Water Treatment
During the study period there were minor interruptions due to power outages, forebay pump malfunctions, lack of water in the forebay, and filter clogging; but over 1.5 million liters of water passed through the filter, an average of over 10,000 liters per day. Various pumping interruptions and slowdowns affected the volume of water filtered during each sampling period (Figure 3). The original goal was to use the same batch of biochar continuously throughout the study, but because of heavy particle loads in the input water, the initial batch of biochar began to clog during the third period after approximately eight weeks. Although the nominal size of the sand in the pre-filter was 0.5 mm, fine particles still occluded the biochar and slowed the water flow. Reduced flow did not seem to reduce treatment efficacy, but additional pre-filters were necessary to maintain adequate flow for the bioreactor channel. New biochar and 100-µm glass fiber particle filters were installed after the third sampling event, effectively dividing the study into a nine-week trial and a twelve-week trial. Because of excessive particle loads in the forebay, filters were changed three times per week when the sand filter was backwashed.
Chemistry
All analytical laboratory blanks were non-detect, and recovery of laboratory control material spikes and surrogate spikes were within acceptable limits with the exception of clothianidin. Clothianidin recoveries ranged from 44.0–59.6%, and were considered qualitative, but maintained in the data set to demonstrate removal through filtration. Analytical results from the fifth sampling period were also omitted because errors in sample handling rendered the data unusable.
A range of 11 to 21 compounds were detected in each sampling period (Table 1). Throughout the study fifteen insecticides were detected, as well as eight fungicides and three herbicides. Some compounds were detected in every event, while others were only detected in a single event. Neonicotinoid insecticides were measured at the highest concentrations, followed by the fungicide mefenoxam and the herbicide bensulide.
The first batch of biochar was used for nine weeks, during which time samples were collected three times. During the first three-week sampling period (1A), the filter system reduced twelve of the thirteen compounds detected in the input sample to below detection limits in the output sample, for an average treatment efficacy of 99.9% (Table 1). This efficacy began to wane in the second (1B) and third (1C) treatment periods. Eleven of seventeen compounds were reduced below detection limits by six weeks of treatment, and eight of fifteen compounds by nine weeks of treatment, for treatment efficacies of 93.1% and 91.8%, respectively. At this stage of the study the biochar was getting increasingly clogged with fine particles and the water flow was getting significantly reduced.
The biochar was replaced and 100-µm fine particle filters were added to the system for the remaining four sampling periods (2A-2D). During period 2A, the new biochar reduced concentrations of 17 of 21 compounds to below detection limits, for a treatment efficacy of 99.3%. Analytical results from the period 2B were unusable, but concentrations of nine of eleven compounds were reduced below detection limits during period 2C (92.9% efficacy), and the concentrations of five of eleven compounds during period 2D (81.3% efficacy).
Load is the total mass of pesticides calculated as a product of the measured concentrations and the water volume treated during each period. Concentrations presented in Table 1 are in microgram or nanogram per liter units, but when multiplied by the amount of water treated, grams of chemicals were removed by the filter. Load reductions ranged from 0.31g to 3.85g per sampling period, and totaled over 13g for the entire study.
Toxicity
All toxicity tests met test acceptability criteria for minimum percent survival in the control (≥90% for C. dubia and H. azteca, ≥80% for C. dilutus, Table 2). Toxicity test organisms exposed to reference concentrations of salts and metals also responded as expected, indicating that all test organisms were healthy and had the appropriate sensitivities during the exposures.
Filter input samples from the forebay were all significantly toxic to at least one test organism during every sampling event. Period 1C was the most toxic with no survival of any of the three organisms (Table 2). Chironomus dilutus was the most sensitive organism to the input water, with complete mortality observed in every input sample. Five of seven input samples were significantly toxic to C. dubia, and four of seven were significantly toxic to H. azteca. All five input samples that were toxic to C. dubia received significant treatment with the filter. Two post-filter samples were still significantly toxic, but the filter increased C. dubia survival from an average of 4% in the input samples to 78.4% in the post-filter samples. Only one input sample caused complete mortality to H. azteca (Period 1C), but within the four toxic input samples, the filter increased average H. azteca survival from 33–77.8%. One post-filter sample was still significantly toxic to H. azteca. Although C. dilutus was the most sensitive organism, the filter improved survival during two periods, but these post-filter samples were still considered significantly toxic. Across all of the sampling periods, the filter improved average C. dilutus survival from 0–18.7%. Because there were no C. dilutus survivors in the input samples, it was impossible to calculate improvements to midge growth, but growth in the post-filter sample from Period 2C was not significantly different from the control, indicating the filter had also removed chronic toxicity. The input samples caused significant toxicity in sixteen of the twenty-one separate toxicity tests that were conducted. The filter increased the overall survival from 12.3–58.3% and significant toxicity was completely removed in six post-filter samples.
