|
Can Woodchip Bioreactors Be Used at a Catchment Scale? Nitrate Performance and Sediment Considerations
Gary W. Feyereisen1,*, Ehsan Ghane2, Todd W. Schumacher1, Brent J. Dalzell1, Mark R. Williams3
Published in Journal of the ASABE 66(2): 367-379 (doi: 10.13031/ja.15496). 2023 American Society of Agricultural and Biological Engineers.
1Soil & Water Management Research Unit, USDA ARS, St. Paul, Minnesota, USA.
2Biosystems and Agricultural Engineering, Michigan State University, East Lansing, Michigan, USA.
3National Soil Erosion Research Laboratory USDA ARS, West Lafayette, Indiana, USA.
*Correspondence: gary.feyereisen@usda.gov
The authors have paid for open access for this article. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License https://creative commons.org/licenses/by-nc-nd/4.0/
Submitted for review on 12 December 2022 as manuscript number NRES 15496; approved for publication as a Research Article and as part of the “Advances in Drainage: Selected Works from the 11th International Drainage Symposium” Collection by Community Editor Dr. Zhiming Qi of the Natural Resources & Environmental Systems Community of ASABE on 11 January 2023.
Mention of company or trade names is for description only and does not imply endorsement by the USDA. The USDA is an equal opportunity provider and employer.
Highlights
Abstract. Denitrifying bioreactors, a structural practice deployed at the field scale to meet water quality goals, have been underutilized and require additional evaluation at the small catchment scale. The objective of this study was to quantify the performance of a large, multi-bed denitrifying bioreactor system sized to treat agricultural drainage runoff (combined drainage discharge and surface runoff) from a 249-ha catchment. Three woodchip bioreactor beds, 7.6 m wide by 41 m long by 1.5 m deep, with cascading inlets, were constructed in 2016 in southern Minnesota, U.S. The beds received runoff for one water year from a catchment area that is 91% tile-drained row crops, primarily maize and soybeans. Initial woodchip quality differed among the three beds, affecting flow and nitrate removal rates. Bioreactor flow was unimpeded by sediment for twelve events from September 2016 to July 2017, during which time 55% of the discharge from the catchment was treated in the bioreactor beds. Average daily nitrate removal rates ranged from 2.5 to 6.5 g-N m-3 d-1 for the three bioreactor beds, with nitrate-N load removal of flow through the beds between 19% and 27%. When accounting for untreated by-pass flow, the overall nitrate-N removal of the multi-bed system was 12.5% (713 kg N). During high-flow events, incoming sediment clogged the reactor beds, decreasing their performance. There was 4,520 kg of sediment trapped in one bed, and evidence suggests the other two trapped a similar load. To solve this problem and prolong the bioreactor’s lifespan, we installed a shutoff gate that activated when inflow turbidity exceeded a threshold value. Finally, the findings indicate that catchment-scale denitrifying bioreactors can successfully remove nitrate load from agricultural runoff, but sediment-prevention measures may be required to extend the bioreactor's lifespan.
Nitrate loading from agriculture contributes to hypoxia in receiving water bodies worldwide (Altieri and Diaz, 2019). The second-largest hypoxic zone globally is in the Gulf of Mexico at the outlet of the USA’s Mississippi River Basin (MRB) (Altieri and Diaz, 2019), where nutrient losses from subsurface tile-drained row crop agriculture contribute heavily to this persistent ecological problem (Rabalais and Turner, 2019). To limit the size of hypoxic zones, nutrient reduction strategies or goals are often developed (United States Environmental Protection Agency, 2010; US EPA, 2011). For example, the US EPA tasked the 12 U.S. states comprising the MRB to develop nutrient reduction strategies that targeted nitrogen (N) and phosphorus (P) load reductions of 45% by 2015 (Mississippi River/Gulf of Mexico Watershed Nutrient Task Force, 2008), which was subsequently extended to 2035 (Mississippi River/Gulf of Mexico WNTF, 2015). However, in the MRB and elsewhere little progress has been made toward meeting nutrient reduction goals within desired timeframes (IDALS, IDNR, and ISU College of Agriculture and Life Sciences, 2020; IEPA, IDOA, and UIE, 2021; MPCA, 2020).
Conservation measures that could be deployed to achieve the desired nutrient reductions from agriculture tend to fall into three broad categories: in-field management, land-use change, and edge-of-field practices. At increasing levels of conservation implementation, edge-of-field practices need to play a proportionately larger role in reduction efforts (Feyereisen et al., 2022a). Meeting nutrient reduction goals will also require multiple or ‘stacked’ conservation practices on most row crop areas (Feyereisen et al., 2022a). Edge-of-field conservation practices, including wetlands (Mitchell et al., 2022), denitrifying bioreactors (Christianson et al., 2021), and saturated buffers (Jaynes and Isenhart, 2014), are typically designed to treat water from individual fields. Although, wetlands can be constructed to treat larger, multiple-field areas (Vymazal and Brezinoava, 2018), and larger denitrification beds have been installed to treat greenhouse effluent in New Zealand (Schipper et al., 2010a) and aquacultural effluent in Denmark (von Ahnen et al., 2018). Large bioreactor beds are currently being installed in Spain to treat N-rich streamflow (Canal Mar Menor, 2021); however, at this time, the work reported herein represents the only catchment-scale bioreactor built and monitored to treat water from multiple fields and farms. Potential advantages of treating water at this scale as opposed to individual fields include more cost-effective treatment (Hartfiel et al., 2021), centralized management, and ease of performance verification.
As the size of the treatment area grows from an individual field to a small catchment, additional landscape processes are likely to contribute to the quantity and quality of water that needs to be treated by the edge-of-field practice. The dominant source of water (e.g., matrix flow vs. preferential flow; Williams and McAfee, 2021), the timing and magnitude of nutrient delivery (Williams et al., 2018), and the influence of other man-made landscape features such as surface inlets, roads, and road ditches may result in varying subsurface drainage discharge and water quality with increasing spatial scale. In particular, surface inlets, which connect ponded overland flow with the subsurface drain system via a vertical pipe, provide a direct path for surface runoff constituents (i.e., sediment) to enter the subsurface drainage system and, as a result, any linked edge-of-field practice. Sediment has been shown to shorten the lifespans of woodchip bioreactors (Christianson, et al., 2020; Duncan, 2022) and reduce the effectiveness of a woodchip bioreactor situated in an agricultural ditch (Maxwell et al., 2022a).
Bioreactor beds typically consist of a trench filled with woodchips through which tile drainage discharge is directed. Nitrate is removed from the water via the process of heterotrophic microbial denitrification (Greenan et al., 2006; Warneke et al., 2011). The conversion of nitrate to the desired endpoint of N2 gas is a time-dependent process, with typical design hydraulic residence times (HRT) of >3 to <48 h (USDA NRCS, 2020). If the HRT is too short, then there can be an increase in the undesirable release of N2O (Davis et al., 2019). In contrast, too long an HRT can result in the removal of all nitrate and the subsequent production of undesirable compounds, e.g., hydrogen sulfide, methane, and methylation of elemental mercury (Kosolapov et al., 2004; Shih et al., 2011).
Maintaining HRT within a desired range is a design decision addressed in the USDA-NRCS bioreactor design standard (USDA NRCS, 2020). One of the two criteria for calculating bioreactor capacity is to size the bed for a minimum 3 h HRT at 15% of the peak discharge of the catchment drainage system. As catchment size and thus flow volume requiring treatment grow, sizing one bioreactor bed to treat the entire catchment will approach some practical limit, especially in terms of length, because head losses increase proportionally with flow path length. Also, sizing one bed to treat larger flow volumes increases the probability of excessively long HRTs during low-flow periods. One approach to overcome these limitations would be to design multiple beds with cascading inlets so that as system flow exceeds bed capacity, overflow cascades to another bed that would operate in parallel. Recently, Maxwell et al. (2022b) reported on a paired bioreactor system with two beds, which they designated “main” and “booster” bioreactors.
