Treatment of highway runoff: engineered filter media for pollutant removal through enhanced sorption

GEORGIA DOT RESEARCH PROJECT 11-23
FINAL REPORT
Treatment of Highway Runoff: Engineered Filter Media for Pollutant Removal through Enhanced Sorption
OFFICE OF RESEARCH RESEARCH & DEVELOPMENT BRANCH

1.Report No.: FHWA-GA-16-11-23

2. Government Accession No.:

3. Recipient's Catalog No.:

4. Title and Subtitle: Treatment of Highway Runoff: Engineered Filter Media for Pollutant Removal through Enhanced Sorption
7. Author(s): Susan E. Burns, Nicole T. Caruso
9. Performing Organization Name and Address: School of Civil & Environmental Engineering Georgia Institute of Technology 790 Atlantic Dr. Atlanta, GA 30332-0355
12. Sponsoring Agency Name and Address: Georgia Department of Transportation Office of Research 15 Kennedy Drive Forest Park, GA 30297-2534
15. Supplementary Notes:

5. Report Date: July 2015
6. Performing Organization Code:
8. Performing Organ. Report No.: RP 11-23; T.O. 02-102
10. Work Unit No.:
11. Contract or Grant No.: RP 11-23; PI# 0010765
13. Type of Report and Period Covered:
Final; 14. Sponsoring Agency Code:

16. Abstract: The work performed in this study focused on the investigation of the use of engineered biofiltration layers to enhance the removal of roadway stormwater runoff contaminants (specifically nutrients, solids, heavy metals, and pH). Six Georgia native grasses as well as one turf grass were tested in the column study, along with a permanently saturated zone for biofiltration enhancement. Results indicated that Big Bluestem, Indiangrass, and Switchgrass, when paired with a permanently saturated zone, removed the highest percentage of total nitrogen across all experiments (4%, 13%, and 18% respectively). These species contained thick and dense root systems that spanned the entire length of the biofilter column. Removal of nitrate was enhanced with a saturated zone, while ammonium removal decreased. A permanently saturated zone increased removal of phosphorus, copper, and zinc (removal of lead was >97% in all cases. The results demonstrate that the addition of active biofiltration layers to BMPs on GDOT rightof-ways can be an important component in the reduction of contaminant loading in stormwater that is being discharged to environmentally sensitive environments. 17. Key Words: Biofiltration, native grasses, 18. Distribution Statement: sorption

19. Security Classification (of this report):
Unclassified

20. Security Classification (of this page):
Unclassified

Form DOT 1700.7 (8-69)

21. Number of Pages:
89

22. Price:

GDOT Research Project No. 11-23
Final Report Treatment of Highway Runoff: Engineered Filter Media for Pollutant Removal through
Enhanced Sorption By
Susan E. Burns, Ph.D., P.E. Professor
Nicole T. Caruso Graduate Research Assistant Georgia Institute of Technology
Contract with Georgia Department of Transportation
In cooperation with U.S. Department of Transportation
Federal Highway Administration July 27, 2015
The contents of this report reflect the views of the author(s) who is (are) responsible for the facts and the accuracy of the data presented herein. The contents do not necessarily reflect the official
views or policies of the Georgia Department of Transportation or the Federal Highway Administration. This report does not constitute a standard, specification, or regulation.
ii

TABLE OF CONTENTS EXECUTIVE SUMMARY ............................................................................................. xiii Acknowledgements.............................................................................................................. xv INTRODUCTION ..........................................................................................................1 Biofilters ................................................................................................................................2 Enhancements .......................................................................................................................4 LITERATURE REVIEW ...................................................................................................5 Role of Vegetation .................................................................................................................5 Nitrogen ................................................................................................................................7 Saturated Zone ......................................................................................................................9
Carbon Addition.................................................................................................................... 11 Phosphorus...............................................................................................................12 Heavy Metals............................................................................................................14 Suspended Solids ......................................................................................................14 Objectives............................................................................................................................15 Experimental Investigation .......................................................................................15 Materials .............................................................................................................................15
Grass Species ........................................................................................................................ 20 Methods ..............................................................................................................................21
Synthetic Stormwater ........................................................................................................... 21 Sampling Schedule................................................................................................................ 23 Sample Analysis .................................................................................................................... 24
iii

RESULTS AND ANALYSIS............................................................................................25 Nitrogen ..............................................................................................................................26 Phosphorus.......................................................................................................................... 36 Heavy Metals.......................................................................................................................42
Copper .................................................................................................................................. 42 Lead....................................................................................................................................... 47 Zinc........................................................................................................................................ 52 Turbidity..............................................................................................................................57 pH .......................................................................................................................................59 Plant Growth .......................................................................................................................62 Summary .............................................................................................................................64 Big Bluestem ......................................................................................................................... 66 River Oats.............................................................................................................................. 66 Cherokee Sedge .................................................................................................................... 66 Pink Muhly ............................................................................................................................ 67 Switchgrass ........................................................................................................................... 67 Indiangrass ............................................................................................................................ 67 CONCLUSIONS ..........................................................................................................68 REFERENCES .............................................................................................................69
iv

List of Tables Table 1: Pollutant Removal Specifications (AMEC Earth and Environmental et al. 2001)4 Table 2: Number of Replicates per Column Configuration.............................................. 18 Table 3: Synthetic Stormwater Formulas ......................................................................... 19 Table 4: Sampling Schedule ............................................................................................. 23 Table 5. Root depth and height measurements for column and field plants..................... 63
v

List of Figures Figure 1: Typical biofiltration design (Figure from AMEC Earth and Environmental et al.
2001). ........................................................................................................................... 3
Figure 2: Transformations of nitrogen in oxic and anoxic environments........................... 8
Figure 4: View of the greenhouse on sampling day with all columns, blue sample buckets, batch mixing can, and shade tarp. ............................................................... 20
Figure 5: Nitrogen species concentration means by plant species with influent concentration (dashed lines represent traditional column effluent dosed with average synthetic stormwater). ............................................................................................... 26
Figure 6: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent saturated columns dosed with average synthetic stormwater). ............................................................................................... 27
Figure 7: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent traditional columns dosed with metals spiked stormwater). ............................................................................................................... 28
Figure 8: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent saturated columns dosed with metals spiked stormwater). ............................................................................................................... 29
Figure 9: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent traditional columns dosed with nutrient spiked stormwater). ............................................................................................................... 30
vi

Figure 10: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent saturated columns dosed with nutrient spiked stormwater). ............................................................................................................... 30
Figure 11: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent traditional columns dosed with an average synthetic stormwater after two weeks of drought conditions)................................... 31
Figure 12: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent saturated columns dosed with an average synthetic stormwater after two weeks of drought conditions)................................... 32
Figure 13: Difference of total nitrogen removal from the traditional configurations dosed with average stormwater............................................................................................ 33
Figure 14: Difference of nitrate removal from the traditional configurations with average stormwater. ................................................................................................................ 34
Figure 15: Difference of ammonia removal from the traditional configurations with average stormwater.................................................................................................... 35
Figure 16: Total phosphorus removal by plant species for columns dosed with average synthetic stormwater.................................................................................................. 37
Figure 17: Total Phosphorus removal by plant species for columns dosed with metal spiked synthetic stormwater. ..................................................................................... 38
Figure 18: Total phosphorus removal by plant species for columns dosed with nutrient spiked synthetic stormwater. ..................................................................................... 39
vii

Figure 19: Total phosphorus removal by plant species for columns dosed with average synthetic stormwater after two weeks of drought conditions. ................................... 40
Figure 20: Total phosphorus removal differences from traditional configuration with average synthetic stormwater. ................................................................................... 41
Figure 21: Copper removal by plant species for columns dosed with average synthetic stormwater. ................................................................................................................ 42
Figure 22: Copper removal by plant species for columns dosed with metal spiked synthetic stormwater.................................................................................................. 43
Figure 23: Copper removal by plant species for columns dosed with nutrient spiked synthetic stormwater.................................................................................................. 44
Figure 24: Copper removal by plant species for columns dosed with average synthetic stormwater after two weeks of drought conditions. .................................................. 45
Figure 25: Copper removal as compared to the traditional configuration dosed with average synthetic stormwater. ................................................................................... 46
Figure 26: Lead removal by plant species for columns dosed with average synthetic stormwater. ................................................................................................................ 47
Figure 27: Lead removal by plant species for columns dosed with metal spiked synthetic stormwater. ................................................................................................................ 48
Figure 28: Lead removal by plant species for columns dosed with nutrient spiked synthetic stormwater conditions. ............................................................................... 49
viii

Figure 29: Lead removal by plant species for columns dosed with average synthetic stormwater after two weeks of drought conditions. .................................................. 50
Figure 30: Lead removal as compared to the traditional configuration with average synthetic stormwater.................................................................................................. 51
Figure 31: Zinc removal by plant species for columns dosed with average synthetic stormwater. ................................................................................................................ 52
Figure 32: Zinc removal by plant species for columns dosed with metal spiked synthetic stormwater. ................................................................................................................ 53
Figure 33: Zinc removal by plant species for columns dosed with nutrient spiked synthetic stormwater.................................................................................................. 54
Figure 34: Zinc removal by plant species for columns dosed with average synthetic stormwater after two weeks of drought conditions. .................................................. 55
Figure 35: Zinc removal as compared to the traditional configuration with average synthetic stormwater.................................................................................................. 56
Figure 36: Turbidity by plant species for columns dosed with average synthetic stormwater. ................................................................................................................ 57
Figure 37: Turbidity by plant species for columns dosed with metals spiked synthetic stormwater. ................................................................................................................ 58
Figure 38: Turbidity by plant species for columns dosed with nutrient spiked synthetic stormwater. ................................................................................................................ 58
ix

