.

Our results hold true when compared with areas with higher temperatures (Schmidt & Clark, 2012, Florida; Woli et al. 2010, Illinois in
summer). The composition of the lter matrix differs between studies, but all matrices are based on wood products (sawdust or
woodchips) with or without addition of gravel, sand or soil. Overall, comparison of our results with international studies suggest that the
design of the Danish bioreactors provides a satisfactory eciency of NO3- removal from drainage water.
Total phosphorus
TP release was generally observed from the different bioreactors (Table 5).
Table 5
Total phosphorus load and removal at the bioreactors.

Page 10/19
Bioreactors Hydrological year Yearly TP
load inlet
TP removal
eciency Yearly TP
removal rate
Daily TP
load inlet
Daily TP
removal rate
g P m-3 yr-1 %g P m-3 yr-1 mg P m-3 d-1 mg P m-3 d-1
Gyldenholm A 2018-2019*
 2019-2020 6.7 -13 -0.87 48.6 -6.3
Gyldenholm B 2018-2019*
 2019-2020 11 -0.4 -0.04 90.9 -0.4
Gyldenholm C 2018-2019*
 2019-2020 13.7 52 7.12 71.7 37.3
Hofmansgave A 2018-2019** 1.50 -4097.9 -61.47 21.7 -890.8
 2019-2020** 3.20 -1730.9 -55.39 50.8 -879.2
Hofmansgave B 2018-2019** 0.96 -3549.8 -34.08 20.9 -740.8
 2019-2020** 1.80 -2233.3 -40.20 37.5 -837.5
Hofmansgave C 2018-2019** 0.77 -4992.6 -38.44 15.7 -784.6
 2019-2020** 1.30 -2626.0 -34.14 23.2 -609.6
Egsmarken 2018-2019 2.80 -1374.8 -38.49 21.7 -298.4
 2019-2020 8.80 27.9 2.46 43.6 12.2
Dundelum 2018-2019 11.80 -844.4 -99.64 91.5 -772.4
 2019-2020 41.70 34.8 14.51 221.8 77.2
Serupgård 2018-2019 13.20 -240.2 -31.71 123.4 -296.3
 2019-2020 21.80 43.4 9.46 100.5 43.6
Mean All years 9.40 -1436.4 -26.73 65.6 -396.4
* Sampling of TP failed.
**The Hofmansgave facility was only in operation from February to April in both monitoring periods due to technical problems. In the start
of the second year, all woodchips were removed and stored, and a new membrane was installed. Hereafter, woodchips were lled into the
bioreactor again.
However, at Serupgård, Egsmarken and Dundelum, this release was especially noticeable in the rst months following bioreactor
establishment, as shown for Serupgård (Fig. 3). Overall, the bioreactors monitored in the project had a mean TP removal of -26.73 g P m-3
yr-1 or -396.4 mg P m-3 d-1,indicating P-loss to the recipients.
Overall, the TP loss from the bioreactors varied between 31.71 and 99.64 g P m-3 year-1 in their rst year of operation. This is in line with the
results of Carstensen et al. (2019), who in a 5-year study of bioreactors also found loss of TP in the rst year, which could be attributed to
signicant of PO43--P losses, while particulate P was retained during the whole 5-year monitoring period. The bioreactors in the present
study were established in spring 2018, a dry year with almost absence of precipitation between May and November. Leaving a bioreactor
with fresh and moist woodchips under aerobic conditions in a dry summer and an autumn with hardly no rainfall may have led to fast
mineralisation of easily digestible organic matter in the woodchips, augmenting even more the temperature in the lter matrix and speeding
up the mineralisation process. As displayed in Fig. 3, the release of TP decreased to zero over a relatively short period (2-3 months),
whereas release of PO43--P lasted up to nine months in the study by Carstensen et al. (2019). At Hofmansgave, the TP loss was observed in
both years, probably due to the technical problems and the handling of the woodchips. Thus, at the start of the second year, all woodchips
had to be removed and placed on the soil surface under aerobic conditions when a new membrane was installed, after which the
woodchips were backlled. Overall, the TP release seemed limited in time and several bioreactors turned from being a P source during the
rst year to being a P sink in the remaining monitoring period (Table 4), which is in agreement with Carstensen et al. (2019). However, the
TP release in the rst year was still substantial and could potentially impact vulnerable downstream recipients. It is therefore important to

Page 11/19
take into consideration the risk of P release for downstream waterbodies, especially when implementing several bioreactors in the same
catchment. To mitigate this P release, use of technical solutions such as the addition of material to trap P (e.g. sand lter) may be a
possibility. Further, in temperate regions, bioreactors should be established in autumn or winter, the optimal timing being immediately
before the start of the drainage season. Hence, most of the woodchips will be water saturated immediately after which anaerobic processes
will take over.