The research reported herein is the first report of a multiple-bed bioreactor system with cascading inlets and parallel beds for treating agricultural drainage discharge at the small catchment scale. Our primary research question was: Can a bioreactor be an effective conservation practice for removing nutrients at the small catchment scale? The two main study objectives were to (i) evaluate the performance of the bioreactor at this spatial scale (when it was flowing, not restricted); and (ii) document issues with bioreactor beds clogging due to accumulated sediment and how this problem was resolved to extend the bioreactor bed life.
Materials and Methods
Site Description
The unique three-bed woodchip bioreactor system was installed in 2016 on a 249-ha catchment drained by a county drain pipe in Faribault County near Blue Earth, Minnesota, U.S. The system is located in a grassy waterway between cropped fields and drains into a 300-meter-long riprapped stream that flows into the Blue Earth River (fig. 1). Each bed is 7.6 m wide, 41 m long, and 1.5 m deep, with a 0.5 mm thick polyethylene film lining. The beds are arranged end to end with 11 m between them. The inlets are cascading; that is, when the system inlet flow exceeds the capacity of the first bed (BR1), the system inlet overflow cascades to the inlet of the second bed (BR2); when the system inlet flow exceeds the first and second beds’ capacities, overflow cascades to the inlet of the third bed (BR3). Overflow from BR3 is returned to the county tile main. Flow from the 60 cm diameter county drain pipe to the beds is controlled by a water table control structure. Flow that exceeds the capacity of the bioreactor system bypasses it. Inlet flows to and outlet flows from each bed are also managed by water table control structures.
The county main tile system in the catchment that feeds the bioreactors was improved in 2015 to address the collapse of a 45 cm diameter clay tile main. During construction, four drop intakes with perforated riser pipe inlets (Hickenbottom, Fairfield, IA, U.S.) were installed in road ditches (fig. S1) and buried in 19 mm diameter washed gravel. The drop intakes convey surface runoff from adjacent fields in this prairie pothole landscape and runoff from the limestone gravel roads into the subsurface county drain pipe. One additional surface intake exists in one of the fields.
  |
| Figure 1. Location of catchment (Left) and layout of three bioreactor beds (right). Each bed is 7.6 m wide by 41 m long by 1.5 m deep. The inlets are plumbed in cascading fashion so when flow exceeds BR1, catchment discharge moves to inlet of BR2, and likewise for inlet of BR3. |
Precipitation was measured at the site with a manual rain gauge and a tipping bucket rain gauge connected to a datalogger during non-freezing months. Precipitation for freezing months and historical meteorological data (1972 to 2017) for Blue Earth, MN, 6 km from the site, were obtained from the Minnesota Department of Natural Resources (MN DNR, 2019).
The construction of the bioreactor system was completed in March 2016. Flow through the beds was initiated 4 April 2016 and continued into July 2016, when tile drainage halted due to insufficient precipitation. Flow data from this period are not available, during which time instrumentation was being installed to measure flow through the county tile main, flow out of each bed, and overflows at BR1 and BR3 inlets. System flow was reestablished 7 September 2016 and continued through 30 July 2017. Data collected during this period, which for this region is essentially one water year, are reported herein.
Flow Measurement
Flow through the county drain pipe was measured at 5-minute intervals with an area-velocity sensor (Model 2150, ISCO Teledyne, Lincoln, NE, U.S.) mounted in a 60 cm diameter smooth-wall PVC pipe. Flow from the outlets of each bed and at the overflow from BR1 and BR3 inlets was measured at 5- minute intervals with a pressure transducer (Model No. CS451, Campbell Scientific Instruments, Logan, UT, U.S.) behind a V-notch weir machined into a 25 cm tall 12.5 cm thick rigid PVC stoplog (Agri-Drain Corp., Adair, IA). Orifice flow through the 13 cm weep hole in the bottom stoplog was added to V-notch flow to calculate the total flow through each bed. Equations for converting stage to flow are found in the Supplementary Information. The accuracy of the pressure transducers was checked weekly using an electronic tape (Model No. 101, Solinst, Georgetown, ON, Canada), and adjustments were made in the datalogger programs when the pressure transducer and electronic tape measurements differed by >6 mm. The area-velocity sensor was connected to a cell modem, and four dataloggers were connected by radio to a second cell modem to allow remote monitoring of the site.
Tracer Test
A tracer test was conducted on BR1 on 13 to14 October 2016 by mixing 4 kg of potassium bromide (KBr) into 20 L of water and pouring it into a pipe inserted into the inlet manifold. Two automated samplers (Model 3700, ISCO Teledyne, Lincoln, NE) were programmed to collect (from the BR1 outlet) 500 ml samples at 10 min intervals from hours 2 through 6, at 15 min intervals for 2 additional hours, and at 1 hour intervals for 14 hours. The samples were measured for Br- concentration via flow injection colorimetry (Lachat QuikChem 8500, Hach Co., Loveland, CO) (QuikChem Method 10-135-21-2-B). A similar test on BR2 on 31 October 2016 was unsuccessful. Flow issues with BR3 precluded tracer testing. The mean tracer residence time, bromide mass recovery, volumetric efficiency, effective porosity, and Morrill Dispersion Index were calculated using the methods of Ghane et al. (2019).
The hydraulic residence time was calculated by dividing the product of the saturated volume of the bed and its effective porosity by the flow rate. The saturated volume was estimated based on the estimated average daily water depth at the bed midpoint. Since these beds are long (41 m), there is head loss along the bed length such that the average bed depth (which we assumed was at the midpoint) was different than what would be assumed by using the outlet depth. We determined this difference from synoptic tape-down measurements in October 2016 and assumed this difference was consistent, using it with the data logged outlet depth to calculate an average daily midpoint depth.
The effective porosity was calculated for BR1 based on the tracer test (0.62). The effective porosity for BR2 and BR3 were taken as 0.55 and 0.50, respectively, based on the professional judgment given each bed’s unique media characteristics and literature (Moorman et al., 2010; Ghane et al., 2019).
Water Quality Monitoring
Water was sampled at the county drain pipe, herein referred to as the system inlet, and at the outlets of BR1, BR2, and BR3. From September to 21 November 2016, samples were collected with autosamplers (Model 3700, ISCO Teledyne, Lincoln, NE) using a flow-based sampling regime. The autosamplers were taken offline before freezing-up and were unavailable in 2017. Thus, from 29 November 2016 through the end of flow in July 2017, a weekly manual sampling strategy was used. At each sampling, duplicates were obtained, one acidified (Clesceri et al., 1998), and one left unacidified. Samples were transported in a cooler and frozen at the local Soil and Water Conservation District within two hours.
The samples were transported monthly from the Soil and Water Conservation District to the USDA-ARS lab in St. Paul, MN, then thawed and tested within 48 hours of being thawed as follows: Filtered (0.45 µm), acidified samples were analyzed (Lachat QuikChem® 8500, Hach Co., Loveland, CO) by flow injection colorimetry for nitrate-N (NO3--N + NO23--N) (QuikChem® Method 10-107-04-1-A) and ammonium-N (NH4-N) (QuikChem® Method 10-107-06-2-A). Filtered, unacidified samples were analyzed for dissolved reactive P (DRP) by colorimetry (QuikChem® Method 10-115-01-1-A) and for sulfate-S (SO4--S) by a turbidimetric method (QuikChem® Method 10-116-10-1-A). The filtered, unacidified samples were also analyzed for dissolved carbon (DC) with combustion followed by an infrared CO2 detector (Model vario TOC Select, Elementar Analysensysteme GmbH, Hanau, Germany). Dissolved inorganic C (DIC) was determined by adding phosphoric acid to and bubbling ultra-zero air through the sample and routing the gases through the infrared CO2 detector. Dissolved organic C (DOC) was determined by subtracting DIC from DC. Unfiltered, acidified samples were digested with an alkaline persulfate protocol (Patton and Kryskalla, 2003) and analyzed for total-N (TN) and total-P (TP) with flow injection colorimetry, by the abovementioned methods. When NH4-N and DRP concentrations were below detection limits, 0.005 mg-N L-1 and 0.003 mg-P L-1, respectively, concentrations were set at half the detection limit.