Figure 39: Turbidity by plant species for columns dosed with average synthetic stormwater after two weeks of drought conditions. .................................................. 59
Figure 40: pH of treated average synthetic stormwater. ................................................... 60 Figure 41: pH of treated metals spiked synthetic stormwater........................................... 61 Figure 42: pH of treated nutrient spiked synthetic stormwater......................................... 61 Figure 43: pH of treated average synthetic stormwater after two weeks of drought. ....... 62 Figure 44: Average nutrient removal across all experiments in the traditional
configurations. ........................................................................................................... 65 Figure 45: Average nutrient removal across all experiments in saturated configurations.65
x

LIST OF SYMBOLS AND ABBREVIATIONS

BMP FHWA GDOT hp L/min LID LOI m2/g mg/L

Best Management Practice Federal Highway Administration Georgia Department of Transportation
horsepower liters per minute Low Impact Design Loss on Ignition square meters per gram milligrams per liter

mg N/L mg P/L NO3NO2NOx NH4+

milligrams of nitrogen per liter milligrams of phosphorus per liter
Nitrate Nitrite Combined Nitrate and Nitrite Ammonium

xi

LIST OF SYMBOLS AND ABBREVIATIONS (Continued)

N2 PVC rpm TKN TN TSS

Nitrogen Gas Polyvinyl chloride rotations per minute Total Kjeldahl Nitrogen
Total Nitrogen Total Suspended Solids

xii

EXECUTIVE SUMMARY
Recent changes to environmental regulations have mandated treatment of stormwater runoff and have required the Georgia Department of Transportation to increase the frequency of installation of treatment structures. These structures, which are known as best management practices (BMPs) also bring ongoing, long term operation and maintenance obligations for GDOT. Currently, the most common type of waterquality treatment structure used by the Department relies predominantly on filtration through a silica sand media. Designing these types of large structures generally requires additional right of way, increased construction area, and impacts to utilities, all of which can cause delays in project schedules and increases in project budgets. If these filtering structures can be constructed with a smaller footprint while still achieving the same water quality outflow as larger structures, a significant cost savings will be realized.
The work performed in this study focused on the investigation of the use of engineered biofiltration layers to enhance the removal of roadway stormwater runoff contaminants (specifically nutrients, solids, heavy metals, and pH). The work performed in this project focused on the design and construction of thirty-two bioreactor columns that were engineered with filtration and sorptive layers that were used to support the growth of grasses (including species native and non-native to Georgia). The columns were designed to facilitate testing under saturated and drought conditions, after the grass species were established. Six Georgia native grasses as well as one turf grass were tested in the column study, along with a permanently saturated zone for biofiltration enhancement. Synthetic stormwater was used in this study. Two months of dosages with
xiii

an average synthetic stormwater were monitored followed by one event with a heavy metal-spiked synthetic stormwater, one event with a nutrient spiked synthetic stormwater, and one event with an average synthetic stormwater after two weeks of drought conditions. Biomass fly ash was also added to columns to determine potential benefits to biofiltration applications.
Results indicated that Big Bluestem, Indiangrass, and Switchgrass, when paired with a permanently saturated zone, removed the highest percentage of total nitrogen across all experiments (4%, 13%, and 18% respectively). These species contained thick and dense root systems that spanned the entire length of the biofilter column. Removal of nitrate was enhanced with a saturated zone, while ammonium removal decreased. Nitrogen leaching was observed from the columns, but that would likely be reduced by utilizing soil of low organic content. Phosphorus, copper, lead, and zinc removal was not correlated with plant species; however, a permanently saturated zone increased removal of phosphorus, copper, and zinc (removal of lead was >97% in all cases, making differences in removal insignificant).
These results support the impact of specific vegetation types on the removal extent of total nitrogen. Saturation provided benefits of total nitrogen, phosphorus, copper, and zinc removal in terms of removal extents as well as consistency of treatment across all experiments. The results demonstrate that the addition of active biofiltration layers to BMPs on GDOT right-of-ways can be an important component in the reduction of contaminant loading in stormwater that is being discharged to environmentally sensitive environments.
xiv

Acknowledgements The authors are pleased to acknowledge the contributions of Mr. Jeremy Mitchell
and Ms. Kasey Henneman; their support to the ongoing operation of the greenhouse was invaluable to the success of the project. In addition, thanks to Mr. Jon D. Griffith, P.G., P.E., for his continued dedication and significant contributions in the formulation of this project and throughout the course of this work.
xv

INTRODUCTION
When rain falls on a paved surface such as a highway, the resulting stormwater runoff can accumulate contaminants such as oil and grease, nutrients, metals, and suspended solids. Because the source of the contamination is distributed, it makes highway stormwater runoff a non-point source of pollution for lakes, rivers, and streams (US EPA 2003). Reduction in these contaminants, specifically nitrogen and phosphorus, is often necessary to prevent overgrowth of algae and eutrophication of the receiving water bodies. Mitigation of the influx of water volumes during a storm event is also required to protect from erosion. Implementation of contaminant reduction and flow volume control can be achieved through the use of best management practices (BMPs). BMPs include permanent infrastructures such as ponds, wetlands, biofilters, sand filters, infiltration trenches, grass channels, and pervious pavements. Also of significant importance for the Georgia Department of Transportation is a reduction in the long-term operation and maintenance costs associated with stormwater BMPs.
Reduction in contaminant concentrations can occur due to a variety of removal mechanisms, including sorption (both adsorption and absorption) to solid surfaces, chemical degradation, and natural attenuation. Sorption represents the accumulation of contaminants at the interface of two phases, and can be either reversible or irreversible. Previous studies of sorption on GDOT right-of-ways have quantified the uptake capacity of zeolites, which are naturally occurring, low cost, highly sorptive aluminosilicate minerals with a high affinity for heavy metals. Gray et al., (2012) demonstrated that
1

zeolites provided a high sorption affinity for lead, copper, and zinc, and could be included as an additional sorptive media at a GDOT BMP (sand filter) located in Canton, Georgia. Inclusion of approximately 2000 lb of zeolite could effectively treat one year's worth of storm water runoff at an additional cost of approximately $150. Natural attenuation is another common process for reduction in contaminant mass or concentration, which can result from a variety of physical, chemical, or biological processes, and is typically used for reduction in the concentration of organic contaminants. While natural attenuation is a process that occurs without any additional engineering of the system, the process of natural attenuation must be monitored closely to guarantee that contaminant transport is contained within expected limits.
Biofilters
Biofilters (also known as bioretention basins, areas, or rain gardens) are highly flexible in application because they can be applied in residential areas, roadway medians, and other urban environments of varying size. Another major benefit to biofiltration usage is the reduced maintenance burden. Vegetation used in a biofilter consists of native grasses, shrubs, and trees that require maintenance 1-2 times per year, as opposed to many turf grasses, which require mowing every 2-3 months after initial establishment. The incorporation of numerous plant species also creates a more aesthetically pleasing environment.
Biofiltration design typically consists of a gravel underdrain system, soil media layers, mulch, vegetation, and an optional sand layer (Figure 1). The design may also include an impermeable liner between the BMP and the native soil in order to retain
2

water in the system or be designed to be in contact with native soil to promote infiltration. Pollutant removal in biofiltration devices is facilitated by a number of physical, chemical, and biological processes. For example, vegetation enhances the biological activity in the soil, thus increasing pollutant removal when compared to that of a typical sand filter (Table 1).
Figure 1: Typical biofiltration design (Figure from AMEC Earth and Environmental et al. 2001).
.
3

Table 1: Pollutant Removal Specifications (AMEC Earth and Environmental et al. 2001)

Contaminant

Biofilter (% Removal)

Sand Filter (% Removal)

Suspended Solids

80

80

Total Phosphorus

60

50

Total Nitrogen

50

25

Heavy Metals

80

50

Typical recommendations for biofiltration construction indicate that a variety of warm season and cool season species should be planted to encourage year-round growth and consistent performance (AMEC Earth and Environmental et al. 2001; Department of Water and Swan River Trust 2007). Species should also be tolerant of flood and drought conditions to prevent frequent replanting. Wetland species may also be considered based on the site characteristics (WEF et al. 2012).
Enhancements
According to the Prince George's County Bioretention Manual (2007), removal processes for pollutants include interception, infiltration, settling, evaporation and transpiration (evapotranspiration}, filtration, absorption, assimilation, and adsorption. Within adsorption, processes include nitrification, denitrification, volatilization, thermal attenuation, degradation, and decomposition. With this large number of processes and the interactions among them, there are many ways to enhance the treatment capabilities

4

of a biofiltration system. Possible enhancements include the selection of nutrient efficient vegetation, implementation of a saturated layer underground by raising the outlet of the underdrains, and engineering the soil media for maximum pollutant uptake.
The objectives of this research program were (1) to provide a literature review on existing stormwater treatment controls operating with enhanced sorption, (2) to conduct laboratory and field tests to evaluate the effectiveness of controls at pollutant removal with comparison between enhanced and standard filter performance and overall costs, and (3) to provide a guidance for selecting and maintaining stormwater filtration enhancements for GDOT applications, especially in proximity to environmentally sensitive receiving waters.
LITERATURE REVIEW
Role of Vegetation
Numerous studies confirm that vegetated filters achieve higher removals of nutrients when compared to non-vegetated filters (Bratieres et al. 2008; Davis et al. 2001; Glaister et al. 2014; Henderson et al. 2007; Lucas and Greenway 2008; Read et al. 2008). Nutrients, particularly nitrogen, may leach from non-vegetated, soil-based filters because vegetation is not available to utilize the nutrients released from the breakdown of organic matter in the soil (Hatt et al. 2007). Vegetation also helps to maintain the hydraulic conductivity of biofilters over time (Hatt et al. 2009), and a thicker root morphology may decrease the impact of clogging (Le Coustumer et al. 2012).
5