There was no single chemical class that caused all of the observed toxicity, but some concentrations of the neonicotinoid imidacloprid and the pyrethroid cypermethrin exceeded organism-specific toxicity thresholds and benchmarks (Technical Appendix). The imidacloprid LC50 for C. dilutus was exceeded in forebay samples collected during period 2C and 2D. The filter completely removed imidacloprid in period 2C, and removed over 90% of this insecticide in period 2D. Period 2C saw the best recovery of C. dilutus survival in the post-filter sample (Table 2). Cypermethrin LC50s for both H. azteca and C. dilutus were exceeded during periods 1C and 2A, with the cypermethrin concentration in period 1C exceeding the H. azteca LC50 by over 250 times. The filter also reduced these concentrations to non-detectable levels. Pyrethroids are hydrophobic contaminants that readily associate with surfaces, and are therefore easier to remove with carbon filter media than more soluble compounds, such as imidacloprid.
Although only two insecticides exceeded organism-specific LC50 values, seven insecticides exceeded the U.S. EPA Aquatic Life Benchmarks (Table 2). Benchmark concentrations are generally more conservative than median lethal concentrations, but an exceedance of these values can still indicate possible contributions to toxicity. The neonicotinoids clothianidin, imidacloprid and thiamethoxam, as well as malathion, bifenthrin, permethrin and methomyl; all exceeded their respective chronic benchmarks for invertebrates, and imidacloprid exceeded its acute invertebrate benchmark. Aquatic life benchmarks are unlike LC50 thresholds in that they are more protective, and likely more indicative of ecosystem health impacts. Benchmark concentrations are estimates below which pesticides are not expected to represent a risk for aquatic life. Input samples from the forebay exceeded benchmark values during every sampling period, but the filter was able to reduce concentrations below benchmark values in 75% of these samples.
Table 1
Concentrations of detected chemicals in pre- and post-filtration samples (A and B, respectively), and calculation of percent reduction through treatment. Shaded cells indicate an exceedance of an organism-specific LC50 or U.S. EPA Aquatic Life Benchmark. Blank cells indicate a chemical was not detected, and ND indicates non-detect in post-treatment sample.
|
|
Period 1A
(7/1/20)
|
Period 1B
(7/22/20)
|
Period 1C
(8/12/20)
|
Period 2A
(9/2/20)
|
Period 2C
(10/14/20 – 6 weeks)
|
Period 2D
(11/4/20)
|
Analyte (ng/L)
|
Type
|
A
|
B
|
% Red.
|
A
|
B
|
% Red.
|
A
|
B
|
% Red.
|
A
|
B
|
% Red.
|
A
|
B
|
% Red.
|
A
|
B
|
% Red.