The nitrate-N load was calculated for each flow interval by multiplying flow by the sample concentration and using the midpoint in flow as the dividing point between samples. Flow through the three beds was assumed to be equal to the outflow since inflow into each bed was not directly measured. The nitrate-N removal rate for each bed was calculated on a daily time step by summing the daily nitrate-N load removed and dividing by the average wetted woodchip volume for the day.
Media and Sediment
There were two woodchip sources for the beds: locally chipped trees, primarily cottonwood (Populus deltoides), taken from a local drainage ditch maintenance project, and a supplier of woodchips typically used for playground surfaces (Courtland Hardware & Hardwoods, LLC, Courtland, MN). The supply of locally chipped material was sufficient to fill BR3, which was constructed first, and approximately half of BR2, which was a mixture of both sources. Constructed last, BR1 was filled with woodchips purchased from the supplier. Woodchips were collected from each bioreactor during construction, known herein as “initial” woodchips, stored at 4°C, oven dried (72 h at 60°C), and sieved (25, 19, 12.5, 9.5, 8, 6.3, 4.75, 3.35, and 1.18 mm) in batches of ˜100 g for 15 minutes (Meinzer II, CSC Scientific Company, Inc., Fairfax, VA). The “cumulative percent finer than” values for each sieve batch were averaged for each bed, and the d10, d50 (median particle size), and d60 were determined by interpolation. The uniformity coefficient (UC) was calculated as d60/d10; smaller values of UC represent more uniform particle size.
Due to an apparent internal blockage of flow, an initial investigation of the BR1 woodchip status was undertaken 24 May 2018 by manually removing the soil cap on a 1 m2 area 2 m from the inlet manifold. Woodchip samples were collected in 30 cm depth increments using a manual posthole digger. The preliminary investigation of the BR1 bed revealed the presence of sediment. Thus, a more thorough investigation of the extent of sedimentation in each of the beds was conducted 18 October 2018.
A mini-excavator removed the 90±25 cm soil cap at 3 and 11 m from the inlets of BR1, BR2, and BR3. Additional holes were dug in BR1 above and adjacent to the inlet manifold, and at 20 and 26 m from the inlet manifold (fig. S2). Time and resource limitations precluded additional sampling of BR2 and BR3. Below each opening in the soil cap, woodchip samples were manually collected in 30 cm increments from the top to the bottom of the bed using a posthole digger. On 21 July 2020, the process was repeated in all beds at 31 and 40 m from the inlet. The spent woodchips, known herein as “final” woodchips, were handled and sieved according to the protocol described in the preceding paragraph.
The sediment content of the initial and final woodchips was estimated as follows: Each sample was dried (72 h at 60°C), ground (initial: 2 mm mesh, Standard Model No. 3 Wiley Mill, Arthur H. Thomas Co., Philadelphia, PA, U.S.; final: 1 mm mesh, ED-5 laboratory Mill, Thomas Scientific, Swedesboro, New Jersey, U.S.), subsampled for moisture content (48 h at 105°C), and combusted in a muffle furnace (4 h at 550°C).
Data Analysis
The county main pipe daily discharge record was divided into 17 flow events, with an event defined as the period beginning the day flow increased and ending the day prior to a subsequent flow increase, provided that flow was less than 50% of peak flow for that event. As defined, these flow events included stormflow and baseflow. The percentages of system flow, nitrate-N removal rate (NRR), nutrient load reductions, and balances were calculated for each flow event. An event from 26 October to 19 November 2016 and four events from 10 February to 19 April 2017 were excluded from analysis due to suppressed flow through the beds that required shutdowns and trial-and-error manipulation of control structure stoplogs. Thus, there are 12 “included events” (table 1). Flow was obstructed in one or more beds in several instances. To examine the potential flow treatment, an assumption was made that the daily flow for all three beds could have been as high as the greatest for that day, up to 100% of the system flow.
| Table 1. Event start/end dates and duration of event, and mean and maximum catchment discharge rates, catchment water yield, and bioreactor outflow by event. Shaded rows were excluded from analysis due to suppressed flow through the bioreactor beds. | |||||||||
| Event No. | Start Date | End Date | Duration (d) | Catchment | |||||
| Mean Flow Rate (l s-1) | Maximum Daily Flow Rate (l s-1) | Water Yield (mm) | |||||||
| BR1 | BR2 | BR3 | |||||||
| Equivalent Depth Treated (mm) | |||||||||
| 1 | 9/7/2016 | 9/21/2016 | 15 | 5.2 | 25.1 | 2.52 | 1.6 | 0.02 | 0.08 |
| 2 | 9/22/2016 | 10/4/2016 | 13 | 36.9 | 68.9 | 15.36 | 4.87 | 2.31 | 0.38 |
| 3 | 10/5/2016 | 10/25/2016 | 21 | 30.2 | 75.2 | 20.96 | 6.33 | 3.68 | 0.68 |
| 4 | 10/26/2016 | 11/19/2016 | 25 | 55.6 | 149.3 | 46.28 | 2.36 | 1.31 | 0.61 |
| 5 | 11/20/2016 | 12/22/2016 | 33 | 17.2 | 34.2 | 19.1 | 7.74 | 2.54 | 0.85 |
| 6 | 12/23/2016 | 1/17/2017 | 26 | 9.6 | 40.8 | 8.34 | 4.49 | 1.98 | 0.03 |
| 7 | 1/18/2017 | 1/29/2017 | 12 | 24.4 | 56.0 | 9.31 | 1.63 | 2.38 | 0.02 |
| 8 | 1/30/2017 | 2/9/2017 | 11 | 13.7 | 35.5 | 4.75 | 0.43 | 1.11 | 0.48 |
| 9 | 2/10/2017 | 2/16/2017 | 7 | 18.2 | 29.1 | 3.79 | 0.14 | 0.32 | 0.45 |
| 10 | 2/17/2017 | 2/27/2017 | 11 | 29.2 | 51.2 | 10.12 | 0.14 | 0.13 | 0.10 |
| 11 | 2/28/2017 | 3/17/2017 | 18 | 35.9 | 81.4 | 21.16 | 1.46 | 0.78 | 0.13 |
| 12 | 3/18/2017 | 4/19/2017 | 33 | 27.6 | 58.7 | 30.56 | 8.61 | 1.77 | 0.13 |
| 13 | 4/20/2017 | 4/29/2017 | 10 | 19.5 | 23.0 | 6.08 | 2.5 | 1.53 | 0.19 |
| 14 | 4/30/2017 | 5/19/2017 | 20 | 26.5 | 72.1 | 17.47 | 4.64 | 3.1 | 0.64 |
| 15 | 5/20/2017 | 6/11/2017 | 23 | 25.8 | 61.4 | 19.68 | 4.71 | 4.2 | 1.21 |
| 16 | 6/12/2017 | 6/21/2017 | 10 | 17.1 | 25.4 | 5.32 | 1.78 | 2.23 | 0.13 |
| 17 | 6/22/2017 | 7/14/2017 | 23 | 7.0 | 26.8 | 5.35 | 1.71 | 1.43 | 0.18 |
Results
Tracer Test
The mean bromide residence time for BR1 was 6.28 h, with a bromide mass recovery of 77%. With a theoretical hydraulic residence time of 7.05 h, the volumetric efficiency was 0.89 (fig. S3). The effective porosity was 0.62, which was used to determine the actual HRT (AHRT) for BR1. The Morrill Dispersion Index (MDI), the ratio of the times at which 90% (t90) to 10% (t10) of the recovered tracer passed through the bed (t90:t10), was 2.8.