Cost eciency analysis
In the economic analyses, Gyldenholm was treated as one bioreactor and Hofmansgave was left out as the data did not cover two
hydrological years due to shift of the membrane (see Table S3 basic costs data). As shown in Table S3, the relative size of the reactor was
around 0.3% of the effective catchment area, which is a bit larger than in the N catalogue where the fraction is set to 0.2% of the catchment
area (Eriksen et al., 2020). The size of the bioreactor relative to the catchment area is still quite low compared to that of constructed open
wetlands where the wetland area constitutes around 1% of the catchment area. The results on bioreactor eciency presented here are
based on only two years of data from the four bioreactors. To consolidate the results and hence the economic calculations, several more
years of monitoring are required. 
The average height of the biomaterial (woodchips) used in our cost analysis was 1.7 m due to the requirement of an unsaturated layer of
0.3–0.5 m woodchips on the top to avoid CH4 emission. With time, the volume of the biomaterial will shrink, requiring addition of another
0.5 m layer of biomaterial roughly every 6 years. The average price for of biomaterial used is DKK134 tonne, which is 8% cheaper than the
price level used in the N catalogue of DKK 145 tonne (Eriksen et al., 2020).
The investments are described in more detail in Table S4, the costs related to the establishment of the work site being the rst. The
establishment costs are often independent of the size of the reactor, whereas the others investments often increase with reactor size. In our
calculations, additional investments were included to cover extra wells (ushing wells), biomaterial and costs related to advisory tasks. It
should be noted that costs related to permissions and archaeological diggings were not included; they are frequently an issue and therefore
included in the standard costs. The consultancy fee is around 20% of the costs, but as no two sites are the same, the need for advice differs
from site to site. 
At Gyldenholm, it was not considered necessary to install a pump, but it appeared that a pump was in actual fact needed as the main inlet
drain pipe was buried surprisingly deep into the ground. The installation of a frequency pump in a 7-m deep well costed DKK 496,000,
resulting in a yearly cost of DKK 36,500, which can be perceived as an additional but perhaps not required cost.
In the calculations, all costs – except those related to consultants, paths and permissions – were scaled according to size. To further
improve the scaling, an initial investment of DKK 105,000 was assumed irrespective of size, based on the initial investments shown in
Table S4. The additional investments were then scaled according to size. The costs related to archaeological digs were scaled following the
agricultural agency guideline, involving a tendency of cost overestimation for smaller bioreactors when these were converted to a size of
0.2 ha.
The advisory part includes a description of the idea, inspection of the site and elaborating a draft project layout. It also comprises
elaboration of a detailed project plan, contract agreements with contractors and project follow-up until the nal delivery, including
coordination between contractors, but the costs stated by the contractors do not include application to authorities. In our case, catchment
advisors were in charge of nding locations and getting permissions. For other projects, these costs could be paid by the system of special
advisors helping to promote constructed wetlands. In our case, the service provided by the catchment ocers was paid by the research
project, but it is omitted from the other calculations as it is free to farmers. 