An extensive study of 20 different Australian native grasses adapted to low nutrient concentrations in native soils determined that grasses vary greatly in their ability to uptake nitrogen and phosphorus (Read et al. 2008). From this study, Carex appressa (Tall Sedge) seemed to be the most effective plant in biofilters, possibly due to the extensive network of fine root hairs which increase the surface area for nutrient uptake. In a follow-up study, strong correlations were found between nitrogen and phosphorus removal and the length of longest root, root soil depth, root mass, percent root mass, and total root length (Read et al. 2010). A study in Austin, Texas, confirmed this result comparing a common native grass, Muhlenbergia lindheimeri (Big Muhly) with a turf grass, Buchloe dactyloides (Buffalograss 609). The study demonstrated that biofilter columns planted with Big Muhly consistently performed better than those without vegetation or those planted with Buffalograss (Barrett et al. 2013).
Vegetation also contributes to the genetic characteristics that are found within the microorganisms in the treatment systems. Nitrification and denitrification are significantly dependent on the genes that are found within the microbial community, and the presence of vegetation has been linked with an order of magnitude increase in 16S r DNA gene concentrations in soil cores, which indicated a greater potential for nitrogen transformations and removal (Chen et al. 2013). This study also indicated that the presence of these genes decreased with depth but to a lesser extent when heavy vegetation was present. Genes for nitrification were much greater than denitrification genes in all sampling locations indicating more favorable conditions for the creation of nitrate and nitrite.
6

Another recent study conducted in Australia indicated that vegetation within biofilters, when paired with a saturated zone, leads to consistent effluent concentrations of all constituents year round through wet and dry periods (Glaister et al. 2014). Nitrogen
Forms of nitrogen that are readily available for plant uptake include the inorganic forms of nitrate (NO3-) and ammonium (NH4+). Organic forms of nitrogen undergo microbial decomposition by ammonification and nitrification to these inorganic forms. In aerobic environments, ammonium is readily converted to nitrate via nitrification (Figure 2). In contrast, denitrification occurs when bacteria utilize nitrate as an electron acceptor to convert nitrate to nitrogen gas (N2) thus removing it from the system. Since oxygen is a more efficient electron acceptor than nitrate, denitrification will occur at significant rates in an anoxic environment, decreasing total nitrogen (TN) concentrations. Nitrate is especially difficult to remove due to its high solubility. Soil has a net negative charge, so the negative charge of nitrate makes sorption unfavorable as opposed to the net positive charge of ammonium which is attracted to clay particles. This requires nitrate to be biologically transformed for removal.
7

Figure 2: Transformations of nitrogen in oxic and anoxic environments (Bernhard 2010).
Sources of organic nitrogen in highway stormwater runoff include vegetable and animal decay, and animal excrement. Sources of inorganic nitrogen include nitrogen fixation by nitrogen-fixing bacteria (and some by lightning strikes), nitrification by nitrifying bacteria, and synthetic fertilizers. Biofiltration studies vary greatly on the results of nitrogen removal. In a study focused on dissolved constituents, vegetated columns resulted in twice as much removal of TN (63-77%) as non-vegetated columns (Henderson et al. 2007). High NOx removal (65%-93%) of vegetated columns was also observed with net zero or leaching observed in non-vegetated columns. Ammonia was removed in all configurations with and without vegetation (72-96%).
8

Hunt et al. (2006) found that nitrate removal can vary greatly at the field scale with 13% removal in a biofilter with sandy fill absent of organic matter and 75% removal in media with abundant organic matter. Based on soil core analysis, the finer soil gradation and organic matter in the second filter may have resulted in pocket saturated zones, facilitating denitrification. Total Kjeldahl Nitrogen (TKN) removal showed an opposite trend resulting in a net equivalent total nitrogen removal of approximately 40%. Addition of organic matter to planting soil has been encouraged to facilitate plant growth; however, organic matter in many cases also contributes to increased leaching of nitrogen compounds during decomposition (Hunt et al. 2012). Studies have also confirmed nitrate leaching from biofiltration experiments (Davis et al. 2006; Zinger et al. 2013). One study measured increasing concentrations of dissolved nitrogen with depth in the filter media (Hatt et al. 2006). Leaching may be due to the decomposition of organic matter and the oxidation of captured ammonia to nitrate.
Saturated Zone
Installation of a saturated zone in the lower layers of a biofilter has been studied as a means to create anoxic conditions for denitrification to occur. Another benefit to this improvement is a lower velocity of water flowing through the filter due to a decrease in hydraulic head. This allows a longer contact time between the media and the pollutants in the stormwater. Similarly, retaining water in the bottom of the filter allows plants to utilize nutrients in this zone over time, potentially increasing removal (Glaister et al. 2014). Access to a constant source of water may also enhance the survival of plant plants during dry periods.
9

In an optimization study of 18 biofiltration columns planted with Carex appressa, varying depths of a saturated anoxic zone in the presence of a carbon source (wood chips) were tested (Zinger et al. 2007). With increasing saturated zone depth, ammonia and organic nitrogen removal slightly decreased while total nitrogen and NOx removal increased. When the saturated anoxic zone was 450 mm in depth for a 900 mm height by 375-mm diameter column, >99% NOx removal was achieved.
In a follow-up study, Zinger et al. (2013) studied the effects of a submerged zone on the removal of nitrogen as well as phosphorus, total suspended solids (TSS), and heavy metals when an existing biofilter was retrofitted with a saturated anoxic zone. In this case, the microcosms were planted with two previously determined nitrogen inefficient species, Dianella revolute and Microlaena stipoides, and one highly efficient species, Carex appressa (Read et al. 2008). Before retrofitting with a saturated zone, ammonia removal was consistently above 90% in all columns. Results for NOx agreed with previous studies (Davis et al. 2001; Read et al. 2008) in that leaching was observed. Dianella and Microlaena columns exhibited TN leaching while 45% to 65% removal from was observed with Carex. After the retrofitting, NOx leaching was reduced, in some cases to net zero, in Dianella and Microlaena columns, and NOx removal was enhanced in Carex columns. Ammonia removal was reduced in Dianella and Microlaena but unaffected in Carex. Dissolved organic nitrogen increased in all cases. Overall results indicated that vegetation choices that enhance nitrogen removal may be more effective than the presence of a saturated zone.
A North Carolina field study (Hunt et al. 2006) compared a constructed bioretention cell containing a saturated zone to a similarly constructed cell without a
10

saturated zone. Results showed no statistically significant differences in the total nitrogen outflow. An increase in ammonium and slight decrease in nitrate was noted when the saturated zone was present. Ammonification occurred at a faster rate under aerobic conditions than under anoxic conditions (Hunt et al. 2006).
In Barrett et al. (2013), filters constructed of masonry sand and loam sand media and planted with Big Muhly with a saturated zone showed slightly increased removal, but all configurations did not consistently increase nitrogen removal. This was a possible result of a submerged layer that was not thick enough to become anoxic consuming only the bottom 6 inches (one third) of the soil media.
Carbon Addition
As described above, a column study for an anoxic zone with a carbon source and a retrofitted column study without a carbon source showed that >99% nitrate removal was achieved with a carbon source (Zinger et al. 2007, 2013). Additionally, concentrations of 16S rDNA for nitrification and denitrification genes were present in high concentrations in areas containing high readily degradable material, suggesting that additional compost in a saturated layer may enhance denitrification (Chen et al. 2013).
In a comprehensive optimization study by Kim et al. (2003), alfalfa, leaf mulch compost, newspaper, sawdust, wheat straw, and wood chips were compared as potential electron donors in a saturated zone. Sulfur-limestone and sulfur-only particles of varying size were also tested as inorganic substrates for chemolithotrophs. This study focused on microbial activity; no plants were involved. All columns performed better than the control column which was submerged without a carbon supplement. Alfalfa and wheat
11

straw both showed removal of greater than 95%; however, high TKN and turbidity was discharged. Results showed newspaper, wood chips, and small sulfur-limestone particles as the most effective electron donors. A flow study also displayed the resilience of bioretention systems; they continued to remove 90% nitrate after recovering from long drought periods (30 and 84 days). The provided explanation was that microbes switched to alternate metabolisms when stormwater was not entering the system and thus needed to recover once nitrogen species were reintroduced. As part of this study, pilot-scale bioretention boxes revealed complete removal of nitrate and nitrite species after remaining in the submerged zone for one week. A drawback of this method is that the carbon and nitrogen source will eventually need to be replaced as it degrades over time. Newspaper may exhibit the best longevity as the main constituent is lignin the ink prevents microbes from attacking the entire cellulose surface(Kim et al. 2003). A quick release and slow release carbon source may need to be combined such as a mixture of sawdust and hardwood mulch for optimum long-term treatment (Glaister et al. 2014).
Phosphorus
Sources of phosphorus in highway stormwater runoff include leaf decay from trees, fertilizers, and lubricants Studies have shown that total phosphorus can be greatly reduced within a biofilter because a majority of phosphorus is associated with particulate matter (Glaister et al. 2014; Hatt et al. 2007). In a study testing six different filter media types, total phosphorus was shown to have high removal in the upper portion of a soilbased filter; however, soluble phosphorus concentrations increased as a semi-synthetic stormwater flowed through the filter (Hatt et al. 2007). A follow up field study indicated
12

that increased levels of phosphorus in the effluent may be due to high phosphorus content of the filter media (Clark and Pitt 2009; Hatt et al. 2009). This agrees with a field study in North Carolina in which phosphorus removal ranged from 65% to -240%, consistent with phosphorus concentration in the soil media (Hunt et al. 2006). Opposing studies have found that media depth showed no effect on total phosphorus or orthophosphate concentrations (Bratieres et al. 2008), or that greater removal of orthophosphate (7080%) is found in the middle to bottom depths of pilot bioretention box filters (Davis et al. 2001). The latter study indicated that removal was likely due to favorable sorption to clay particles at a neutral pH. When TSS was not added to the synthetic stormwater mixture, 80% removal of total phosphorus was observed with 90-100% removal of orthophosphate (Henderson et al. 2007).
The effect of a saturated zone is unclear for phosphorus removal. Barrett et al. observed increased removal in the presence of a saturated zone (2013) while other studies indicate increased mobility of sorbed (where the term sorption is used to indicate mechanisms of both adsorption and absorption) phosphorus from soil surfaces (Clark and Pitt 2009; Zinger et al. 2013).
In a study comparing biofiltration media, fly ash was found to have a high potential for sorption of phosphorus when added as a supplement to the soil column. Fly ash was always mixed with soil since an entire column of fly ash can cause low hydraulic conductivity (Zhang et al. 2008).
13

