Investigation of Nutrient Removal
Mechanisms of a Constructed Gravel
Wetland Used for Stormwater Control in a
Northern Climate
Prepared by
The University of New Hampshire Stormwater Center
In Collaboration with
The New England Interstate Water Pollution Control
Commission
Prepared under Support from U.S. Environmental Protection
Agency Assessment and Watershed Protection Grant
March 2010
Table of Contents
ACKNOWLEDGEMENTS ....................................................................................................... iv EXECUTIVE SUMMARY ........................................................................................................ 1 INTRODUCTION ...................................................................................................................... 1 BACKGROUND ........................................................................................................................ 2 Nitrogen .................................................................................................................................. 4 Analyte .................................................................................................................................... 5 Analytical Method .................................................................................................................. 5 Method Detection Limit ......................................................................................................... 5 Total Dissolved Nitrogen ........................................................................................................ 5 Phosphorus .............................................................................................................................. 5 SAMPLING METHODS ............................................................................................................ 6 Sample Collection and Analysis ......................................................................................... 7 Water Quality Measurements ............................................................................................. 8 Summary and Performance Statistics ................................................................................. 9 HISTORY AND MAINTENANCE ......................................................................................... 10 Aircraft Deicer Experiment................................................................................................... 10 System Maintenance ............................................................................................................. 10 Forebay Maintenance ............................................................................................................ 10 RESULTS ................................................................................................................................. 12 Long Term and Seasonal Results .......................................................................................... 12 Nitrogen ............................................................................................................................ 13 Phosphorus ........................................................................................................................ 15 Nutrient Cycling and Removal Mechanisms ........................................................................ 16 Nitrogen ............................................................................................................................ 16 Phosphorous ...................................................................................................................... 21 Real Time Water Quality Parameters ................................................................................... 21 Regression Analyses ............................................................................................................. 21 CONCLUSIONS....................................................................................................................... 23 REFERENCES ......................................................................................................................... 25 Appendix A--- Nitrogen speciation within gravel wetland ................................................. 27 Appendix B--- Phosphorus speciation in gravel wetland..................................................... 30 Appendix C--- Regression of EMC and Analysis of Variance ............................................ 33 Appendix D--- TSS Distributions for the first three years of operation............................... 36 Page i
September 2004-January 2007 .............................................................................................. 36 May-November ..................................................................................................................... 37 November-April .................................................................................................................... 38 Appendix E--- Total Phosphorus distributions for the first three years of operation .......... 39 September 2004-January 2007 .............................................................................................. 39 May-November ..................................................................................................................... 40 November-April .................................................................................................................... 41 Appendix F--- Nitrate distributions for the first three years of operation ........................... 42 September 2004-January 2007 .............................................................................................. 42 May-November ..................................................................................................................... 43 November-April .................................................................................................................... 44 Appendix G--- UNHSC Subsurface Gravel Wetland Design Specifications ...................... 45 Appendix H--- UNHSC Gravel Wetlands Planting Assessment – October 2006.............. 571 Page ii
Table of Figures
Figure 1: Nitrogen entering the wetland in runoff or plant detritus is converted to NH4, NO2
and NO3 in the oxygenated forebay. NO3 is converted to N2 in the low oxygen subsurface
cells. ........................................................................................................................................ 3
Figure 2: (a, left) Sampling locations at the UNHSC field facility. Samples were collected
from the influent, effluent, and 3 locations within the wetland; (b, right) Locations of water
quality measurements within the subsurface gravel wetland, Durham, NH ........................... 6
Figure:3: Oxygen concentrations in the subsurface gravel wetland effluent .......................... 9
Figure 4: TSS performance for the subsurface gravel wetland September 2004 –July 2009 13
Figure 5: Nitrate performance for the subsurface gravel wetland September 2004 –July
2009; yellow shaded area indicates period of ADF (deicer) experiment; vertical red lines
indicate period of forebay maintenance. ............................................................................... 14
Figure 6: Total phosphorous performance for the subsurface gravel wetland September
2004 –July 2009; yellow shaded area indicates period of ADF (deicer) experiment; vertical
red lines indicate period of forebay maintenance. ................................................................ 15
Figure 7: Total phosphorus concentration in effluent as a function of pH .......................... 16
Figure 8: Examination of nitrogen speciation (ug N/l) as it travels through gravel wetland
throughout storm. GW1 was not sampled during this event................................................ 17
Figure 9: Mass loading for DRO, Zn, NO3, TSS as a function of normalized storm volume
for two storms: (a) a large 60 mm rainfall over 1685 minutes; (b) a smaller 15 mm storm
depth over 490 minute. DRO=diesel range organics, Zn= zinc, NO3= nitrate, TSS= total
suspended solids.................................................................................................................... 18
Figure 10: Examination of phosphorous speciation (ug/l) as it travels through gravel
wetland throughout storm ..................................................................................................... 20
Figure 11: Real-time water quality measurements in the gravel wetland, collected weekly
April 2007-October 2008 ...................................................................................................... 22
Figure 12: TSS Regression of Event Mean Concentrations with ANOVA for Annual (a),
Summer (b), and Winter (c) from September 2004- January 2007 ...................................... 33
Figure 13: NO3 Regression of Event Mean Concentrations with ANOVA for Annual,
Summer, and Winter from September 2004- January 2007 ................................................. 34
Figure 14: TP Regression of Event Mean Concentrations with ANOVA for Annual,
Summer, and Winter from September 2004- January 2007 ................................................. 35
Table of Tables
Table 1: Subsurface Gravel Wetland Sizing and Design Criteria.......................................... 3
Table 2: Laboratory analytical methods and expected performance data for each analyte. ... 5
Table 3: Summary Performance Statistics for Gravel Wetland for September 2004- January
2007....................................................................................................................................... 11
Page iii
ACKNOWLEDGEMENTS
The study was conducted under funding from the FY06 EPA Assessment and Watershed
Protection Program Grants (AWPPG) as a NPS Program National Priority. The project was a
collaboration with the New England Interstate Water Pollution Control Commission and the
University of New Hampshire Stormwater Center.