Table 6
Annual costs of a bioreactor (DKK yr-1)

Page 12/19
Name of reactor Serupgård Egsmarken Dundelum Gyldenholm Share of total (%)
Bioreactor and initial biomaterial 26,548 45,855 47,883 112,067 57
Additional well 756 1,041 0 0 0
Additional pump 0 0 0 36,504 9
Additional biomaterial 4,458 9,420 12,582 20,125 11
Electricity etc. 0 0 0 20,000 5
Cleaning pipes 1,500 1,500 1,500 1,000 1
Loss of income 102 198 246 539 0
Consultant  11,916 10,531 10,531 29,727 15
Total costs 45,280 68,546 72,742 219,962 100
Total costs without pump 45,280 68,546 72,742 163,458 86
Note: For Gyldenholm, the running costs related to pump electricity was DKK 20,000 yr-1 and moving of the lawn DKK 1000. For Serupgård,
Egsmarken and Dundelum, the running costs were DKK 1,500 y-1 for drain cleaning.
Note: DKK 100 = € 13.44 = $ 14.41 (exchange rate on 25.4.2022) 
Table 6 shows the calculated annual costs of which the largest are related to the initial investment in bioreactor construction and
biomaterial, while the second largest item is the advisory costs. The additional purchase of more biomaterial comes next. If a pump is
needed, the costs increase as is the case for Gyldenholm where pump investment and electricity increase the costs, potentially also the
consultancy costs.
The costs eciency gures presented in Table 7 show some but not large differences in the N reduction (standard effect for constructed
wetlands (CW) of 0.2 ha) between the sites when they are scaled to the same size. It is, however, worth noticing that the N effect at three of
the four sites differs between the two years. Thus, the effect in year 2 is much higher at these three sites, which is partly due to the weather,
but the removal percentage in year 1 and year 2 is fairly similar. Some of the basins at Gyldenholm have a low removal percentage,
reducing the average. 
The P effect is negative in the rst year and subsequently positive (see Table 7). It will, therefore, often take a number of years before the
amount removed will equal the amount of P lost during year 1. The P reduction is limited to around 1-9 kg P per year, and the P removal
ranges between 21 and 43%.The P cost eciency calculations range from DKK 8,300 to 7,300 per kg P based on year 2.
Table 7
N and P reduction and cost eciency for woodchip bioreactors.
Name of reactor Serupgård Egsmarken Dundelum Gyldenholm
Reactor size (m2)271 526 652 1431
TN-reduction (kg N/year) 123 162 266 612
Standard effect (kg N/yr/0.2 CW ha) 908 617 817 855
TN-Removal % 27 38 48 17
TP-reduction 1st year (kg P/yr) -9 -20 -70 -30
TP-reduction 2nd year (kg P/yr) 3 1 9 3
Cost eciency (DKK/kg N) 368 423 273 359
Cost eciency (DKK/kg N)without pump 368 423 273 300
Cost eciency (DKK/kg P) 17.415 52.727 8.266 73.321

Page 13/19
Note: The P effect is typically negative for the rst year since P from the biomaterial is released into the nearby stream, so year 2 effect has
been used instead. 
In the present study, the costs per kg N removed varied from DKK 273 to 423 (Table 7). For a bioreactor in Australia,White et al. (2022)
found expenses of AUS$ 17.83 (DKK 95) per kg NO3-N removed, which is around 22 to 32% of the costs calculated in our study. This is
partly due to a much lower investment in the Australian study (DKK 35,000 compared to around DKK 500,000) than the investment in the
Danish bioreactors (ca. 600 m2in size). Moreover, costs related to replacement of biomaterial and advisory costs were not included in the
analysis by White et al. (2022). However, further studies are needed to compare the applied factors in the cost effectiveness analyses to
consolidate the big differences.