|
Acetamiprid (µg/L)
|
Neonicotinoid
|
|
|
|
|
|
|
|
|
|
0.091
|
ND
|
-100
|
|
|
|
|
|
|
Clothianidin (µg/L)
|
Neonicotinoid
|
0.425
|
ND
|
-100
|
2.37
|
0.062
|
-97
|
1.04
|
0.023
|
-97.8
|
0.550
|
ND
|
-100
|
0.547
|
ND
|
-100
|
0.815
|
ND
|
-100
|
Imidacloprid (µg/L)
|
Neonicotinoid
|
0.481
|
ND
|
-100
|
1.42
|
0.040
|
-97
|
0.942
|
0.024
|
-97.5
|
1.35
|
0.013
|
-99.1
|
3.97
|
ND
|
-100
|
1.73
|
0.148
|
-91.4
|
Thiamethoxam (µg/L)
|
Neonicotinoid
|
1.62
|
0.026
|
-98
|
0.807
|
0.124
|
-85
|
1.56
|
0.201
|
-87.1
|
2.46
|
0.043
|
-98.2
|
1.17
|
0.049
|
-95.8
|
0.128
|
0.054
|
-58.0
|
Dimethoate (µg/L)
|
Organophosphate
|
|
|
|
0.034
|
ND
|
-100
|
|
|
|
|
|
|
|
|
|
|
|
|
Malathion (µg/L)
|
Organophosphate
|
0.030
|
ND
|
-100
|
|
|
|
|
|
|
0.070
|
ND
|
-100
|
|
|
|
|
|
|
Bifenthrin (ng/L)
|
Pyrethroid
|
1.04
|
ND
|
-100
|
2.12
|
ND
|
-100
|
2.03
|
ND
|
-100
|
2.60
|
ND
|
-100
|
3.66
|
ND
|
-100
|
1.88
|
ND
|
-100
|
Cypermethrin (ng/L)
|
Pyrethroid
|
|
|
|
|
|
|
581
|
ND
|
-100
|
6.94
|
ND
|
-100
|
1.04
|
ND
|
-100
|
|
|
|
Etofenprox (ng/L)
|
Pyrethroid
|
|
|
|
|
|
|
|
|
|
100
|
ND
|
-100
|
|
|
|
|
|
|
Permethrin (ng/L)
|
Pyrethroid
|
|
|
|
3.60
|
ND
|
-100
|
5.29
|
ND
|
-100
|
10.2
|
ND
|
-100
|
|
|
|
|
|
|
Chlorantraniliprole (µg/L)
|
Ryanoid
|
0.480
|
ND
|
-100
|
0.377
|
0.066
|
-83
|
1.22
|
0.157
|
-87.1
|
1.59
|
0.032
|
-97.9
|
1.01
|
ND
|
-100
|
0.327
|
0.099
|
-69.8
|
Fipronil Amide (µg/L)
|
Phenylpyrizole
|
|
|
|
0.011
|
ND
|
-100
|
|
|
|
|
|
|
|
|
|
|
|
|
Methomyl (µg/L)
|
Carbamate
|
0.102
|
ND
|
-100
|
0.037
|
ND
|
-100
|
0.165
|
ND
|
-100
|
2.00
|
ND
|
-100
|
|
|
|
|
|
|
Methoxyfenozide (µg/L)
|
Diacylhydrazine
|
0.153
|
ND
|
-100
|
0.128
|
0.050
|
-61
|
0.176
|
0.059
|
-66.5
|
0.167
|
ND
|
-100
|
0.822
|
ND
|
-100
|
0.125
|
0.104
|
-16.8
|
Azoxystrobin (µg/L)
|
Fungicide
|
0.045
|
ND
|
-100
|
0.045
|
ND
|
-100
|
0.489
|
ND
|
-100
|
0.142
|
ND
|
-100
|
0.044
|
ND
|
-100
|
0.044
|
ND
|
-100
|
Boscalid (µg/L)
|
Fungicide
|
1.28
|
ND
|
-100
|
0.699
|
0.025
|
-96
|
0.534
|
0.023
|
-95.8
|
0.499
|
ND
|
-100
|
0.408
|
ND
|
-100
|
0.330
|
ND
|
-100
|
Fenamidone (µg/L)
|
Fungicide
|
0.031
|
ND
|
-100
|
|
|
|
|
|
|
|
|
|
0.150
|
ND
|
-100
|
|
|
|
Fludioxonil (µg/L)
|
Fungicide
|
|
|
|
|
|
|
0.810
|
ND
|
-100
|
0.613
|
ND
|
-100
|
|
|
|
0.081
|
ND
|
-100
|
Mefenoxam (µg/L)
|
Fungicide
|
0.068
|
ND
|
-100
|
2.24
|
0.811
|
-64
|
1.88
|
1.02
|
-45.5
|
2.99
|
0.282
|
-90.6
|
0.960
|
0.706
|
-26.5
|
1.99
|
0.831
|
-58.2
|
Propiconazole (µg/L)
|
Fungicide