Project Sponsors
US Environmental Protection Agency
Project Team
UNH Stormwater Center
Robert Roseen, PhD, PE
Alison Watts, PhD, PG
James Houle
Kim Farah, PhD
Thomas Ballestero, PhD, PE, PG, PH, CGWP
NEIWPCC
Laura Chan
Heather Gilbert
Marianna Vulli
Mike Jennings
Clair Whittet
Technical Advisory Committee
Betsy Dake, RI DEM
Michelle Daley, UNH
Sarina Ergas, University of Southern Florida (formerly UMass)
Bill McDowell, UNH
Megan Moir, City of Burlington, VT (formerly VT DEC)
Thelma Murphy, EPA Region 1
Sally Snyder, CT DEP
Kerry Strout, NEIWPCC
EPA Project Officer
Bryan Rittenhouse, EPA HQ
Page iv
Investigation of Nutrient Removal
Mechanisms of a Constructed Gravel
Wetland Used for Stormwater Control in a
Northern Climate
Final Report by the University of New Hampshire Stormwater Center
EXECUTIVE SUMMARY
Gravel wetlands have been found to be very effective at removing many contaminants,
including nutrients, from stormwater runoff. The University of New Hampshire Stormwater
Center has been monitoring nitrogen and phosphorus removal in a subsurface horizontal flow
gravel wetland since 2004. The long term monitoring shows that the annual median nitrate
removal rate is >95% for the first three years, prior to system maintenance, with summertime
performance > 98% and reduced winter performance of about 30%. Reduced performance in
the winter is likely due to reduced plant uptake and microbial activity in the winter, and
increased load from decomposing plant detritus in the late fall and winter. Nitrate removal
declined after three growing seasons with an observed increase in nitrate concentration in the
forebay indicating rerelease due to decomposition of organic matter. Long term phosphorus
data shows an annual median removal rate of 55% for the first three growing seasons with
increased performance in the winter of > 70%, likely related to increased dissolved oxygen
concentrations in winter runoff and to influent pH, while summertime TP removal is lower, at
about 50%. Detailed monitoring within the wetland system conducted from 2007-2009 found
that organic nitrogen is successfully being converted to nitrogen gas and removed from the
system. Concentrations of nutrients and suspended sediments in the forebay indicate the need
for periodic maintenance to remove standing vegetation. The style of maintenance and degree
of disturbance of sediments was also found to be important with respect to resuspension.
INTRODUCTION
This report presents the results of a nutrient removal study conducted by the University of
New Hampshire Stormwater Center (UNHSC) in collaboration with the New England
Interstate Water Pollution Control Commission (NEIWPCC). Stormwater runoff from urban
development and other human activities contributes to nonpoint source pollution within
watersheds and has a significant impact on water quality. Developing an efficient and effective
management strategy can be challenging for watershed managers. Best management practices
(BMPs), are increasingly being implemented, but detailed performance data on nutrient
removal and the importance of maintenance is often lacking. Managers and engineers need
the best available information on the nutrient removal capabilities of BMP options in order to
make the most effective management decision.
Page 1
A detailed study of the long-term sustainability of nutrient treatment in a horizontal subsurface
gravel wetland was conducted from late 2007 through the summer of 2009 in a wetland
constructed by the UNHSC in 2004. Phosphorous and nitrogen were measured at select
representative locations in the influent, the forebay, wells within the subsurface gravel
substrate, and in the effluent during eight storms from late 2007 to 2009. Samples were
collected and analyzed for total phosphorous, dissolved phosphorus, ortho-phosphate, total
nitrogen, particulate nitrogen, dissolved organic nitrogen, nitrite, nitrate, ammonia, and
dissolved organic nitrogen. Additional routine analyses were collected under the UNHSC
general monitoring program and include total suspended solids (TSS), total petroleum
hydrocarbons-diesel (TPH-D), and total zinc (Zn). Information collected during this study, and
under the UNHSC’s long term monitoring program have been used to identify and evaluate
nutrient removal processes, as well as long term and seasonal performance trends.