Heavy Metals
In many biofiltration studies, indicator heavy metals have included copper, lead, and zinc. Common sources of copper include wear of bearings and brake linings, moving engine parts, fungicides, and insecticides (Burns 2012). Lead sources include automobile exhaust, wear of tires and bearings, and lubricating agents while zinc sources include oil, grease, and wear of tires (Burns 2012).
Results from multiple studies showed that metals removal was very high in biofiltration systems (Hatt et al. 2009; Hsieh and Davis 2005; Mitchell et al. 2011; Zinger et al. 2013). Removal was typically attributed to accumulation in soil and mulch due to their high organic matter content. Metal concentrations in the upper mulch layer were 2-3 times greater than measured in the soil media in a Massachusetts study (Davis et al. 2001). Studies also indicated that metals assimilation into plant material accounted for 5% or less of heavy metals removal (Davis et al. 2001; Dietz and Clausen 2006). Increased removal of heavy metals was observed in biofilters with a saturated zone in rain gardens (Dietz and Clausen 2006), while no effect was observed in column study (Zinger et al. 2013). Plants were observed to have very weak to no correlation with heavy metals removal (Read et al. 2010).
Suspended Solids
Sources of solids include wear of pavements and vehicles as well as atmospheric depositions (Burns 2012). Studies reviewed indicated a minimum of 76% TSS removal
14

by biofiltration (Barrett et al. 2013; Bratieres et al. 2008; Hatt et al. 2009; Hsieh and Davis 2005; Mitchell et al. 2011).
Objectives
Studies have shown that the species of vegetation within a biofilter has an impact on the performance, most specifically for nitrogen removal. Studies have shown varying success for total nitrogen removal with the installation of a saturated zone; more successful performance was observed when a carbon source was added within this layer. Phosphorus removal also varied as a function of the soil media and the presence of a saturated zone. In all studies (field and column), the removal of heavy metals and suspended solids was high.
Vegetation native to the southeastern United States has not been studied for biofiltration performance. The principal objective of this work was to identify the nutrient uptake efficiency of common Georgia native grasses as well as the inclusion of a saturated anoxic zone with an additional carbon source in typical Georgia topsoil biofiltration system. In addition to the study of native species, the additional plants chosen for study were selected in accordance with GDOT specifications.
Experimental Investigation
Materials The soil used to support plant growth consisted of gravel, sand, and mulch.
Number 7 coarse aggregate was donated by the Vulcan Materials Company (Forest Park, Georgia). Number 10 sand was obtained from Sand-Rock Transit (Atlanta, Georgia).
15

Both sand and gravel sources were pre-approved by GDOT (GDOT 2014). Topsoil and hardwood mulch were obtained from Green Brothers Earthworks (Marietta, Georgia). All materials were used as received. Three columns were tested with biomass fly ash incorporated into the soil substrate. The biomass ash used was formed from the combustion of forest, sawmill, and urban wood waste (Yeboah et al. 2014). The ash had a residual carbon content of 22.4%, loss on ignition (LOI) of 46.7%, and specific surface area of 116 m2/g (Yeboah et al. 2014).
Biofiltration columns were constructed from polyvinyl chloride (PVC) pipe with dimensions 813-millimeter height by 203-millimeter diameter (32-inch height by 8-inch diameter). Drainage outlets of 12.7 millimeters-( inch-) diameter were installed approximately 38 millimeters (1.5 inch) above the bottom of each column. Before packing, the columns were cleaned of all coolants and oils used during construction and rinsed in a hydrochloric acid solution. Columns were then thoroughly rinsed with tap water. Rubber test caps were added to the bottom of the columns and tightened. Gravel was added to the bottom of the columns to form a 6-inch layer thickness and hand tamped. Number 10 natural concrete sand was then added to 16 of the columns to form a 10-inch layer thickness over the drainage gravel. In 16 additional columns, natural sand was hand mixed with 5% hardwood mulch by volume and added for a 10-inch layer thickness over the drainage gravel (as constructed Figure 3).
16

8" Diameter
2" Ponding 2" Hardwood Mulch

8" Diameter
2" Ponding 2" Hardwood Mulch

12" Topsoil

12" Topsoil

10" Sand or Sand + 5% Mulch

1" Ash 9" Sand or Sand + 5% Ash

6" Gravel

6" Gravel

Figure 3: Biofiltration column configurations. 32 columns of configuration A (3A left) configuration and 3 columns containing
biomass ash in configuration B (3B left).
Three of the tested columns contained biomass fly ash. In one fly ash column, a natural sand layer was added to a 9-inch thickness with a 1-inch layer of biomass ash (Figure 3B). In the second column, a 9-inch layer of natural sand with 5% biomass ash by volume was added, with a 1-inch layer of biomass ash. The third column contained a 10inch layer of the sand/biomass mixture (Figure 3A), rather than mulch. All columns contain a 12-inch layer of silt loam topsoil (low-plasticity organic) with a hydraulic conductivity of 1.4E-4 cm/s and 8.3% organic matter, a 2-inch layer of hardwood mulch, and 2-inches of space for water ponding. A summary of all column configurations can be found in Table 2.

17

Table 2: Number of Replicates per Column Configuration

Saturated

Traditional with

Plant Species

Saturated

Traditional

with 5% ash 1" ash layer

in sand

above sand

None

2

2

-

-

Bermuda Grass

2

2

-

-

Big Bluestem

2

2

-

-

River Oats

2

2

-

-

Cherokee Sedge 2

2

-

-

Pink Muhly

2

2

-

-

Switchgrass

2

2

-

-

Indiangrass

2

2

2

1

Chemicals used to prepare the synthetic stormwater (Table 3) included lead nitrate (Pb(NO3)2), cupric nitrate hemipentahydrate (Cu(NO3)22.5H2O), zinc nitrate hexahydrate (Zn(NO3)26H2O), sodium nitrate (NaNO3), ammonium nitrate (NH4NO3), glycine (C2H5NO2) and sodium phosphate diabasic anhydrous (Na2HPO4). Sodium metabisulfite (Na2S2O5) was used for dechlorination of tap water. All chemicals were of certified grade from Fisher Scientific.

18

Pollutant

Table 3: Synthetic Stormwater Formulas

Field

FHWA

Measured Study

(Burns (Driscoll

2012)

1990)

Average

Metals

Synthetic

Spike

Stormwater (9/1/14)

Nutrient Spike (9/22/14)

Source Chemical

Total Phosphorus (mg P/L) 0.08-1.29 0.65 Total Nitrogen (mg N/L) 1.20-3.40 -

0.74 0.17 0.74

3.60

Na2HPO4

3.19 0.42 4.66 0.59 15.90 0.15 -

Nitrate (mg N/L)

0.65-1.20 0.61

Ammonium (mg N/L)

-

-

Organic Nitrogen (mg N/L) -

1.60

1.52 0.28 2.27 0.05 3.80

0.65 0.18 1.03

3.33

1.01

1.36

8.77

NaNO3 NH4NO3 C2H5NO2

Copper (mg/L) Lead (mg/L)

0.03

0.05

0.11 0.03 0.33

0.16

0.01

0.34

0.25 0.07 0.73

0.35

Cu(NO3)2 Pb(NO3)2

Zinc (mg/L)

0.12

0.20

0.44 0.13 1.58

0.71

Zn(NO3)2

An 8 foot by 12 foot Palram Snap & Grow greenhouse was constructed to protect the columns from precipitation (Figure 4). The greenhouse was located on the Georgia Tech campus in Atlanta, GA. The doors to the greenhouse remained open and two of the three back panels were removed to allow air to flow freely throughout the building. Columns were elevated in the greenhouse with cinderblocks to encourage drainage by gravity into sampling buckets. A 30% shade tarp was placed over the top of the greenhouse for temperature control.