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Pyraclostrobin (µg/L)
|
Fungicide
|
|
|
|
0.102
|
ND
|
-100
|
0.040
|
ND
|
-100
|
0.101
|
ND
|
-100
|
|
|
|
|
|
|
Tebuconazole (µg/L)
|
Fungicide
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Thiabendazole (µg/L)
|
Fungicide
|
|
|
|
|
|
|
|
|
|
0.023
|
ND
|
-100
|
|
|
|
|
|
|
Trifloxystrobin (µg/L)
|
Fungicide
|
|
|
|
|
|
|
|
|
|
0.045
|
ND
|
-100
|
|
|
|
|
|
|
Bensulide (µg/L)
|
Herbicide
|
0.063
|
ND
|
-100
|
0.051
|
ND
|
-100
|
6.14
|
0.045
|
-99.3
|
0.082
|
ND
|
-100
|
|
|
|
|
|
|
Diuron (µg/L)
|
Herbicide
|
0.050
|
ND
|
-100
|
0.026
|
ND
|
-100
|
|
|
|
0.034
|
ND
|
-100
|
|
|
|
0.028
|
ND
|
-100
|
Prometryn (µg/L)
|
Herbicide
|
|
|
|
0.043
|
ND
|
-100
|
|
|
|
|
|
|
|
|
|
|
|
|
Mean Reduction (%)
|
|
|
-99.9
|
|
|
-93.1
|
|
|
-91.8
|
|
|
-99.3
|
|
|
-92.9
|
|
|
-81.3
|
Volume Treated (L)
|
|
172,464
|
|
193,207
|
|
274,517
|
|
269,530
|
|
438,442
|
|
153,635
|
Load Reduction (g)
|
|
|
0.75
|
|
|
1.39
|
|
|
3.85
|
|
|
3.38
|
|
|
3.43
|
|
|
0.31
|
Table 2
Mean survival results summary for testing with the daphnid (C. dubia), amphipod (H. azteca) and midge (C. dilutus). Samples were collected from the forebay (A) and post filter (B). Control indicates negative dilution water control used in toxicity tests. Grey shading indicates significant toxicity. SD indicates standard deviation. NA indicates not analyzed.
|
Period 1A (7/1/20)
|
Period 1B (7/22/20)
|
Period 1C (8/12/20)
|
Period 2A (9/2/20)
|
Period 2B (9/23/20)
|
Period 2C (10/14/20)
|
Period 2D (11/4/20)
|
C. dubia
|
Survival (%)
|
|
|
|
|
|
|
Sample
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
A
|
0
|
0
|
20
|
24
|
0
|
0
|
0
|
0
|
0
|
0
|
96
|
9
|
92
|
11
|
B
|
100
|
0
|
96
|
9
|
76
|
9
|
72
|
11
|
48
|
18
|
96
|
9
|
96
|
9
|
Control
|
100
|
0
|
100
|
0
|
92
|
18
|
100
|
0
|
100
|
0
|
100
|
0
|
96
|
10
|
H. azteca
|
Survival (%)
|
|
|
|
|
|
|
|
|
|
|
|
|
Sample
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
A
|
93
|
7
|
42
|
26
|
0
|
0
|
32
|
15
|
96
|
5
|
96
|
5
|
58
|
11
|
B
|
100
|
0
|
98
|
4
|
17
|
12
|
100
|
0
|
98
|
4
|
100
|
0
|
96
|
9
|
Control
|
90
|
10
|
100
|
0
|
96
|
9
|
94
|
5
|
100
|
0
|
98
|
4
|
100
|
0
|
C. dilutus
|
Survival (%)
|
|
|
|
|
|
|
|
|
|
|
|
|
Sample
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
A
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
B
|
58
|
23
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
71
|
20
|
0
|
0
|
Control
|
83
|
10
|
90
|
8
|
96
|
8
|
92
|
7
|
92
|
7
|
100
|
0
|
90
|
8
|
C. dilutus
|
Growth (mg/ind.)