We compared the project costs to the standard costs reported by Danish authorities (see Table S5), including the costs of woodchip
bioreactors given in the Danish guidelines for the smallest (500 m2) and largest site (2500 m2) (Eriksen et al., 2020). These guideline costs
were included as there is a low number of existing woodchip bioreactors, implying that the guideline costs could help to improve standard
gures as well as identify where the costs of the analysed sites were larger than expected. In the comparison, the bioreactors were all
adjusted to the same size (0.2 ha). 
In general, the investments were larger than expected even when using the smallest case of 500 m2 in the N catalogue (Eriksen et al., 2020).
They were also much larger than anticipated in previous analyses (Jacobsen and Gachango, 2013). This might indicate that in reality the
investments may be higher than expected although the investments based on the standard costs given in Table S2 are also derived from
real projects. In some cases, additional wells and pumps can generate additional costs, which may signicantly increase the costs per year.
It seems that the advisory costs in the four projects were much larger than the standard costs. This could perhaps be due to higher
complexity and limited help from the catchment ocers as ours was also a research project. Catchment ocers do indicate that the
administrative costs included in the standard costs in Table S2 only cover the most required tasks and thus not all project coordination
issues.
In Fig. 4, investments in the project bioreactors were compared with the basic investments in the Danish Agricultural Agency guideline
(Table S5). The smaller reactors required the lowest investment, but the investments in small reactors are the highest in the same-size
comparisons (0.2 ha).
The larger investments in the projects analysed in this paper could be related to the adopted pilot approach and the fact that the companies
involved had not engaged in this type of project before. Our project shows that no two sites are the same; thus, identication and
management of the best location may take time. Furthermore, the costs in the guide from the Agricultural Agency may be on the low side to
avoid overcompensation.
The initial bioreactor is supplemented with 0.5 m biomaterial in years 6, 12 and 18. Catchments ocers suggested that the initial level of
biomaterial should rather be 2.25 m (and not 1.7 m) to follow the requirements set by the Agricultural Agency. The advisory costs are much
larger than expected, possibly related to the pilot nature of the project. Aarhus University had almost the same help from catchment ocers
that farmers might have received. Normally, farmers undertake project coordination themselves, but in our project this was carried out by a
consultant at a cost that the farmer might not be compensated for. This indicates that in many projects the consultancy costs may be
higher than the amount included in the standard costs. This is also because the farmer does not want to bear the responsibility for the nal
acceptance of the project by the Agricultural Agency.
Conclusions
The performance of ve eld-based bioreactor facilities, two bioreactors with three subunits and three with just one single bioreactor, was
reported for two hydrological years. The TN removal rate varied from 1.49 to 5.37 g N m-3 d-1 with an overall average of 2.90 g N m-3 d-1.
The yearly N removal eciency varied from 6 to 55%, with a mean of 30.5% across bioreactors and years. In the rst year of operation, the
bioreactors had a TP loss rate of 298.4 to 890.8 mg P m-3 d-1. However, in the second year, most bioreactors retained phosphorous at rates
varying from 12.2 to 77.2 mg P m-3 d-1.
Overall, the annual costs of the four bioreactors in the present project were about twice the average costs compared to other nitrogen
reduction measures. This was mainly due to larger investments and higher consultancy costs and, in some cases, also investment in an
additional pump. The reduction costs per kg N (with pump) based on the actual size was, on average, ca. DKK 350 per kg N, which is
around 50% higher than the standard costs used by the Danish authorities for bioreactor establishment.

Page 14/19
The cost eciency linked to the P removal in year 2 showed large variations from around DKK 8,000 to 73,000 per kg P for the four
bioreactor facilities.
Declarations
FUNDING
The project with all its results and this manuscript were funded by Ministry of Environment and Food of Denmark and Department of
Agroecology, Aarhus University.
AVAILABILITY OF DATA AND MATERIAL
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable
request.
DECLARATIONS
Conict of interest The author declare no competing interests.