19

Figure 4: View of the greenhouse on sampling day with all columns, blue sample buckets, batch mixing can, and shade tarp.
Grass Species Native grass species were selected based on ability to withstand flood and drought
conditions, sunlight needs, and availability. Species tested in this study included Andropogon gerardii (Big Bluestem), Muhlenbergia capillaris (Pink Muhly), Chasmanthium latifolium (River Oats), Panicum virgatum (Switchgrass), Sorghastrum nutans (Indiangrass), and Carex cherokeensis (Cherokee Sedge). Cynodon dactylon (Bermuda grass) was used as a control. River oats are listed on the GDOT Native Grass Seeding Table for cool weather planting, while switch grass and Indiangrass are listed for
20

warm weather planting (Section 700, GDOT Standard Specifications). All grasses except Indiangrass were obtained in quart containers from Niche Gardens (Chapel Hill, NC), while Indiangrass was donated in 4 in. plugs from Baker Environmental Nursery (Hoschton, GA). Three Indiangrass plugs were approximately equal in size to one quart container. Nursery soil was removed from root systems to extent possible before transplant to biofilter columns. All species were planted in four test columns. Two columns were designed with a saturated layer, and two columns were designed as free draining (traditional). Indiangrass was planted in columns containing biomass ash. Columns planted with Bermuda grass were cut down to approximately 1.5-inch height approximately once per month to avoid decomposition of grass in the column and to replicate mowing in field conditions.
Methods
Synthetic Stormwater
Synthetic stormwater was used to provide comparable, consistent control of the inflow constituents and to reduce experimental artifacts. The concentration of contaminants (Table 3) was formulated with reference to a previous characterization study of highway stormwater runoff performed at the Canton Sand Filter at the interchange of I-575 and GA 20 in Canton, Georgia (Burns 2012), as well as the average concentrations from the Federal Highway Administration's (FHWA) characterization study of North Carolina, Florida, and Tennessee highway runoff (Driscoll et al. 1990). Suspended solids were not added in this study due to the variable concentration of contaminants that has been observed with this practice. Instead, dissolved constituents were the main concern, since it is well established that TSS are removed at extents >88% in biofiltration columns (Barrett et al. 2013; Bratieres et al. 2008; Davis et al. 2001). The
21

performance of the columns was measured in terms of removal of dissolved contaminants.
Tap water was added to a polyethylene batch can (Figure 4) container and mixed with 1725 rotations per minute (rpm), 1/3-horsepower (hp) mixer motor, and 48-inch dual propeller shaft for approximately five minutes. Free chlorine was then measured using free chlorine micro check test strips (HF Scientific). Sodium metabisulfite (Na2S2O5) was added for dechlorination at a ratio of 1.34 parts Na2S2O5 per 1.0 part residual chlorine and mixed for a minimum contact time of five minutes (US EPA 2000). Chlorine concentrations were then measured to assure dechlorination was achieved before adding contaminants. Contaminants were mixed with the dechlorinated tap water for approximately fifteen minutes.
Synthetic stormwater was pumped from the tank through submersible pumps and delivered to the plants through a inch PVC irrigation system at a rate of approximately 0.4 L/min. Each column received approximately 11 liters with each watering event (AMEC Earth and Environmental et al. 2001). Watering was typically completed in two doses with an hour in between to avoid overfilling, particularly in ash columns, which had an observed lower hydraulic conductivity. Stormwater was sampled as it came out of the irrigation system. Samples were collected in 3.5-gallons buckets after complete drainage. Buckets were mixed thoroughly and a 1-liter sample was transported to the lab in polyethylene bottles for analysis.
22

Sampling Schedule

All grasses were planted by April 17, 2014. Plants were watered with

approximately 2.9 gallons of tap water twice weekly. A 30% shade tarp was installed to

reduce temperatures inside the greenhouse to ambient outdoor temperatures. The first

synthetic stormwater dosing was on June 6, 2014, with continued dosing twice a week

(typically on Mondays and Thursdays). A whitefly infestation was identified on June 19,

2014. The plants were hosed with tap water to wash whiteflies from the leaves and stems.

An insecticidal soap was sprayed lightly on the leaves to remove any surviving flies. No

impact of the insecticidal soap was observed on the measured level of dissolved metal or

nutrient concentrations. The saturated layer condition was imposed on June 30, 2014.

After a final pressure rinse on July 2, 2014, the whiteflies seemed to be eliminated with

continued monitoring thereafter through yellow sticky traps. Treated sampling was

conducted on a regular schedule throughout the summer months (Table 4).

Test Condition

Table 4: Sampling Schedule
Number of Samples Sample Dates

Average Stormwater

4

7/7/14, 7/21/14, 8/4/14, 8/18/14

Metals Spike

1

9/1/2014

Nutrient Spike

1

9/22/2014

Average after 2-week drought 1

10/6/2014

Watering with stormwater spiked with heavy metals was performed on September 1, 2014. Watering with stormwater spiked with nutrients was performed on September 22, 2014 (09-22-14). Watering ceased after the 09-22-14 watering event to impose
23

drought conditions. Saturated layers were depleted due to plant uptake and evaporation by the end of the drought period. On October 6, 2014, a final sample was collected after a dosing with an average synthetic stormwater mixture.
Plant height was measured at the end of the study for all columns. A column of each configuration was also cut open to measure the depths and observe the density of root growth within the column. Soil was shaken loose of root systems to determine maximum root depth within the soil column. All grasses except Bermuda grass were planted at an operational underground sand filter in Canton, Georgia to compare survival in the greenhouse to survival in an outdoor biofiltration setting.
Sample Analysis
Collected samples were immediately tested for pH (XL60, Accumet) and turbidity (TB-200, Orbelco). They were then filtered through 0.45 m Millipore nylon syringe filters. Samples were analyzed for nitrate, nitrite, phosphate, and ammonium via ion chromatography (ICS-1100, Dionex). An AS22 column and AERS 500 suppressor were utilized for anions with a 4.5mM sodium carbonate: 1.4mM sodium bicarbonate eluent. A CS16 with an ERS 500 suppressor and 36 mM methanesulfonic acid eluent was used for ammonium analysis. Samples were digested via Standard Methods 4500-P J, Persulfate method (APHA et al. 2012) to convert all nitrogen and phosphorus forms to nitrate and orthophosphate, respectively. Digested samples were measured through ion chromatography for total nitrogen while total phosphorus was measured through spectrophotometry (UV-1800, Shimadzu) at 880 nm via Standard Methods 4500-P E (APHA et al. 2012). Samples were prepared with 5% nitric acid and 1 ppm yttrium for
24

analysis of copper, lead, and zinc through inductively coupled plasma optical emission spectroscopy (Optima 8000, Perkin Elmer).
The results from all columns were averaged between two replicates, except the traditional Indiangrass plus biomass ash column which did not have a replicate (biomass column was a proof of concept test). For nitrogen, effluent concentrations are shown as mg N/L by form (nitrate + nitrite, ammonium, or organic) to indicate total nitrogen make up. Results for the first four collection dates in which columns were consistently dosed with an average synthetic stormwater are averaged and presented in the results section. For the synthetic stormwater spiked with metals, synthetic stormwater spiked with nutrients, and average stormwater after a two-week drought, replicate columns were averaged for the single sampling event. Turbidity and pH data were collected for all sampling events except the first on July 7, 2014.
RESULTS AND ANALYSIS
Outflow from the biofiltration columns was collected and analyzed for nitrogen, phosphorus, and metal removal extents. The following figures summarize removal results as a function of species and test conditions. Removal was calculated by subtracting the effluent concentration by the influent concentration and dividing by the influent concentration. The data are first analyzed for contaminant removal, and then presented in terms of removal for each plant species.
25

Nitrogen Nitrogen results are displayed in terms of concentration in order to highlight the
proportion of nitrate and nitrite, ammonia, and organic nitrogen in relation to total nitrogen. During the first two months of stormwater monitoring, total nitrogen leaching was commonly observed. Removal was achieved in columns planted with Big Bluestem, Switchgrass, and Indiangrass in saturated columns. Significantly greater removal (p < 0.0002) was found in the saturated condition as compared to the traditional condition (Figure 5 and Figure 6).

Nitrogen Concentration (mg N/L)

12

10

Organic Nitrogen

Ammonia

8

Nitrate + Nitrite

6

4

2

0

Plant Species
Figure 5: Nitrogen species concentration means by plant species with influent concentration (dashed lines represent traditional column effluent dosed with
average synthetic stormwater).

26

Nitrogen Concentration (mg N/L)

12

10

Organic Nitrogen

Ammonia

8

Nitrate + Nitrite

6

4

2

0

Plant Species
Figure 6: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent saturated columns dosed with average synthetic stormwater).
In the traditional columns, total nitrogen leaching was common across all species, ranging from 4% to 76% net export. In the saturated condition, total nitrogen concentration ranged from 5% export to 70% removal in the case of Big Bluestem. The Indiangrass column with biomass ash lenses resulted in an average 32% removal of total nitrogen in the aerobic condition and 11% removal in the saturated condition. Ammonia concentrations typically decreased, while nitrate concentrations increased in the case of traditional columns, which was consistent with the nitrogen degradation processes occurring in aerobic conditions. The saturated layer increased denitrification as expected with nitrate removals ranging from -55% to 46% in the traditional configuration and 27% to 79% in the saturated configuration; however, increased concentrations of ammonia were observed. The presence of biomass ash seemed to be more effective in the reduction of nitrogen species in the traditional condition when compared to the saturated condition,
27

with removal extents of 75% and 54% respectively. Organic nitrogen in the soil seemed to largely contribute to leaching. Big Bluestem, Switchgrass, and Indiangrass all showed positive removal extents in descending order.
When the stormwater inflow was spiked with heavy metals, an overall increase of nitrogen leaching was observed, especially in both the traditional and saturated configurations of columns planted with Bermuda grass (Figure 7 and Figure 8).

Nitrogen Concentration (mg N/L)

50

45

Organic Nitrogen

40

Ammonia

35

Nitrate + Nitrite

30

25

20

15

10

5

0

Plant Species
Figure 7: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent traditional columns dosed with
metals spiked stormwater).