Experimental Conditions, Operation, and Hydrology
Precipitation for the calendar year 2016, 1025 mm, was the fourth greatest in the past 45 years (average 814 mm) at Blue Earth, Minnesota, and precipitation for the study period (1 September through 31 July), 937 mm, was the sixth greatest (average 701 mm). During this period, catchment drainage discharge (depth equivalent water yield) was 246 mm (or 26.3%) of precipitation. There were liquid precipitation events in each calendar month of the study period, and the precipitation total for the eight months (1 September 2016 through 30 April 2017), was 616 mm, the highest in the previous 45 years. Temperatures throughout the study period were warmer than the 45-year average for nine of the 11 months. The warmer (January, +3.4°C; February, +6.5°C) and wetter than normal conditions supported flow throughout the winter months (fig. 2), which is unusual for this climate.
In the Fall of 2016 (September and October), water control structure stop logs were adjusted at bed outlets and inlet overflows to balance the outflows among the beds. The design flow rate (8.8 l s-1) was achieved in BR1, but not in BR2 nor BR3, apparently due to woodchip sizing constraints (see the Media characterization subsection). The 89 mm storm on 25 October 2016 dramatically increased catchment discharge; afterward, bioreactor outflows diminished for three weeks (fig. 2). During March and April 2017, outflow from the bioreactor beds was again impeded. Stoplog settings were adjusted, and the beds drained and refilled so that bed outflow was re-established for the remainder of the growing season.
After a dry period in Fall 2017 (beginning in August), county main pipe discharge returned in Spring 2018 (March). Outflow rates through the bioreactor beds were again diminished. Adjustments made to the bed inlets/outlets and draining the beds throughout Spring 2018 were unsuccessful in reestablishing bioreactor bed outflow. On 25 June 2018, a 91 mm storm ended meaningful flow through the beds. During this event, surface runoff from the catchment inundated the grass waterway where the beds were located (fig. 1) and entered the beds through inlet and outlet water control structures and manifold cleanout ports.
The catchment water yield during the included events was 134 mm (table S1a) or 336,000 m3 (table S1b). During this time, the total outflow depth through the bioreactor beds was 74 mm (184,000 m3), or 55% of the system flow. On the day of peak flow for the included events, the percentage of the peak flow treated through the three beds ranged from 21% to 69% (average 41%). On an event basis, the portion of drainage discharge treated ranged from 42% (Event 8) to 78% (Events 6 and 16), with an event-based average of 58%. The cumulative treated outflow depth was greatest to least in order of BR1, BR2, and BR3: 42, 26, and 5 mm, respectively. Daily outflow through BR1 exceeded that of BR2 and BR3 during 80% of the study period. If we assume that the bed with the greatest outflow rate represents the upper limit for treatment in each event, and if all three beds operated at that capacity, then this three-bed bioreactor could have treated 106 mm of depth equivalent water yield (79% of drainage discharge; table S1a).
 |
| Figure 2. Daily precipitation, county main pipe discharge and water temperature, and bioreactor bed outflows from September 2016 through July 2017. Shaded areas represent periods excluded from analysis due to suppressed flow through bioreactor beds. |
The median event HRTs for BR1 ranged from 5.7 to 56 h, BR2 ranged from 6.3 to 26.9 h, and BR3 ranged from 15.5 to 95 h (table S2). Although BR1 received greater flow than BR2, median event HRTs were shortest for BR2 for 7 of the 12 events, in part due to lower bed water table depths (data not shown) and therefore lower wetted pore volumes. In BR3, outflow rates were restricted and bed water table depths were reduced; design HRTs of 8 to 16 hours could not be consistently achieved despite the manipulation of the inlet and outlet stoplogs. The event-based average bioreactor bed outflow rates ranged from 97 to 933 m3 d-1 for BR1, 4 to 555 m3 d-1 for BR2, and 3 to 132 m3 d-1 for BR3 (table S1b). Average event bioreactor outlet water temperatures varied from 18°C (September 2016) to 5°C (January 2017) (table S2).
Nitrate-N concentration, Load Reduction, and Removal Rate
Nitrate-N concentrations in the county main pipe ranged from 10.2 to 20.5 mg N L-1 based upon automated flow-weighted mean (FWM) sampling (prior to 22 November 2016) and 10.5 to 18.3 mg-N L-1 during the weekly manual sampling period (fig. 3). The FWM county main concentration was 16.0 mg-N L-1 over the study period. During the automated FWM sampling period, outflow concentrations for BR1 ranged from 2.9 to 16.4 mg-N L-1, BR2 ranged from 3.3 to 16.4 mg-N L-1, and BR3 ranged from 9.9 to 14.7 mg-N L-1. Weekly manual sampling concentration ranges were similar: BR1 ranged from 0.003 to 16.1 mg-N L-1, BR2 ranged from 3.5 to 15.3 mg-N L-1, and BR3 ranged from 4.9 to 15.0 mg-N L-1.
For the entire study period, the nitrate-N load was 9,790 kg (39.3 kg-N ha-1) from the catchment. For the 12 included events, the catchment nitrate-N load was 5,720 kg, of which the bioreactors treated 55% of the flow and removed 713 kg (12.5%) of nitrate-N. Of the water passing through the bioreactor beds (excluding bypass flow), overall nitrate-N load removal was 23% over the 12 events (table S3). Nitrate-N load removal for the individual beds was 19% for BR1, 27% for BR2, and 25% for BR3 with event-based ranges of 13% to 37% for BR1, 13% to 91% for BR2, and 10% to 81% for BR3 (table 2 and table S3). The average daily NRR were 4.9 for BR1, 6.5 for BR2, and 2.5 g-N m-3 d-1 for BR3, with event-based ranges of 0.7 to 8.6 for BR1, 1.6 to 11.1 for BR2, and 0.9 to 4.1 g-N m-3 d-1, for BR3 (table S2). Assuming that the outflow rate through the three beds was increased to that of the greatest bed (table S1a) and the nitrate-N load removal percentage remained the same, the bioreactors would have removed 18% of the catchment nitrate-N load.
Other Nitrogen, Phosphorus Balances
Ammonium-N was exported from BR1 for every event, from BR2 for 10 of 12 events, and from BR3 for 8 of 11 events (table S4). The greatest export for BR1 (1.7 kg NH4-N) and BR2 (1.9 kg NH4-N) occurred during the 20 days of Event 3. Outflow concentrations tended to be low, averaging 0.16 mg-N L-1 for BR1, 0.19 mg-N L-1 for BR2, and 0.08 mg-N L-1 for BR3 (data not shown). The total-N load balance mirrored that of nitrate-N (table 2 and table S5). Over the study period, DRP and TP were removed by the bioreactors. The bioreactors removed DRP for 15 of all 17 events (table S6) and TP for 13 of all 17 events (table S7). Total-P was released by all three bioreactors on Event 1, after a dry period; BR1 and BR2 released DRP and TP during Event 9.
The TP load from the catchment over the entire study period was 95.6 kg (0.38 kg-P ha-1), of which 41% was exported during discharge Event 4 in response to an 89 mm precipitation event on 25 October 2016. This event also caused the first instance of flow depression through the bioreactor beds. During the study period, the sum of TP loads into and out of the beds was 18.2 and 7.8 kg-P, a 57% reduction of TP. There were instances where bioreactor outlet TP concentrations exceeded inlet TP concentrations near event peak flows (fig. 4).