AUTHOR CONTRIBUTION STATEMENT
F. Plauborg, J. Audet, C.C. Hoffmann and B. Jacobsen wrote the main text of the manuscript. J. Audet and M.H. Skjødt provided the Tables
and Figures. All authors reviewed the manuscript.
References
1. Addy, K., A.J. Gold, L.E. Christianson, M.B. David, L.A. Schipper, and N.A. Ratigan. 2016. Denitrifying bioreactors for nitrate removal: A
meta-analysis. J. Environ. Qual. 45:873–881. https://doi.org/10.2134/jeq2015.07.0399.
2. Agricultural Agency. 2019. Standardomkostninger for minivådområder med ltermatrice - Beregning af standardomkostninger på
baggrund af konkrete projekter. Landbrugsstyrelsen, EU & Erhverv, Ref. September 2019.
3. Agricultural Agency. 2020. Minivådområdeordningen 2020 – Etablering af åbne minivådområder og minivådområder med ltermatrice.
https://lbst.dk/leadmin/user_upload/NaturErhverv/Filer/Landbrug/Natur_og_miljoe/Minivaadomraader/Minivaadomraadevejledning-
2020.pdf
4. Andersen, H.E., Heckrath, G. (redaktører). 2020. Fosforkortlægning af dyrkningsjord og vandområder i Danmark. Aarhus Universitet,
DCE – Nationalt Center for Miljø og Energi, 340 s. -Videnskabelig rapport nr. 397. http://dce2.au.dk/pub/SR397.pdf
5. Audet, J., Jeglot, A., Elsgaard, L., Maagaard, A.L., Sorensen, S.R., Zak, D., Hoffmann, C.C. 2021. Nitrogen removal and nitrous oxide
emissions from woodchip bioreactors treating agricultural drainage waters. Ecological Engineering 169.
https://doi.org/10.1016/j.ecoleng.2021.106328.
. Bruun, J. D., Kjærgaard, C. & Hoffmann, C. C. 2016a. Nitrogen removal in permeable woodchips lters affected by hydraulic loading
rate and woodchips ratio. Journal of Environmental Quality 45(5), 1688-1695.
7. Bruun, J. D., Hoffmann, C. C., Kjærgaard, C. & Pugliese, L. 2016b. Solute transport and nitrate removal in full scale subsurface ow
constructed wetlands treating agricultural drainage water. Ecological Engineering, 97, 88-97. DOI: 10.1016/ j.ecoleng.2016.07.010
. Carstensen, M.V., Larsen, S.E., Kjaergaard, C., Hoffmann, C.C. 2019. Reducing adverse side effects by seasonally lowering nitrate
removal in subsurface ow constructed wetlands. Journal of Environmental Management 240, 190-197.
https://doi.org/10.1016/j.jenvman.2019.03.081.
9. Carstensen, M. V., Hashemi, F., Hoffmann, C. C., Zak, D., Audet, J., & Kronvang, B. 2020. Eciency of mitigation measures targeting
nutrient losses from agricultural drainage systems: A review. Ambio, 49(11), 1820-1837.
https://link.springer.com/article/10.1007/s13280-020-01345-5
10. Christianson, L., Bhandari, A., Helmers, M., Kult, K., Sutphin, T., Wolf, R., 2012. Performance evaluation of four eld-scale agricultural
drainage denitrication bioreactors in Iowa. Trans. ASABE 55, 2163–2174.
11. Dalgaard, T., Hansen, B., Hasler, B., Hertel, O., Hutchings, N.J., Jacobsen, B.H., Jensen, L.S., Kronvang, B., Olesen, J.E., Schjorring, J.K.,
Kristensen, I.S., Graversgaard, M., Termansen, M., Vejre, H. 2014. Policies for agricultural nitrogen management-trends, challenges and
prospects for improved eciency in Denmark. Environmental Research Letters 9, 115002 (16pp). https://doi.org/10.1088/1748-
9326/9/11/115002.