|
|
|
|
|
|
|
|
|
|
|
|
|
Sample
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
A
|
NA
|
NA
|
NA
|
NA
|
NA
|
NA
|
NA
|
NA
|
NA
|
NA
|
NA
|
NA
|
NA
|
NA
|
B
|
0.35
|
0.12
|
NA
|
NA
|
NA
|
NA
|
NA
|
NA
|
0.10
|
NA
|
0.76
|
0.54
|
NA
|
NA
|
Control
|
0.93
|
0.47
|
0.88
|
0.61
|
3.66
|
1.88
|
1.83
|
1.04
|
3.84
|
0.57
|
1.11
|
0.48
|
1.97
|
1.49
|
The bioreactor forebay, which supplied water to the filter, receives agricultural runoff from over 1000 acres. The crops on these fields use a variety of pesticides for the treatment of insects, fungi and weeds. It is the mixture of these compounds that ultimately caused the toxicity. Although CDFA was able to measure 61 compounds, this analyte list was still limited compared to the number of compounds applied in the watershed. The toxicity of some compounds, such as pyrethroids, can be additive because they have the same mode of action, and this additivity can be calculated. For example, a toxic unit approach can be used to calculate the relative contribution of each pyrethroid based on the measured concentration divided by the LC50 value. A single toxic unit would be expected to cause 50% mortality. The sum of pyrethroid toxic units for H. azteca explains three of the four significant reductions of survival. Much of the other observed toxicity cannot be easily explained, but is clearly caused by the mixture of compounds draining to the watershed. Complex mixtures are the main reason that conducting toxicity testing is imperative to provide a quantitative measurement and to determine potential impacts on the receiving system.
Bioreactor Treatment
The main purpose of the wood chip bioreactor is to provide a substrate for bacteria to reduce nutrients through denitrification or other biochemical processes (Krone et al. 2022). Post-filter water was passed through a single channel of wood chips to provide additional treatment for nutrients, and to determine if the wood chips would provide additional pesticide removal. The average concentration of nitrate in the forebay (Table 3, Sample A) was 26.1 mg/L, and the average concentration in the post-filter samples was 27.5 mg/L, indicating no nutrient reduction took place in the filter. Nitrate is extremely soluble and was not expected to bind to the carbon during the short residence period in the filter. The processes that reduce nitrate in the wood chip bioreactor require bacteria, heat and residence time, which were not feasible with the small biochar filter. Average concentrations at the downstream end of the wood chip channel were 5.6 mg/L, indicating a 79% reduction in nitrate.
Table 3
Nitrate concentrations (mg/L) measured in samples from the forebay (A), post filter (B) and post wood chips (C).
Sample
|
Period 1A (7/1/20)
|
Period 1B (7/22/20)
|
Period 1C (8/12/20)
|
Period 2A (9/2/20)
|
Period 2B (9/23/20)
|
Period 2C (10/14/20)
|
Period 2D (11/4/20)
|
A
|
25.7
|
38.4
|
18.0
|
29.0
|
39.0
|
28.0
|
17.3
|
B
|
24.0
|
39.8
|
18.2
|
26.5
|
35.8
|
39.4
|
17.3
|
C
|
1.80
|
7.60
|
9.30
|
9.40
|
4.60
|
4.40
|
1.10
|
Percent Reduction
|
-93.0
|
-80.2
|
-48.3
|
-67.6
|
-88.2
|
-84.3
|
-93.6
|
In most cases, the wood chip channel did not further reduce contaminant concentrations, but added compounds back to the post-filter input water (data not shown). For example, during period 1A, the concentrations of all but one compound were reduced to below detection limits. Only thiamethoxam was not completely removed (Table 1). Of the thirteen compounds detected in the forebay input water, twelve were non-detects in the post-filter sample, but seven were again detected in the post wood chip sample. Clothianidin and azoxystrobin were detected at concentrations higher than those detected in the forebay, and two other compounds, propiconazole and prometryn had not been detected in the forebay, but were detected downstream of the wood chips. There was still a net reduction of chemical concentrations between sample A and sample C, but the return of compounds by the wood chips indicate the bioreactor can be a source of previously-bound contaminants if uncontaminated water is used for the input.
Several samples of water that had passed through the wood chips had high enough concentrations of pesticides added back to them to cause significant toxicity (data not shown). Periods 1B, 1C and 2B had significant treatment of toxicity to C. dubia in the post-filter sample, and had toxicity returned in the post wood chip sample. This was also true of H. azteca in period 1C. Significant treatment of C. dilutus toxicity only occurred during periods 1A and 2C, but complete mortality was observed in all post wood chip samples.
The wood chips in this system had been put in place to treat nitrate in the spring of 2017. Since that time, the bioreactor has processed millions of liters of water, and continues to successfully reduce nitrate concentrations. The wood chips have also served as a substrate and carbon source for the binding of agricultural chemicals, and these chemicals appear to be leaching from the wood chips into the clean water that is passing through the channel. This situation might have been alleviated by placing the biochar filter downstream of the wood chip channel, but the primary objective of this study was to determine the raw treatment efficacy of the filter as a standalone unit. It was assumed the bioreactor would serve its purpose with nitrate reduction, but it also ultimately lessened the overall load reduction.