Dissolved organic Carbon Concentration and Load
Dissolved organic carbon concentrations ranged from 2.2 to 20.4 mg-C L-1 in the county main during the study period and tended to increase during or immediately following event flow peaks (fig. 5). Bioreactor outlet DOC concentrations ranged from 2.7 to 35.3 mg-C L-1 for BR1, 3.3 to 14.3 mg-C L-1 for BR2, and 4.4 to 17.9 mg-C L-1 for BR3, and were in nearly all cases greater than inlet DOC concentrations. Prior to the study period when flow measurements were unavailable (April through August 2016), manual samples were collected semi-weekly or weekly. Average DOC concentrations during this pre-measurement period were 16.1 for BR1, 52.0 for BR2, and 22.9 mg-C L-1 for BR3. Similarly calculated simple averages for the study period were 7.9, 7.7, and 8.1 mg-C L-1, respectively. The DOC load into and out of the bioreactors during the study period was 973 and 1231 kg-C, respectively, for a net export of 212 kg-C, or 22% of the inlet load (table S8). Including bypass flow, the bioreactors increased the DOC load from the catchment by 11%.
 |
| Figure 3. Nitrate-N concentrations at county main pipe (bioreactor inlet) and bioreactor outlets. Concentrations prior to 22 November 2016 were from automated FWM sampling; after that date, concentrations were from weekly manual sampling. Shaded areas represent periods excluded from analysis due to suppressed flow through bioreactor beds. |
Sulfate Load
The event mean sulfate-S removal over the study period was 4% for BR1, 6% for BR2, and 8% for BR3 (table 2). Flow problems and correspondingly long HRTs in BR3 resulted in greater reducing conditions and thus sulfate-S conversion. Although the event mean was greater for BR2 than BR1, the total load removal for BR2 was 3% versus 6.4% for BR1 (table S9).
| Table 2. Descriptive statistics for bioreactor outlet (BR1, BR2, and BR3) event-based nutrient load removal for included events. Load removal for individual events is shown in supplementary tables S1 to S9. Negative values indicate nutrient export. | ||||
| Parameter | Load Removal (%) | |||
| Min | Median | Mean | Max | |
| NO3-N | ||||
| BR1 | 13 | 21 | 21 | 37 |
| BR2 | 13 | 29 | 33 | 91 |
| BR3 | 10 | 35 | 38 | 81 |
| TN | ||||
| BR1 | 16 | 18 | 20 | 34 |
| BR2 | 19 | 28 | 32 | 80 |
| BR3 | 16 | 31 | 36 | 75 |
| DRP | ||||
| BR1 | 5 | 63 | 56 | 80 |
| BR2 | -222 | 89 | 55 | 97 |
| BR3 | 47 | 90 | 82 | 97 |
| TP | ||||
| BR1 | -53 | 58 | 46 | 87 |
| BR2 | -405 | 81 | 32 | 91 |
| BR3 | -63 | 72 | 57 | 93 |
| DOC | ||||
| BR1 | -38 | -14 | -11 | 14 |
| BR2 | -442 | -42 | -70 | 5 |
| BR3 | -201 | -53 | -72 | -3 |
| SO4-S | ||||
| BR1 | -12 | 5 | 4 | 17 |
| BR2 | -2 | 1 | 6 | 50 |
| BR3 | -19 | 6 | 8 | 28 |
Media Characterization
The d50 of the initial woodchips was 14.5 for BR1, 12.9 for BR2, and 10.9 mm for BR3 (table 3). The UC was 2.3 for BR1, 3.7 for BR2, and 2.8 for BR3. The BR2 media was less uniform (greater UC) than BR1 and BR3 media. This was expected given that BR2 was a mixture of BR1 and BR3 media, which differed in mean particle size. Correspondingly, the particle size distribution curve for BR2 lay between BR1 and BR3 (fig. S4). The portion of particles finer than 6.3 mm was similar in BR2 (20%) and BR3 (22%), but comparatively less in BR1 (7%), indicating that hydraulic conductivity would likely be greater in BR1. The dxx values of the final woodchips at 11 and 31 m from the inlet were smaller than for the initial woodchips for all three bioreactors. The UC increased for all three bioreactors, indicating that the particle size distribution had widened.
Ash and Sediment Content
Ash contents of the initial woodchips were 0.71% for BR1, 2.0% for BR2, and 5.6% for BR3. Given that the BR1 woodchips were “clean,” and the ash content was consistent with literature values (Misra et al., 1993), we assumed no sediment in the BR1 initial woodchips, but sediment concentrations were 1.3% for BR2 and 5.2% for BR3, indicating the locally chipped media contained some sediment at the outset. The sediment content was closely related to the ash content (table S10), so we will focus on the sediment content henceforth. For BR1, the greatest sediment contents were observed above and adjacent to the inlet manifold, ranging from 10% to 36%. Of the remaining six sample locations along the bed, four of them had the greatest sediment content at 0.3 to 0.6 m depth, second from the top of the bed. Sediment content ranged from 2.1% to 12.6% for BR2, with no clear pattern by depth. For BR3, sediment content ranged from 2.7% to 23.1%, with greater values in the uppermost increment (0 to 0.3 m) at 31 and 40 m, and at the bottom increment at 10 m.
 |
There were 4,530 kg of sediment trapped along the BR1 bed. Sediment was concentrated at the inlet manifold; however, beginning with the segment centered on the 3 m sample location, accumulation was nearly linear (fig. 6). Assuming a sediment density of 1500 kg m-1, the sediment would occupy 3 m3, which is only 2.4% of the pore space assuming a porosity of 0.40 (Ghane et al., 2016) and a water table depth in the bed of 1.0 m. The pattern of sediment accumulation along the lengths of the beds varied among the bioreactors (fig. 7). The sum of sediment trapped in the four common segments sampled was 2780 kg for BR1, 2860 kg for BR2, and 1900 kg for BR3.
Discussion
Nitrate Removal Performance
In this study, we provide the first report on the performance and challenges of a multi-bed bioreactor system treating discharge from a small agricultural catchment. Study results revealed that drainage discharge from a small catchment can be treated with a multi-bed woodchip bioreactor at NRRs comparable to those situated at the edge of individual fields reported in meta-analyses: Addy et al. (2016), median 4.7 g-N m-3 d-1, 5% to 95% confidence interval 2.9 to 7.3 g-N m-3 d-1; and Christianson et al. (2021), median 5.1 g-N m-3 d-1, and 95% of NRRs <15 g-N m-3 d-1. Although the nitrate-N load removal efficiency of the system was low (12.5%) relative to many other studies—having been impacted negatively by hydraulic issues and temperature—the mass removal (713 kg-N) was multiple times greater than reports from smaller field-scale bioreactors (Christianson et al., 2012; David et al., 2016).
The study period was wetter than normal for this area of Minnesota, U.S. The rainfall-driven precipitation events from January through March 2017 were unusual, and water temperatures were cold; NRRs and nitrate-N load removal efficiencies were correspondingly low. These results may be relevant to more temperate regions where flow throughout the winter is common (e.g., northern Europe and the mid-to-southern tier of the U.S. Midwest). For the study locale, the climate is trending toward warmer winters and greater spring precipitation; thus, these results provide insight into operation under future expected conditions.
Sediment Problems and Solutions
The entrance of sediment into the subsurface drainage system feeding the bioreactor, primarily through surface inlets in the catchment, greatly shortened the operating life of the beds. Similarly, Christianson et al. (2020) attributed sedimentation (along with a breakdown in woodchip particle size) as a cause of reduced hydraulic performance and bioreactor failure in Iowa, U.S. after nine years. Event 4 in the current study was initiated by an 89 mm rainfall on fields that had recently been harvested. Runoff-induced soil erosion during this event is hypothesized to have led to increases in suspended sediment in the drainage water through connectivity via both field and road ditch surface inlets. In the aftermath of Event 4, we did not understand that the cause of reduced flow through the bioreactor beds was sediment (fig. 2) and posited other hydraulic-related causes. Post-humous understanding leads us to hypothesize that a slug of sediment entered the beds at the onset of the storm and was gradually transported along the beds over a period of three to four weeks until flow was re-established to near design capacity. It appears that a rain event in February 2017, which was rain on frozen soil, created a similar scenario at the onset of Event 9.