Page 15/19
12. DS 221. 1975. Dansk Standard nr DS 221. Bestemmelse af nitrogen efter oxidation med peroxodisulfat. December 1975.
https://webshop.ds.dk/da-dk/standarder/standard/ds-2211975
13. DS 292. 1985. Dansk Standard nr DS 292. Total phosphor i vand. Marts 1985. https://webshop.ds.dk/da-dk/standard/ds-2921985
14. EEA. 2018. European Waters Assessment of Status and Pressures. EEA Report No 7/2018. European Environment Agency,
Luxembourg.
15. Elgood, Z., Robertson, W.D., Schiff, S.L., 2010. Greenhouse gas production in a stream bed bioreactor for nitrate removal. J. Ecol. Eng.
36, 1575–1580.
1. Eriksen, J., Thomsen, I. K., Hoffmann, C. C., Hasler, B., Jacobsen, B. H. 2020. Virkemidler til reduktion af kvælstofbelastningen af
vandmiljøet. Aarhus Universitet. DCA – Nationalt Center for Fødevarer og Jordbrug. 452 s. – DCA rapport nr. 174
https://dcapub.au.dk/djfpdf/DCArapport174.pdf
17. Finansministeriet. 2019. Vejledning i samfundsøkonomisk konsekvensvurdering. https://fm.dk/udgivelser/2017/august/vejledning-i-
samfundsoekonomiske-konsekvensvurderinger/
1. Gosch, L., Liu, H.J., Lennartz, B. 2020. Performance of a Woodchip Bioreactor for the Treatment of Nitrate-Laden Agricultural Drainage
Water in Northeastern Germany. Environments 7 (9). https://doi.org/10.3390/environments7090071
19. Hoffmann, C.C. Larsen, S.E. and Kjaergaard, C. 2019. Nitrogen removal in woodchip-based biolter of variable design treating
agricultural drainage discharge. Journal of Environmental Quality 48, 1881-1889. doi:10.2134/jeq2018.12.0442
20. Hoffmann, C.C., Zak, D., Kronvang, B., Kjaergaard, C., Carstensen, M.V., Audet, J. 2020. An overview of nutrient transport mitigation
measures for improvement of water quality in Denmark. Ecological Engineering 155. https://doi.org/10.1016/j.ecoleng.2020.105863
21. Jacobsen, B.H., Gachango, F.G. 2013. Vurdering af omkostningseffektiviteten ved inivådområder med inltrationsmatrice.
Frederiksberg: Institut for Fødevare- og Ressourceøkonomi, Københavns Universitet. IFRO Dokumentation, Nr. 2013/1 https://static-
curis.ku.dk/portal/les/122604294/IFRO_Dokumentation_2013_1.pdf
22. Jaynes, D.B., T.C. Kaspar, T.B. Moorman and T.B. Parkin. 2008. In situ bioreactors and deep drain-pipe installation to reduce nitrate
losses in articially drained elds. J. Environ. Qual. 37: 429-436.
23. Ministry of Environment and Food. 2020.
Vurdering af kvælstondsatsen
. Februar 2020.
https://mfvm.dk/leadmin/user_upload/MFVM/Landbrug/Afrapportering_af_kvaelstofudvalgets_arbejde.pdf (accessed 25.4.2022)
24. Ministry of Environment. 2021. Forslag til vandområdeplanerne 2021-2027. December 2021.
https://mim.dk/media/226716/vandomraadeplanerne-2021-2027.pdf (accessed 25.4.22)
25. Motarjemi, S.K., Rosenbom, A.E., Iversen, B.V., Plauborg, F. 2021. Important factors when simulating the water and nitrogen balance in a
tile-drained agricultural eld under long-term monitoring. Science of the Total Environment 787, 147610.
https://doi.org/10.1016/j.scitotenv.2021.147610.