Longevity
It is difficult to determine how long a batch of carbon will successfully treat agricultural runoff. Whether the filter media is granulated activated carbon or biochar, there are several factors that will affect carbon efficacy, including water volume, flow rate, contaminant load and particle load. The lower threshold of acceptable treatment that will trigger replacing the carbon needs to be determined according to individual treatment goals. In the case of granulated activated carbon, the earliest breakthrough of contaminants into the effluent might indicate replacement (U.S. EPA 1991), but with biochar use in an agricultural setting, it would be costly and difficult to constantly monitor chemical concentrations downstream of the carbon treatment. A more likely solution would be to over-engineer the system for a worst-case scenario, safe in the knowledge that the filter could last for a known period. In the case of the filter used in this study, multiple units could be placed in parallel to increase capacity.
Fouling of the biochar with particles caused this study to be divided into two treatment efficacy study periods. This situation effectively created two experimental replicates. The waning treatment efficacy from both periods was similar. The first period lasted nine weeks and saw the efficacy go from 99.9–91.8%. During the second period of twelve weeks, the efficacy went from 99.3–81.3%. The downward efficacy slope of each study period was approximately -1.6, indicating that barring additional complications from clogging, the biochar used in either treatment period would have lasted approximately 34 weeks under similar irrigation volumes before reaching 50% efficacy, and approximately 58 weeks to reach 10% efficacy.
Creating a filter with a larger biochar bed would increase the efficacy and longevity. A larger system could also allow for a mixture of other materials with the biochar to increase flow and decrease clogging. Recent experiments with mixtures of biochar and pumice show increased flow efficiency (unpublished data), and the use of biochar and wood chip mixtures in recharge basins have demonstrated increased infiltration efficiency over biochar alone (Andrew Fisher, UC Santa Cruz, personal communication). Further study is needed to determine an optimal mixture of flow and treatment, but current studies show that if water can come in contract with the biochar, treatment will occur (Phillips et al. 2021).
Cost Benefit Analysis
Understanding the economic advantages and disadvantages of this type of management practice will be helpful in evaluating management decisions related to this technology. This type of practice can be small and portable, and could potentially be used in a variety of locations either individually, in series or in parallel. Initial cost and maintenance will vary based on need, but improved water quality will reduce economic grower liability for producers under regulatory frameworks addressing nonpoint source runoff. This management practice could potentially be added to the list of practices recommended and subsidized by agencies like the Natural Resources Conservation Service (NRCS) through their Environmental Quality Incentives Program (EQIP), or through grant funded projects by Resource Conservation Districts or local agencies, thus providing economic incentives for installation.
The current practice utilized 600 liters of biochar in a specialized filter housing, but costs could be minimized by using a similar volume of biochar in less expensive, and more easily manageable 55-gallon drums. Ideally, particle-free runoff will be pumped from a sediment trap or pond, but if the water source contains a heavy particle load, optional sand and glass-fiber filters can be used to reduce suspended particles. Open-topped plastic drums would be plumbed with diffusers to distribute water over the biochar mixture. Water would be collected from the bottom of the biochar column and conveyed to a drainage.
Producers of diverse crops can benefit from carbon filtration of their runoff. Improved water quality in tailwater runoff will lead to reductions of pesticide concentrations and occurrences of pesticide-related toxicity. Reductions in pesticide loads will promote healthier invertebrate communities, with cascading benefits to fish, birds, and other wildlife in extended food webs. Migratory birds, wetland habitats, and anadromous fish are all expected to benefit directly and/or indirectly from reduced pesticide inputs to waterways. Use of the filter for treating agricultural runoff may also be attractive to growers previously unable to commit to established conservation practices. A mobile treatment system, either offered as a rented service or as a purchased product, could be used on properties without the ability to install vegetated treatment systems, or as a final step in an integrated system to remove multiple classes of pesticides. Use of a filter system for treating agricultural runoff will not promote bacterial vectors associated with food safety concerns. Potential drawbacks of using the filter system are unknown costs, particularly in the form of maintenance. Variable conditions on a farm might lead to excessive particle loads in the input water, which would lead to increased labor and materials costs for keeping the pre-filters clean and preventing the biochar from clogging. Local users and advisors can determine the required maintenance to adequately suit their treatment goals.