Conditions from August 2017 to March 2018 were very dry. When flow returned in the spring of 2018 (March), bioreactor outflow problems were encountered again until a storm on 25 June essentially plugged the beds beyond self-repair. The June 2018 storm caused substantial damage to the drainage systems in Faribault County. Federal funds secured for system repairs included resources to exhume and replace the woodchips (fig. S5), since the August 2018 investigation showed that sedimentation was evident along the length of the beds. The replacement medium specified was that of the initial BR1 medium, with a relatively large median particle size and uniform distribution (table 3). The woodchips were replaced in August 2020, and the beds were re-instrumented. Drainage flow was minimal in 2021 but returned in 2022 (see third paragraph below).
 |
| Figure 8. County tile main discharge, cumulative bioreactor outflow, turbidity, and sediment-exclusion gate status. Peak turbidity values beyond y-axis scale were 314 (23 Apr 2017) and 570 NTU (29 Apr 2017). Linear actuator, which had been installed in 2020 when beds were rebuilt, failed in closed position for event beginning 23 April 2022 and replaced 26 April 2022. |
Two efforts to reduce future sediment problems were incorporated into the repair of the bioreactor beds. First, the inlet manifold pipe was changed from a 30 cm diameter corrugated, dual-walled pipe with 12 mm holes, to a 30 mm diameter PVC pipe with well-screen slits. We found that the sediment was packed tightly between the ridges of the corrugations in the original design. To further keep sediment away from the entrance into the beds, the inlet manifolds were encased in a 60 cm by 60 cm trench of cleaned 19 cm crushed rock (fig. S6).
In a second effort to address sedimentation problems, we incorporated a sediment sensing and exclusion system. During the two years of waiting for funds to make the repairs, we monitored turbidity in the catchment discharge (Model OBS 3+, Campbell Scientific, Logan, UT, U.S.) and found that turbidity rose abruptly at the onset of events but also dropped relatively quickly, from a few hours to less than a day later. We attached a 12 V DC linear actuator (Model PA-06-8-180, Progressive Automations, Arlington, WA, U.S.) powered by a solar-charged battery to the manual gate that closed flow to the bioreactor beds, forcing all the catchment discharge down the bypass (fig. S6). The datalogger to which the turbidity sensor and linear actuator were attached was programmed to shut off the flow to the bioreactors when turbidity rose to >20 NTUs and to open flow after turbidity subsided to <10 NTUs for more than one hour. Based on our experience, we recommend for catchment-scale bioreactors that a pre-construction investigation of potential sedimentation sources is completed and that sediment-prevention measures are included in the bioreactor design as needed.
The sediment-exclusion system successfully shut off flow to the bioreactor beds for the first few events after its installation, beginning 12 April, 23 April, 29 April, and 11 May 2022, for 39, 64, 45, and 29 hours, respectively (fig. 8). During the second event, the linear actuator, which was installed in 2020, failed in the closed position. The shutdown would have been 8 hours in duration based on the control logic. Using a relationship between turbidity and total suspended solids (TSS) we developed in 2019, for this five-week period we estimated a TSS load at the catchment outlet of 5,818 kg, of which 5,021 kg (86%) bypassed the bioreactors while the shutoff gate was closed, correcting for the failed linear actuator. While the sediment-exclusion gate is closed, the bioreactors are being bypassed, which is a lost treatment opportunity. For this same period, 33% of catchment discharge bypassed the bioreactors. We are uncertain how well the four events mentioned above represent the long-term sediment response of the catchment. Construction occurred on the drainage system during this time, potentially introducing sediment into the system, and turbidity readings were elevated relative to our previous two years of monitoring.
While bypassing the bioreactor beds during periods of high turbidity reduces treatment opportunities, these periods tend to coincide with large flow events during which most of the flow would bypass anyway. For example, peak flow rates in the CD62 catchment reach up to eight times the flow rate that can be treated in the three bioreactor beds, so the lost treatment opportunity, in this case, is one-eighth of the peak flow rate. Additionally, shutting off flow to the beds and subsequently draining them may increase the NRR substantially. Maxwell et al. (2019) showed that draining laboratory columns for 24 hours per week more than doubled the NRR over continuously saturated columns. Thus, despite being operated only 6/7ths of the time, the drained treatment removed significantly more N mass load. In the future, we plan to investigate the opposing effects of bypass flow (decrease) and NRR (increase) of the sediment-exclusion system on nitrate removal, so that catchment managers can better understand the net impact when planning to meet water quality goals.
Media
The impact of size and consistency of media in maintaining flow and nitrate-N removal was observed in flow Events 1, 2, and 3 (fig. 2). Although there was sufficient system flow to maintain robust flow in all three beds, BR3 outflows were typically one-third or less the design flow rate. We attribute this primarily to the relatively small median particle size of the woodchip media and the presence of soil mixed in with the woodchips initially. Throughout the study, BR3 underperformed hydraulically; nevertheless, the system treated a fraction of catchment flow close to the design expectation of 60% of catchment flow.
Total P, DOC
During non-peak portions of the hydrograph, the beds consistently removed TP. Since TP is typically associated with particulates, we speculate that the bioreactor beds were removing TP in part because they were accumulating sediment. With the sediment exclusion system in place, overall TP reductions would presumably be smaller given that most of the sediment will be bypassing the bioreactor beds. After several weeks of dry down, all three bioreactor beds released TP during Event 1, then consistently reduced TP loads. During the 12 included events, only BR3 released TP during one additional event (Event 6). During the excluded events in February, wherein bioreactor outflow was restricted, there were three additional instances of TP release, each during peak flow. During wintertime weekly sampling (when automated sampling was shut down), it is possible that there were uncaptured episodic releases of TP and DRP between sampling events.
Others have reported a lack of evidence for consistent TP removal in bioreactors (Rambags et al., 2016; Schipper et al., 2010a) and evidence of TP export (David et al., 2016). Contrarily, Feyereisen et al. (2023) reported consistent TP removal in pilot-scale field bioreactors operated under a constant flow rate. Dialameh and Ghane (2022) showed that investigation of P dynamics in drainage systems requires a sub-daily sampling strategy to minimize uncertainty in P load estimation. Phosphorus removal and release dynamics in bioreactors need further investigation with high-resolution sampling.
Loss of DOC upon startup of woodchip bioreactors is a known pollution trading issue (Schipper et al., 2010b). We observed higher DOC concentrations in the pre-measurement period than after it, confirming the observations of others (David et al., 2016; Robertson et al., 2005). By the end of our study, DOC concentrations had reduced to relatively stable levels. The additional DOC available in a bioreactor's first months of operation also tends to support higher denitrification rates (David et al., 2016). Our findings showed that NRRs for BR1 appeared to be dropping toward the end of the study, while that was not the case for BR2. One possible explanation is the physical interference that sedimentation may have had with C availability for denitrification. We observed substantial sediment coating of the post-study woodchips (fig. S7). The NRRs for BR2 were greater than for BR1 (10 of 12 events), but the reason is unclear. Because there were fewer pore volumes that passed through BR2, perhaps C availability characteristic of “young” bioreactors was greater than for BR1. Alternatively, the BR2 media included smaller particle sizes that may have more readily released C.
The strengths of this study include the consistent measurement of flow and nutrient removal for three bioreactor beds over one water year of operation at a spatial scale not previously demonstrated. Two items that would have further benefited the study include measurement of bed water table depth and of total suspended solids (TSS). We did not have data logging pressure transducers within the beds (only at the inlet and outlet structures), so our estimate of the wetted gross bed volume for purposes of calculating NRR, especially during low flow periods, has additional uncertainty. We now have three transducers in each bed and recommend that others who are researching this conservation practice include these at the project’s outset. We did collect TSS samples throughout the experiment; however, the samples were stored frozen, and solid precipitates formed, invalidating our results.