28

Nitrogen Concentration (mg N/L)

25 Organic Nitrogen
20 Ammonia Nitrate + Nitrite
15
10
5
0
Plant Species
Figure 8: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent saturated columns dosed with metals
spiked stormwater).
The metal spiked synthetic stormwater contained a slight increase of nitrate (1.52 to 2.27 mg N/L) and total nitrogen (3.19 to 4.66 mg N/L) because metal chemicals were in the form of nitrate. Watering with this stormwater resulted in much larger amounts of total nitrogen leachate from the columns in both the saturated and traditional configurations. Saturated columns exported nitrogen at removal extents of -7% to -115% for native grasses, -237% for Bermuda grass, and -136% in the control. Traditional columns ranged from -283 to 53% for Pink Muhly and Indiangrass respectively. The leaching nitrogen was predominantly in the form of organic nitrogen because NOx removal and NH4 removal increased in all columns.
The third stormwater dosage type with spiked nutrient concentrations resulted in similar trends to that of the metal spiked stormwater, when compared to the average stormwater experiments (Figure 9 and Figure 10).
29

Nitrogen Concentration (mg N/L)

50

45

Organic Nitrogen

40

Ammonia

35

Nitrate + Nitrite

30

25

20

15

10

5

0

Nitrogen Concentration (mg N/L)

Plant Species

Figure 9: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent traditional columns dosed with
nutrient spiked stormwater).

25

Organic Nitrogen

20

Ammonia

Nitrate + Nitrite

15

10

5

0

Plant Species
Figure 10: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent saturated columns dosed with nutrient spiked stormwater).
30

All nitrogen concentrations were intended to be increased five times the average synthetic stormwater mixture; however, nitrate concentrations were measured to be approximately 2.5 times the average. With these increased concentrations, TN removal suffered slightly in most cases with overall removal in the traditional columns ranging from -174% with Cherokee Sedge and -17% with Big Bluestem. In the saturated condition, TN removal ranged from 2% to 42% with Switchgrass and Big Bluestem respectively. Removal of nitrate was greatly enhanced up to 72% and 92% with Big Bluestem in the traditional and saturated conditions, respectively. Similarly, percent removal for ammonium and Big Bluestem were 67% traditional and 74% saturated.
Lastly, results for the final experiment in which columns were dosed with an average stormwater mixture after two weeks of drought have mixed results as compared to the regular dosing of average stormwater (Figure 11 and Figure 12).

Nitrogen Concentration (mg N/L)

50

45

Organic Nitrogen

40

Ammonia

35

Nitrate + Nitrite

30

25

20

15

10

5

0

Plant Species
Figure 11: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent traditional columns dosed with an
average synthetic stormwater after two weeks of drought conditions).

31

Nitrogen Concentration (mg N/L)

25

20

Organic Nitrogen

Ammonia

15

Nitrate + Nitrite

10

5

0

Plant Species
Figure 12: Nitrogen species concentration means in effluent by plant species with influent concentration (dashed lines represent saturated columns dosed with an
average synthetic stormwater after two weeks of drought conditions).
To compare the trends of the different column configurations and stormwater doses, the traditional configuration was used as the baseline, and removals for different column configurations were subtracted from that of the traditional configuration with average stormwater (Figure 13 for total nitrogen, Figure 14 for nitrate + nitrite, and Figure 15 for ammonium). The traditional column configuration with metals spiked synthetic stormwater was the most common condition observed for high extents of total nitrogen leaching. Saturation tended to increase TN removal in almost all cases, except in the presence of biomass fly ash.
The carbon addition of mulch in the saturated layer will degrade over time altering the life of the biofilter. In the long term, decaying roots will become readily degradable carbon sources as plants turnover their root systems through the growing season.
32

Figure 13: Difference of total nitrogen removal from the traditional configurations dosed with average stormwater.
33

Figure 14: Difference of nitrate removal from the traditional configurations with average stormwater.
34

Figure 15: Difference of ammonia removal from the traditional configurations with average stormwater.
35

Nitrate was removed at variable extents within vegetated columns ranging from -17% to 92% with the highest removal observed in the saturated, nutrient spiked experiment and lowest rate observed in traditional columns after a drought period. Across all experiments, nitrate removal ranged from 43% to 92% in saturated vegetated columns and -17% to 81% in traditional vegetated columns. Control columns with Bermuda grass or non-vegetated had nitrogen removals ranging from -139% (traditional) to 74% (saturated). The presence of biomass ash decreased the removal of nitrate in all experiments. High removal extents of ammonium were observed in the traditional columns with aerobic conditions as opposed to saturated columns. The highest removals of total nitrogen were typically observed from Big Bluestem, Switchgrass, and Indiangrass. Phosphorus
Total phosphorus removal ranged from 50 to 90% in the saturated condition and 47 to 67% in the traditional, free-draining condition (Figure 16). Greater removal in the saturated condition ( p < 0.001) supported the findings of Bratieres et al. (2008) but conflicted with those of Zinger et al. (2013). High removal in the traditional condition also agreed with previous work (Henderson et al. 2007). Columns including ash exhibited the lowest removal efficiency for phosphorus. In most experiments, the non-vegetated columns exhibited similar removal as compared to vegetated columns, indicating that sorption may be the primary mechanism for removal of phosphorus.
36

Percent Removal

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

Saturated Traditional

Figure 16: Total phosphorus removal by plant species for columns dosed with average synthetic stormwater.
Increased concentrations of heavy metals in stormwater runoff were accompanied by increased removal of total phosphorus in both the saturated and traditional configurations (Figure 17).

37

Percent Removal

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

Saturated Traditional

Figure 17: Total Phosphorus removal by plant species for columns dosed with metal spiked synthetic stormwater.
Higher removal in the presence of heavy metals likely indicated precipitation reactions. These precipitation reactions are commonly used to immobilize heavy metals, specifically lead, in contaminated soils (Fang et al. 2012). When phosphorus concentrations were increased to approximately five times "average" concentrations, removal varied greatly (Figure 18).

38

Percent Removal

100% 80% 60% 40% 20% 0% -20% -40% -60% -80%

Saturated Traditional

Figure 18: Total phosphorus removal by plant species for columns dosed with nutrient spiked synthetic stormwater.
This may indicate a greater impact of plant uptake than in the average condition as the removal extent for the non-vegetated column in the saturated condition was significantly lower (40%) when compared to the grass species (57% to 94%). In this case, traditional columns typically noted an export of phosphorus for all species except Pink Muhly and Switchgrass. Ash columns exhibited high amounts of export (-44% removal) in the traditional condition. After the two week drought period, the saturated columns demonstrated higher removal than the traditional columns (Figure 19).

39

Percent Removal

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

Saturated Traditional

Figure 19: Total phosphorus removal by plant species for columns dosed with average synthetic stormwater after two weeks of drought conditions. Saturated columns maintained an average of 80 to 90% removal during average
conditions; however, traditional column removal dropped from around 60% to around 40% on average. Ash columns resulted in the least removal in both traditional and saturated conditions after the drought period.
When comparing all removals to that of the traditional configuration under average conditions (Figure 20), the results demonstrated that the two conditions of traditional configuration under high nutrient conditions and the traditional configuration after drought conditions showed greatly reduced performance. In contrast, a configuration with a permanent saturated layer enhanced performance. An increased concentration of heavy metals demonstrated the greatest enhancement to phosphorus removal; it is believed this was due to precipitation reactions within the filter media, but is a removal mechanism that will be explored in more detail in future studies.
40

Figure 20: Total phosphorus removal differences from traditional configuration with average synthetic stormwater.
41

Heavy Metals
Copper
For all column configurations and all synthetic stormwater formulas, copper was removed at extents greater than or equal to 82%. Removals in the traditional column, with monitored conditions were greater than 92%, with the saturated condition resulting in removals greater than 99% (Figure 21).

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 21: Copper removal by plant species for columns dosed with average synthetic stormwater.
The increase in removal observed in the saturated columns was statistically significant (p < 0.001).

42

Spiked metals concentrations in stormwater resulted in increased metals removal (Figure 22).

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 22: Copper removal by plant species for columns dosed with metal spiked synthetic stormwater.
Increased nutrient concentrations in the influent resulted in the lowest removal extents of copper in the traditional configuration but still larger than 82% in all cases (Figure 23).

43

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 23: Copper removal by plant species for columns dosed with nutrient spiked synthetic stormwater.
After two weeks of drought, copper removal was consistent with that of the
average stormwater conditions (Figure 24).

44

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 24: Copper removal by plant species for columns dosed with average synthetic stormwater after two weeks of drought conditions.
Through all experiments, variation in the uptake of copper among different plants species in comparison to non-vegetated columns was not observed (p > 0.05). This is consistent with previous findings, which demonstrated that plant uptake accounted for 5% or less for metal removal in a biofiltration setting (Davis et al. 2014; Dietz and Clausen 2006). In comparing copper removal by column configuration, saturation of the column resulted in the statistically significant increase of performance (p < 0.001) (Figure 25). The highest removal extents were observed when metals concentrations were increased in the saturated configuration, while lowest removals occurred in traditional columns with high nutrient concentrations. Variation in copper removal was typically within 10% of removals in the traditional column configuration with average stormwater.
45

Figure 25: Copper removal as compared to the traditional configuration dosed with average synthetic stormwater.
46

Lead
Results for lead removal in all configurations were greater than 97%, with no statistically significant differences observed in plant species or column configuration (Figure 26 through Figure 29). Concentrations were reduced from 250 ppb to 4 ppb on average across all columns during average synthetic stormwater dosing with one maximum effluent concentration of 18 ppb in one saturated Cherokee Sedge column.

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 26: Lead removal by plant species for columns dosed with average synthetic stormwater.

47

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 27: Lead removal by plant species for columns dosed with metal spiked synthetic stormwater.
Lead concentrations were reduced from 730 ppb to 6 ppb on average across all columns after the metals spiked synthetic stormwater dosing. Maximum effluent concentrations of 18 ppb were observed in saturated Indiangrass with biomass ash columns.