2. MST. 2015. The Danish Environmental Protection Agency. https://eng.mst.dk/
27. Møller, A.B., Beucher, A., Iversen, B.V., Greve, M.H. 2018. Predicting articially drained areas by means of a selective model ensemble.
Geoderma 320, 30-42. https://doi.org/10.1016/j.geoderma.2018.01.018
2. Olesen, S.E. 2009. Kortlægning af potentielt dræningsbehov på landbrugsarealer opdelt efter landskabselement, geologi, jordklasse,
geologisk region samt høj/lavbund. Aarhus University.
29. Plauborg, F., Hoffmann, C.C., Audet, J., Skjødt, M. 2021. Designmanual for minivådområder med træis ltermatrice baseret på MMM
projektet. DCA - Nationalt Center for Fødevarer og Jordbrug, DCA rapport; Nr. 190, 47 pp. (In Danish).
https://dcapub.au.dk/djfpdf/DCArapport190.pdf
30. Schipper, L.A., Barkle, G.F., Vojvodic‐Vukovic, M. 2005. Maximum rates of nitrate removal in a denitrication wall. Journal of
environmental quality, 34(4), 1270-1276.
31. Schipper, L.A., W.D. Robertson, A.J. Gold, D.B. Jaynes, and S.C. Cameron. 2010. Denitrifying bioreactors: An approach for reducing
nitrate loads to receiving waters. Ecol. Eng. 36:1532–1543. https://doi.org/10.1016/j.ecoleng.2010.04.008.
32. Schmidt, C.A. and M.W. Clark. 2012. Ecacy of a denitrication wall to treat continuously high nitrate loads. Ecol. Eng. 42: 203-211.
33. van Driel, P.W., W.D. Robertson and L.C. Merkley. 2006. Denitrication of agricultural drainage using wood-based reactors. Trans.
ASABE 49: 565-573.
34. WFD. 2000. Water Framework Directive (WFD) 2000/60/EC: Directive 2000/60/EC. http://ec.europa.eu/environment/water/water-
framework/index_en.html
35. White, S.A., Morris, S.A., Wadnerkar, P.D., Woodrow, R.L., Tucker, J.P., Holloway, C.J., Conrad, S.R., Sanders, C.J., Hessey, S., Santos, I.R.
2022. Anthropogenic nitrate attenuation versus nitrous oxide release from a woodchip bioreactor. Environmental Pollution 300.

Page 16/19
https://doi.org/10.1016/j.envpol.2022.118814
3. Woli, K.P., M.B. David, R.A. Cooke, G.F. McIsaac and C.A. Mitchell. 2010. Nitrogen balance in and export from agricultural elds
associated with controlled drainage systems and denitrifying bioreactors. Ecol. Eng. 36: 1558-1566.
Figures
Figure 1
Map showing the location of the ve bioreactor facilities.

Page 17/19
Figure 2
Top: Total nitrogen (TN) load per day at the inlet of the bioreactors relative to hydraulic load per day. Middle: TN removal (total removal per
day) relative to hydraulic load per day. Bottom: TN removal (percent removal per day) relative to hydraulic load per day. The relationships
between the variables were assessed using linear regressions before and after transformation of the hydraulic load using natural
logarithms. GA, GB, GC are Gyldenholm A, B, C, respectively. HA, HB, HC are Hoffmansgave A, B, C, respectively. Se is Serupgård. Eg is
Egsmarken. Du is Dundelum.

Page 18/19
Figure 3
Total P release (black line) and hydraulic load (blue line) from the bioreactor at Serupgård during the two hydrological years 2018-2019 and
2019-2020.

Page 19/19
Figure 4
Investments in woodchip bioreactors depending on reactor size. Investments in biomaterial over time are included for all projects. Minimum
standard costs exclude investment in pumps, but this is included in the maximum standard costs.
Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.
Supplementaleldbasedwoodchipbioreactors.docx