Conclusions/Implications
Lessons learned from this study and recommendations for use of denitrifying bioreactors at the catchment scale include:
- Evaluate obvious and potential sources of sediment, and eliminate them if possible
- A simple sediment exclusion system can be installed to preserve the bioreactor bed life
- Inlet and outlet manifolds can be designed to manage sediment within the bed
- Woodchip media recommendation: larger, uniform particle size without grit or fines
- System plumbing design must consider potential short circuiting
- This scale provides opportunities for centralized management and nutrient reduction verification
A multi-bed denitrifying bioreactor with cascading inlets can treat flows from a small catchment while avoiding excessive HRTs. From this work, which was conducted during a historically wet, cool season, it seems reasonable that >60% and upwards of 75% of flows from a small catchment may be treated, with catchment nitrate-N load removals of 14% to 20%. The net effect of a sediment exclusion system on nitrate-N load removal could decrease or increase these numbers and is likely situation specific.
Supplementary Information
The supplementary materials mentioned in this article are available for download from the ASABE Figshare repository at: https://doi.org/10.13031/21964643
Acknowledgments
We gratefully acknowledge Chad Viland (Faribault County SWCD), Scott Matteson (Minnesota Department of Agriculture), Merissa Lore (Faribault County Drainage Authority), Chuck Brandel and Mark Origer (ISG, Inc.), Sarah Porter (USDA-ARS), Dr. Ulrike Tschirner (University of Minnesota), and Michelle Stindtman. We acknowledge and thank the Minnesota Corn Research and Promotion Council for financial support and the Faribault County Drainage Authority, Faribault County Soil and Water Conservation District, and Minnesota Department of Agriculture for in-kind support.
Nomenclature
AHRT = actual hydraulic residence time
CD62 (County Ditch 62) = the county drain pipe through which the study catchment discharges
DOC = dissolved organic carbon
DRP = dissolved reactive phosphorus
dxx = xx percent finer than
FWM = flow-weighted mean
HRT = hydraulic residence time
MDI = Morrill Dispersion Index
NTU = Nephelometric Turbidity Units
NRR = nitrate-nitrogen removal rate
TN = total nitrogen
TP = total phosphorus
TSS = total suspended solids
UC = uniformity coefficient
References
Addy, K., Gold, A. J., Christianson, L. E., David, M. B., Schipper, L. A., & Ratigan, N. A. (2016). Denitrifying bioreactors for nitrate removal: A meta-analysis. J. Environ. Qual., 45(3), 873-881. https://doi.org/10.2134/jeq2015.07.0399
Altieri, A. H., & Diaz, R. J. (2019). Chapter 24 - Dead zones: Oxygen depletion in coastal ecosystems. In C. Sheppard (Ed.), World seas: An environmental evaluation. Volume III: Ecological issues and environmental impacts (2nd ed., pp. 453-473). Academic Press. https://doi.org/10.1016/B978-0-12-805052-1.00021-8
Canal Mar Menor. (2021). Desnitrificación mediante Biorreactores. Retrieved from https://canalmarmenor.carm.es/?actuaciones/medida-prioritarias/medida-3-2/
Christianson, L., Bhandari, A., Helmers, M., Kult, K., Sutphin, T., & Wolf, R. (2012). Performance Evaluation of Four Field-Scale Agricultural Drainage Denitrification Bioreactors in Iowa. Transactions of the ASABE, 55(6), 2163-2174
Christianson, L. E., Cooke, R. A., Hay, C. H., Helmers, M. J., Feyereisen, G. W., Ranaivoson, A. Z.,... von Ahnen, M. (2021). Effectiveness of denitrifying bioreactors on water pollutant reduction from agricultural areas. Trans. ASABE, 64(2), 641-658. https://doi.org/10.13031/trans.14011
Christianson, L. E., Feyereisen, G. W., Hay, C., Tschirner, U. W., Kult, K., Wickramarathne, N. M.,... Soupir, M. L. (2020). Denitrifying bioreactor woodchip recharge: Media properties after nine years. Trans. ASABE, 63(2), 407-416. https://doi.org/10.13031/trans.13709
Christianson, L.E., R.D. Christianson, C. Díaz-García, G.M. Johnson, B.M. Maxwell, R.A.C. Cooke, N. Wickramarathne, and L. Gentry. Accepted 2023. Denitrifying bioreactor in situ woodchip bulk density. Accepted for Journal of ASABE.
Clesceri, L.S., Greenburg, A.E., & D.A. Eaton, D.A. (eds.). (1998). Standard Methods for the Examination of Water and Wastewater. 20th ed. Am. Public Health Assoc., Washington, DC.
David, M. B., Gentry, L. E., Cooke, R. A., & Herbstritt, S. M. (2016). Temperature and substrate control woodchip bioreactor performance in reducing tile nitrate loads in east-central Illinois. J. Environ. Qual., 45(3), 822-829. https://doi.org/10.2134/jeq2015.06.0296
Davis, M. P., Martin, E. A., Moorman, T. B., Isenhart, T. M., & Soupir, M. L. (2019). Nitrous oxide and methane production from denitrifying woodchip bioreactors at three hydraulic residence times. J. Environ. Manag., 242, 290-297. https://doi.org/10.1016/j.jenvman.2019.04.055
Dialameh, B., & Ghane, E. (2022). Effect of water sampling strategies on the uncertainty of phosphorus load estimation in subsurface drainage discharge. J. Environ. Qual., 51(3), 377-388. https://doi.org/10.1002/jeq2.20339
Duncan, S. (2022). Assessment of aged woodchip bioreactor physical and hydraulic properties. MS thesis. Brookings, SD: South Dakota State University, Agricultural and Biosystems Engineering.
Feyereisen, G. W., Hay, C. H., Christianson, R. D., & Helmers, M. J. (2022). Frontier: Eating the metaphorical elephant: Meeting nitrogen reduction goals in Upper Mississippi River Basin States. J. ASABE, 65(3), 621-631. https://doi.org/10.13031/ja.14887
Feyereisen, G. W., Wang, H., Wang, P., Anderson, E. L., Jang, J., Ghane, E., Coulter, J. A., Rosen, C. J., Sadowsky, M. J., & Ishii, S. (2023). Carbon supplementation and bioaugmentation to improve denitrifying woodchip bioreactor performance under cold conditions. Ecol. Eng., accepted 2 February 2023.
Ghane, E., Feyereisen, G. W., & Rosen, C. J. (2019). Efficacy of bromide tracers for evaluating the hydraulics of denitrification beds treating agricultural drainage water. J. Hydrol., 574, 129-137. https://doi.org/10.1016/j.jhydrol.2019.02.031
Ghane, E., Feyereisen, G. W., & Rosen, C. J. (2016). Non-linear hydraulic properties of woodchips necessary to design denitrification beds. Journal of Hydrology, 542, 463-473. https://doi.org/10.1016/j.jhydrol.2016.09.021
Greenan, C. M., Moorman, T. B., Kaspar, T. C., Parkin, T. B., & Jaynes, D. B. (2006). Comparing carbon substrates for denitrification of subsurface drainage water. J. Environ. Qual., 35(3), 824-829. https://doi.org/10.2134/jeq2005.0247
Hartfiel, L. M., Soupir, M. L., & Rosentrater, K. A. (2021). Techno-economic analysis of constant-flow woodchip bioreactors. Trans. ASABE, 64(5), 1545-1554. https://doi.org/10.13031/trans.14300
IDALS, IDNR, and ISU College of Agriculture and Life Sciences. (2020). Iowa Nutrient Reduction Strategy 2018-19 Annual Progress Report (INRC 0017A). Des Moines, IA: Iowa Department of Agriculture and Land Stewardship (IDALS).