48

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 28: Lead removal by plant species for columns dosed with nutrient spiked synthetic stormwater conditions.

49

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 29: Lead removal by plant species for columns dosed with average synthetic stormwater after two weeks of drought conditions.
A comparison of all enhancements shows differences within 2% of the traditional configuration with average stormwater (Figure 30). Differenced between the saturated and traditional configuration were not considered statistically significant ( p > 0.05). A slight decrease in performance was observed in the nutrient spiked conditions (p < 0.01 in saturated and p < 0.0006 in traditional).

50

Figure 30: Lead removal as compared to the traditional configuration with average synthetic stormwater.
51

Zinc
Zinc showed the greatest variation in removal for the three heavy metals measured. This is consistent with the column studies performed by Davis et al. (2001), which found zinc to have the lowest sorption to a sandy loam soil at neutral pH. Average removal for traditional columns ranged from 67% to 87%, while removal in the saturated columns ranged from 81% to 93% (Figure 31).

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 31: Zinc removal by plant species for columns dosed with average synthetic stormwater.
Consistent with the trends for copper and lead, zinc also showed increased removal when heavy metal concentrations were spiked (Figure 32). In this condition,
52

removals were greater than 94% and 97% in the traditional and saturated conditions,

respectively.

100%

Saturated Traditional

80%

Percent Removal

60%

40%

20%

0%

Figure 32: Zinc removal by plant species for columns dosed with metal spiked synthetic stormwater.
High variation was observed in the zinc removal with nutrient spiked stormwater and average stormwater after a drought period (Figure 33 and Figure 34). While the addition of soluble phosphorus may have caused increased precipitation of lead phosphate and thus immobilization in the soil, one study showed the increase of leachable zinc from heavy metal contaminated soil media (Fang et al. 2012).

53

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 33: Zinc removal by plant species for columns dosed with nutrient spiked synthetic stormwater.

54

Percent Removal

100% 80% 60% 40% 20% 0%

Saturated Traditional

Figure 34: Zinc removal by plant species for columns dosed with average synthetic stormwater after two weeks of drought conditions.
The lowest removal extent of zinc (34%) was observed in a traditional Switchgrass column after a two week drought period. When comparing all zinc removals to that of the traditional configuration with average stormwater, trends observed for copper and lead remain consistent with zinc (Figure 35) with the highest removal observed in the saturated, metals spiked conditions. Zinc results exhibited far greater variability when compared to lead and copper, making removal trends less evident.

55

Figure 35: Zinc removal as compared to the traditional configuration with average synthetic stormwater.
56

Turbidity
Although suspended solids were not added to the synthetic stormwater, noticeable differences in the turbidity of samples were observed upon collection. Turbidity was measured as a surrogate for total suspended solids as gravimetric analysis would have required large volumes of water for the low concentrations observed. Data for all but the first sampling event are shown in Figure 36 through Figure 39 below. Inflow turbidity was measured as 1.11 0.37 NTU except in the case of the metals spiked stormwater which was 4.68 NTU.

Turbidity (NTU)

90

80

Saturated

70

Traditional

60

50

40

30

20

10

0

Figure 36: Turbidity by plant species for columns dosed with average synthetic stormwater.

57

Turbidity (NTU)

90

80

Saturated

70

Traditional

60

50

40

30

20

10

0

Figure 37: Turbidity by plant species for columns dosed with metals spiked synthetic stormwater.

90

80

Saturated

70

Traditional

60

50

40

30

20

10

0

Turbidity (NTU)

Figure 38: Turbidity by plant species for columns dosed with nutrient spiked synthetic stormwater.
58

Turbidity (NTU)

250

Saturated

200

Traditional

150

100

50

0

Figure 39: Turbidity by plant species for columns dosed with average synthetic stormwater after two weeks of drought conditions.
For all sampling events, the saturated configuration resulted in less turbidity in the effluent than the traditional configuration. This is attributable to decreased velocities created by the saturated zone, which allows fine solids to settle and filter from the effluent. Columns configured with a saturated layer and an ash/sand mixture consistently yielded very low turbidity results. pH
Effluent pH was measured immediately after sample collection. Influent pH was adjusted to 7.0. The pH typically remained just less than 7.0 in all cases except those that
59

p H

included biomass ash (Figure 40 through Figure 43). The pH exceeded 7.0 consistently in biomass columns with a saturated zone.
Saturated 9
Traditional 8 7 6 5 4 3 2 1 0
Plant Species
Figure 40: pH of treated average synthetic stormwater.
60

p H

9

Saturated

Traditional

8

7

6

5

4

3

2

1

0

Plant Species
Figure 41: pH of treated metals spiked synthetic stormwater.

Saturated

9

Traditional

8

7

6

5

4

3

2

1

0

p H

Plant Species
Figure 42: pH of treated nutrient spiked synthetic stormwater.
61

p H

Saturated 9
Traditional 8 7 6 5 4 3 2 1 0
Plant Species
Figure 43: pH of treated average synthetic stormwater after two weeks of drought. When compared to Indiangrass columns without ash, the observed increase of pH
in the presence of biomass ash was statistically significant in both the traditional and saturated configurations (p < 0.006 and p < 0.0005 respectively). The increase in pH was likely not large enough to enhance precipitation of metals from solution through complexation with hydroxides.
Plant Growth
Grasses at the Canton, GA sand filter were planted on May 4, 2014; however, mowing took place in early June. Reported heights are considered from approximately 3inch height on June 21, 2014 until a final height measurement on August 13, 2014 along with measured heights of greenhouse grown grasses (Table 5). Measurements for greenhouse grown grasses were taken on October 16, 2014.
62

Table 5. Root depth and height measurements for column and field plants. Root depth Root depth Plant Height, Plant Height, Plant Height,

Saturated

Traditional Saturated

Traditional Field

(inches)a

(inches)a

(inches)b

(inches)b

(inches)c

Bermuda

12

10

12

12

n/a

Big Bluestem 28

28

56.5 2.5 38 11

59

River Oats

15

16

41.5 5.5 37 2

16

Cherokee Sedge 22

13

35 3

31.5 0.5 14

Pink Muhly

12

12

48

42.5 1.5 19

Switchgrass

28

28

48 1

48 1

39 2.5

Indiangrass

22

12

53 19

52 12

57 7

Indian + Ash

24.5

28

64

63

n/a

Notes: a - Maximum soil media depth was 28 inches with 0 to 12 inches of planting soil, 12 to 22 inches sand or sand/carbon source mixture, and 22 to 28 inches drainage gravel. Measurement after six months of growth (April to October 2014) b Measurement after six months of growth (April to October 2014) c Measurement after two months of summer growth (June through August 2014)

63

Since the greenhouse grasses had approximately four months more time for growth, the measured heights were generally lower for the field planted varieties. Big Bluestem and Indiangrass reached comparable heights, while all others were shorter at comparable lengths of growth time.
Soil was shaken loose from root systems to determine the greatest root depths in the entire soil column. Root depth varied significantly in the grasses, with roots from the Big Bluestem and Switchgrass penetrating completely and vigorously through the gravel layer. Indiangrass and Cherokee Sedge showed a significantly greater root depth in the saturated configuration than traditional. Indiangrass also seemed to penetrate deeper into the sand layer and even through the gravel in the presence of biomass ash. This root density may explain some variation in pollutant uptake in accordance with previous studies on plant characteristics (Read et al. 2010). Field constructed biofilters will contain much greater depths of planting media which may lead to growth remaining in the planting soil rather than growing through the sand layer. Summary
Nutrient removals for all four experiments were averaged to estimate the performance of each grass species across all conditions (Figure 44 and Figure 45). Heavy metal removal was excluded as variation among plant species was not evident. All native grasses performed more efficiently in terms of nitrogen removal as compared to control non-vegetated and Bermuda grass columns.
64

100% 50% 0% -50%
-100% -150% -200% -250% -300% -350% -400%

Total Nitrogen Nitrate + Nitrite Ammonia Total Phosphorus

Figure 44: Average nutrient removal across all experiments in the traditional configurations.

100% 50% 0% -50%
-100% -150% -200% -250% -300% -350% -400%

Total Nitrogen Nitrate + Nitrite Ammonia Total Phosphorus

Figure 45: Average nutrient removal across all experiments in saturated configurations.
65

Big Bluestem Among the six native grasses tested, Big Bluestem was among the top three
performing grasses in the saturated condition and top two in the traditional condition based on average removal across all experiments. This same trend applies to ammonium removal. In terms of NOx, Big Bluestem had the top removal in both traditional (45%) and saturated (85%) conditions across all experiments. Big Bluestem also had a maximum total phosphorus removal of 92% in the saturated condition. In both column configurations, Big Bluestem roots penetrated the entire depth of the soil media. River Oats
River Oats had medium to low removal of nutrients in comparison to other species studied. It had the lowest total phosphorus removal of all native species on average. River Oats leached total nitrogen in both saturated and traditional conditions (-23% and -91% respectively). NOx removal was 34% in the traditional and 78% in the saturated conditions which were in the middle range of all species. River Oats had shallow root systems in the columns extending slightly deeper than the topsoil layer with many thick roots along the inner perimeter of the column. Cherokee Sedge
Cherokee sedge had the lowest removal of total nitrogen and NOx across all experiments. It also showed the second lowest removal of phosphorus (83%), just behind River Oats in the saturated condition. This sedge had some of the shallowest and thinnest
66