IEPA, IDOA, and UIE. (2021). Illinois Nutrient Loss Reduction Strategy Biennial Report 2019-2020. Springfield, Illinois: Illinois Environmental Protection Agency and Illinois Department of Agriculture; University of Illinois Extension. Retrieved from https://www2.illinois.gov/epa/topics/water-quality/watershed-management/excess-nutrients/Documents/NLRS-2021-Biennial-Report-FINAL.pdf
Jaynes, D. B., & Isenhart, T. M. (2014). Reconnecting tile drainage to riparian buffer hydrology for enhanced nitrate removal. J. Environ. Qual., 43(2), 631-638. https://doi.org/10.2134/jeq2013.08.0331
Kosolapov, D. B., Kuschk, P., Vainshtein, M. B., Vatsourina, A. V., Wießner, A., Kästner, M., & Müller, R. A. (2004). Microbial processes of heavy metal removal from carbon-deficient effluents in constructed wetlands. Eng. Life Sci., 4(5), 403-411. https://doi.org/10.1002/elsc.200420048
Maxwell, B. M., Birgand, F., Schipper, L. A., Christianson, L. E., Tian, S., Helmers, M. J.,... Youssef, M. A. (2019). Increased duration of drying-rewetting cycles increases nitrate removal in woodchip bioreactors. Agric. Environ. Lett., 4(1), 190028. https://doi.org/10.2134/ael2019.07.0028
Maxwell, B. M., Cooke, R. A., & Christianson, L. E. (2022b). Paired denitrifying bioreactors with wide orientation for increased drainage flow capacity. J. Environ. Manag., 319, 115768. https://doi.org/10.1016/j.jenvman.2022.115768
Maxwell, B., Christianson, L., Cooke, R. A., Foltz, M., Wickramarathne, N., Chacon, R., & Christianson, R. (2022a). Nitrate removal and woodchip properties across a paired denitrifying bioreactor treating centralized agricultural ditch flows. Water, 14(1), 56. https://doi.org/10.3390/w14010056
Misra, M. K., Ragland, K. W., & Baker, A. J. (1993). Wood ash composition as a function of furnace temperature. Biomass Bioenergy, 4(2), 103-116. https://doi.org/10.1016/0961-9534(93)90032-Y
Mississippi River/Gulf of Mexico Watershed Nutrient Task Force. (2008). Gulf hypoxia action plan 2008 for reducing, mitigating, and controlling hypoxia in the Northern Gulf of Mexico and improving water quality in the Mississippi River Basin. Washington, DC: Mississippi River/Gulf of Mexico Watershed Nutrient Task Force. Retrieved from https://www.epa.gov/ms-htf/gulf-hypoxia-action-plan-2008
Mississippi River/Gulf of Mexico Watershed Nutrient Task Force. (2015). Mississippi River Gulf of Mexico Watershed Nutrient Task Force new goal framework. Washington, DC: Mississippi River/Gulf of Mexico Watershed Nutrient Task Force. Retrieved from https://www.epa.gov/ms-htf/hypoxia-task-force-new-goal-framework
Mitchell, M. E., Newcomer-Johnson, T., Christensen, J., Crumpton, W., Richmond, S., Dyson, B.,... Forshay, K. J. (2022). Potential of water quality wetlands to mitigate habitat losses from agricultural drainage modernization. Sci. Total Environ., 838, 156358. https://doi.org/10.1016/j.scitotenv.2022.156358
MN DNR. (2019). Minnesota location selector. Retrieved from http://climateapps.dnr.state.mn.us/mapClim2007/MNlocApp.asp
Moorman, T. B., Parkin, T. B., Kaspar, T. C., & Jaynes, D. B. (2010). Denitrification activity, wood loss, and N2O emissions over 9 years from a wood chip bioreactor. Ecol. Eng., 36(11), 1567-1574. https://doi.org/10.1016/j.ecoleng.2010.03.012
MPCA. (2020). 5-year progress report on Minnesota’s Nutrient Reduction Strategy. Doc. no. wq-s1-84a. Retrieved from https://www.pca.state.mn.us/water/five-year-progress-report
Patton, C. J., & Kryskalla, J. R. (2003). Methods of analysis by the U.S. Geological Survey National Water Quality Laboratory: Evaluation of alkaline persulfate digestion as an alternative to Kjeldahl digestion for determination of total and dissolved nitrogen and phosphorus in water. Water Resources Investigations Report no. 03-4174. Denver, CO: United States Geological Survey. https://doi.org/10.3133/wri034174
Rabalais, N. N., & Turner, R. E. (2019). Gulf of Mexico hypoxia: Past, present, and future. Limnol. Oceanogr. Bull., 28(4), 117-124. https://doi.org/10.1002/lob.10351
Rambags, F., Tanner, C. C., Stott, R., & Schipper, L. A. (2016). Fecal bacteria, bacteriophage, and nutrient reductions in a full-scale denitrifying woodchip bioreactor. J. Environ. Qual., 45(3), 847-854. https://doi.org/10.2134/jeq2015.06.0326
Robertson, W. D., Ford, G. I., & Lombardo, P. S. (2005). Wood-based filter for nitrate removal in septic systems. Trans. ASAE, 48(1), 121-128. https://doi.org/10.13031/2013.17954
Schipper, L. A., Cameron, S. C., & Warneke, S. (2010a). Nitrate removal from three different effluents using large-scale denitrification beds. Ecol. Eng., 36(11), 1552-1557. https://doi.org/10.1016/j.ecoleng.2010.02.007
Schipper, L. A., Robertson, W. D., Gold, A. J., Jaynes, D. B., & Cameron, S. C. (2010b). Denitrifying bioreactors — An approach for reducing nitrate loads to receiving waters. Ecol. Eng., 36(11), 1532-1543. https://doi.org/10.1016/j.ecoleng.2010.04.008
Shih, R., Robertson, W. D., Schiff, S. L., & Rudolph, D. L. (2011). Nitrate controls methyl mercury production in a streambed bioreactor. J. Environ. Qual., 40(5), 1586-1592. https://doi.org/10.2134/jeq2011.0072
United States Environmental Protection Agency. (2010). Chesapeake bay total maximum daily load for nitrogen, phosphorus and sediment. US EPA. Retrieved from https://www.epa.gov/chesapeake-bay-tmdl/chesapeake-bay-tmdl-document
United States Environmental Protection Agency. (2011). Lake Erie binational nutrient management strategy: Protecting Lake Erie by managing phosphorus. Prepared by the Lake Erie LaMP Work Group Nutrient Management Task Group. Retrieved from https://www.epa.gov/sites/default/files/2015-09/documents/binational_nutrient_management.pdf
USDA NRCS. (2020). Conservation Practice Standard Denitrifying Bioreactor Code 605 (605-CPS). Washington, DC: NRCS.
von Ahnen, M., Pedersen, P. B., & Dalsgaard, J. (2018). Performance of full-scale woodchip bioreactors treating effluents from commercial RAS. Aquacult. Eng., 83, 130-137. https://doi.org/10.1016/j.aquaeng.2018.10.004
Vymazal, J., & Brezinová, T. D. (2018). Treatment of a small stream impacted by agricultural drainage in a semi-constructed wetland. Sci. Total Environ., 643, 52-62. https://doi.org/10.1016/j.scitotenv.2018.06.148
Warneke, S., Schipper, L. A., Matiasek, M. G., Scow, K. M., Cameron, S., Bruesewitz, D. A., & McDonald, I. R. (2011). Nitrate removal, communities of denitrifiers and adverse effects in different carbon substrates for use in denitrification beds. Water Res., 45(17), 5463-5475. https://doi.org/10.1016/j.watres.2011.08.007
Williams, M. R., & McAfee, S. J. (2021). Water storage, mixing, and fluxes in tile-drained agricultural fields inferred from stable water isotopes. J. Hydrol., 599, 126347. https://doi.org/10.1016/j.jhydrol.2021.126347
Williams, M. R., Livingston, S. J., Penn, C. J., Smith, D. R., King, K. W., & Huang, C.-h. (2018). Controls of event-based nutrient transport within nested headwater agricultural watersheds of the western Lake Erie basin. J. Hydrol., 559, 749-761. https://doi.org/10.1016/j.jhydrol.2018.02.079