diameter roots, particularly in the traditional condition, which may help explain low nutrient uptake.
Pink Muhly
Pink Muhly removal of all nutrients typically ranked in the middle of all native species. On average across all experiments, Pink Muhly leached total nitrogen in both traditional and saturated conditions (-116% and -4% respectively). Total phosphorus removal was 50% in the traditional columns and 85% in the saturated columns. Pink Muhly had the shallowest root system of any native species within the columns.
Switchgrass
Switchgrass typically had a very high performance with all nutrients when compared to other species. Switchgrass roots penetrated the entire depth of the column with dense roots in the lower sand and gravel of the traditional column. In the traditional configuration, Switchgrass showed one of the highest total nitrogen leachate extents with -131% removal; however, it had the maximum total nitrogen removal of 18% in the traditional configuration. This may be due to the difference in root density between the saturated and traditional columns in the lower portions of the filter. Switchgrass was the top total phosphorus remover in the traditional configuration (55%) and second in the saturated (86%).
Indiangrass
Lastly, Indiangrass was among the top two performers of total nitrogen (-50% traditional and 13% saturated) and top three performers of NOx (25% traditional and 72%
67

saturated) across all experiments. Indiangrass removed ammonia at some of the lowest percentages. While generally more shallow than Switchgrass and Big Bluestem, roots of Indiangrass were very dense in the planting soil layer as compared to Cherokee Sedge and Pink Muhly.
CONCLUSIONS
On average across all experiments, total nitrogen was removed at the highest percentage of 18% with Switchgrass and was leached at an average of 143% with Bermuda grass when both were grown with a saturated layer in the soil column. These results demonstrated the significant differences in nitrogen removal based on the vegetation type. Among these six native species, top recommendations include Big Bluestem and Switchgrass for consistent, high removal extents. Saturation increased the removal of NOx in combination with any of the plants used in this study. Saturation also increased the removal of total phosphorus, typically to greater than 80% removal. Copper, lead, and zinc showed minimum removal extents of 82%, 97%, and 34%, with the highest removal in the saturated configuration. The removal of metals showed no correlation with plant species which indicates that sorption to the soil media was likely the primary mechanism responsible for metal removal. Biomass ash performance varied greatly through all experiments, with some high nutrient removals observed in the aerobic condition. Further field study should be performed to verify these results and eliminate some of the inherent errors that column studies allow.
68

REFERENCES
AMEC Earth and Environmental, Center for Watershed Protection, Debo and Associates, Jordan Jones and Goulding, and Atlanta Regional Commission. (2001). Georgia Stormwater Management Manual Volume 2: Technical Handbook.
APHA, AWWA, and WPCF. (2012). Standard Methods for the Examination of Water and Wastewater. Washington D.C.
Barrett, M. E., Limouzin, M., and Lawler, D. F. (2013). "Effects of media and plant selection on biofiltration performance." Journal of Environmental Engineering, 139(4), 462470.
Bernhard, A. (2010). "The Nitrogen Cycle: Processes, Players, and Human Impact." Nature Education, 3(10), 25.
Bratieres, K., Fletcher, T. D., Deleti, A., and Zinger, Y. (2008). "Nutrient and sediment removal by stormwater biofilters: a large-scale design optimisation study." Water Research, 42(14), 39303940.
Burns, S. E. (2012). Stormwater Controls for Pollutant Removal on GDOT Right-OfWay. Forest Park, GA.
Chen, X., Peltier, E., Sturm, B. S. M., and Young, B. (2013). "Nitrogen removal and nitrifying and denitrifying bacteria quantification in a stormwater bioretention system." Water Research, 47(4), 16911700.
69

Clark, S., and Pitt, R. (2009). "Storm-Water Filter Media Pollutant Retention under Aerobic versus Anaerobic Conditions." Journal of Environmental Engineering, 135(5), 367372.
Le Coustumer, S., Fletcher, T. D., Deleti, A., Barraud, S., and Poelsma, P. (2012). "The influence of design parameters on clogging of stormwater biofilters: a large-scale column study." Water Research, 46(20), 67436752.
Davis, A. P., Shokouhian, M., Sharma, H., and Minami, C. (2001). "Laboratory study of biological retention for urban stormwater management." Water Environment Research, 73, 514.
Davis, A. P., Shokouhian, M., Sharma, H., and Minami, C. (2006). "Water quality improvement through bioretention media: nitrogen and phosphorus removal." Water Environment Research, 78(3), 28493.
Davis, A. P., Shokouhian, M., Sharma, H., Minami, C., and Winogradoff, D. (2014). "Water quality improvement through bioretention: lead, copper, and zinc removal." Water Environment Research, 75(1), 7382.
Department of Water and Swan River Trust. (2007). Stormwater Management Manual for Western Australia Chapter 9: Structural Controls.
Dietz, M. E., and Clausen, J. C. (2006). "Saturation to improve pollutant retention in a rain garden." Environmental Science & Technology, 40(4), 13351340.
70

Driscoll, E. D., Shelley, P. E., and Strecker, E. W. (1990). Pollutant loadings and impacts from highway stormwater runoff, Volume III: Analytical Investigation And Research Report. Oakland, CA.
Fang, Y., Cao, X., and Zhao, L. (2012). "Effects of phosphorus amendments and plant growth on the mobility of Pb, Cu, and Zn in a multi-metal-contaminated soil." Environmental science and pollution research international, 19(5), 16591667.
Georgia Department of Transportation (GDOT). (2014). Qualified Products List.
Glaister, B. J., Fletcher, T. D., Cook, P. L. M., and Hatt, B. E. (2014). "Co-optimisation of phosphorus and nitrogen removal in stormwater biofilters: the role of filter media, vegetation and saturated zone." Water Science & Technology, 69(9), 19611969.
Gray, C.S., Burns, S.E., and Griffith, J.D. (2012). "The use of natural zeolites as a sorbent for treatment of dissolved heavy metals in stormwater runoff." GeoCongress 2012, American Society of Civil Engineers, Oakland, CA, pp. 3978 - 3987.
Hatt, B. E., Deleti, A., and Fletcher, T. D. (2007). "Stormwater reuse: designing biofiltration systems for reliable treatment." Water Science & Technology, 55(4), 201.
Hatt, B. E., Fletcher, T. D., and Deleti, A. (2009). "Hydrologic and pollutant removal performance of stormwater biofiltration systems at the field scale." Journal of Hydrology, 365(3-4), 310321.
71

Hatt, B. E., Siriwardene, N., Deletic, a., and Fletcher, T. D. (2006). "Filter media for stormwater treatment and recycling: the influence of hydraulic properties of flow on pollutant removal." Water Science & Technology, 54(6-7), 263.
Henderson, C., Greenway, M., and Phillips, I. (2007). "Removal of dissolved nitrogen, phosphorus and carbon from stormwater by biofiltration mesocosms." Water Science & Technology, 55(4), 183191.
Hsieh, C., and Davis, A. P. (2005). "Multiple-event study of bioretention for treatment of urban storm water runoff." Water Science and Technology, 51(3-4), 177181.
Hunt, W. F., Davis, A. P., and Traver, R. G. (2012). "Meeting hydrologic and water quality goals through targeted bioretention design." Journal of Environmental Engineering, 138(6), 698708.
Hunt, W. F., Jarrett, A. R., Smith, J. T., and Sharkey, L. J. (2006). "Evaluating bioretention hydrology and nutrient removal at three field sites in North Carolina." Journal of Irrigation and Drainage Engineering, 132(6), 600608.
Kim, H., Seagren, E. A., and Davis, A. P. (2003). "Engineered Bioretention for Removal of Nitrate from Stormwater Runoff." Water Environment Research, 75(4), 355367.
Lucas, W. C., and Greenway, M. (2008). "Nutrient retention in vegetated and nonvegetated bioretention mesocosms." Journal of Irrigation and Drainage Engineering, 134(5), 613624.
72

Mitchell, G. F., Riefler, R. G., and Russ, A. (2011). "Removal, by vegetated biofilter, of medium and low concentrations of pollutants from simulated highway runoff." Transportation Research Record, 2262, 214224.
Prince George's County. (2007). Bioretention Manual. Prince George's County Maryland.
Read, J., Fletcher, T. D., Wevill, T., and Deleti, A. (2010). "Plant traits that enhance pollutant removal from stormwater in biofiltration systems." International Journal of Phytoremediation, 12(1), 3453.
Read, J., Wevill, T., Fletcher, T. D., and Deleti, A. (2008). "Variation among plant species in pollutant removal from stormwater in biofiltration systems." Water Research, 42(4-5), 893902.
United States Environmental Protection Agency (US EPA). (2000). Wastewater Technology Fact Sheet: Dechlorination. Washington, D.C.
United States Environmental Protection Agency (US EPA). (2003). Protecting Water Quality from Urban Runoff Fact Sheet. Washington D.C.
Water Environment Federation (WEF), American Society of Civil Engineers (ASCE), and Environmental & Water Resources Institute. (2012). Design of Urban Stormwater Controls. McGraw Hill, Alexandria, Virginia, 313322.
73

Yeboah, N. N. N., Shearer, C. R., Burns, S. E., and Kurtis, K. E. (2014). "Characterization of biomass and high carbon content coal ash for productive reuse applications." Fuel, 116, 438447.
Zhang, W., Brown, G. O., Storm, D. E., and Zhang, H. (2008). "Fly-Ash-Amended Sand as Filter Media in Bioretention Cells to Improve Phosphorus Removal." Water Environment Research, 80(6), 507516.
Zinger, Y., Blecken, G.-T., Fletcher, T. D., Viklander, M., and Deleti, A. (2013). "Optimising nitrogen removal in existing stormwater biofilters: Benefits and tradeoffs of a retrofitted saturated zone." Ecological Engineering, 51, 7582.
Zinger, Y., Fletcher, T. D., Deleti, A., Blecken, G.-T., and Viklander, M. (2007). "Optimisation of the nitrogen retention capacity of stormwater biofiltration systems." Novatech Conference, Lyon, France, 893900.
74