In early 2007, just prior to the initiation of the nutrient cycling study, an experiment was
conducted that resulted in significant impairment of the gravel wetland functionality, which in
turn influenced the outcomes of this study. The experiment examined the treatment
performance of engineered wetlands to remove propylene glycol (aircraft deicer) from
stormwater. This study placed an exceptionally high biological oxygen demand (BOD) load
on the system, and stimulated excessive biomass production in the form of plant and microorganism growth in the forebay and within the treatment cells. While the impact of the glycol
study is not entirely understood, the impact upon system functionality was substantial.
Performance trends following the glycol experiment are therefore difficult to substantiate,
however nutrient cycling trends and mechanisms were studied, and the confounding issues led
to a greater understanding of maintenance components and forebay functionality of the
subsurface gravel wetland system.
Constructed wetlands can be one element of an effective strategy for sustainable site design as
they are able to efficiently mimic the processes that occur in natural wetlands and have the
capacity to provide important biochemical cycling functions necessary to treat nonpoint source
pollution. Constructed wetlands can be used to reduce pollutant loading from a number of
varying land cover types including urban areas, agricultural, transportation and recreational
areas such as golf courses (Roussea et al. 2004). There are two main types of constructed
wetlands; surface and subsurface flow. Surface flow treatment wetlands are modifications of
traditional ponds with surface flow directed through a vegetated marsh system. Sub-surface
horizontal flow wetlands move stormwater through a permeable substrate such as gravel or
sand which functions as a filtration system. The surface of the substrate is vegetated, allowing
root growth and subsequent plant uptake. These systems combine biochemical removal
processes coupled with physical filtration operations, making them very effective options for
stormwater treatment.
BACKGROUND
The UNHSC subsurface gravel wetland is constructed with a pre-treatment sedimentation
forebay where stormwater runoff enters the system. The broad shallow forebay design allows
for settling of coarse particles and focused maintenance. Water then passes through a conduit
Page 2
Influent
Organic N from runoff and plant debris
Nitrification
NH4→N02→ NO3
Aerobic Zone
Forebay and surface of wetland
Denitrification
→N2 (gas)
Anaerobic Zone
Subsurface gravel
Figure 1: Nitrogen entering the wetland in runoff or plant detritus is converted to NH4, NO2 and NO3 in the
oxygenated forebay. NO3 is converted to N2 in the low oxygen subsurface cells.
Table 1: Subsurface Gravel Wetland Sizing and Design Criteria
Design specifications
Rainfall-runoff depth
Catchment Area
Treatment Peak Flow
10-Year Peak Storm Flows
Water Quality Volume
Resident Subsurface Water Volume
Total Treatment Volume
Treatment Volume Drain Time
Value
SI (English)
25.4 (1)
0.4 (1)
2450 (1)
8570 (3.5)
92 (3264)
22 (768)
114 (4032)
24-48
Unit
SI (English)
mm (in)
hectares (acre)
m3/day (cfs)
m3/day (cfs)
cubic meters (cubic feet)
cubic meters (cubic feet)
cubic meters (cubic feet)
Hours
into the first of two subsurface gravel cells. Water ponds on the surface of the cell, much as in
a dry pond, where additional settling and aerobic transformations occur. Water then moves
vertically into the gravel substrate through perforated standpipes. The flow moves
horizontally through the gravel and exits the system through a subsurface drain. The gravel
remains saturated at all times due to an elevated outlet structure thereby creating anaerobic
conditions. Effluent flow leaving the system is controlled by a hydraulic outlet structure which
Page 3
imposes orifice control. The subsurface gravel wetland design specifications are provided in
Appendix G---, and are available on the UNHSC website (www.unh.edu/erg/cstev). The
design criteria are summarized in Table 1 and system components are illustrated in Figure 1.
The system follows the specification listed in the Appendix G. The wetland was planted with a
combination of seeds and potted plants as listed in Appendix H. The plants selected were
native species chosen to provide a habitat function. The underlying soils of the system are
marine clays, with very low hydraulic conductivity, and as such no liner was needed.
Nitrogen
Nitrogen occurs in the environment in numerous forms; the most common are soluble or
particulate organic nitrogen, nitrate (NO3), ammonium (NH4), nitrite (NO2), and nitrogen gas
(N2O or N2). Nonpoint sources of nitrogen in urban and suburban environments include
atmospheric deposition, plant debris, pet waste and landscaping fertilizer. A small amount of
nitrogen enters the wetland directly through rainfall onto the system, rather than runoff, but
this amount is negligible (less than five percent). Nitrogen is transformed within the wetland
system in several ways, but is only removed by plant uptake and subsequent removal of the
plant material and biomass harvesting, or by transformation to gaseous forms which diffuse to
the atmosphere. Nutrients utilized by plants are removed from the water or sediments and
stored in plant tissue, either above or below ground (USEPA 2000, Garcia et al. 2004). When
the plant senesces in the fall, the leaves and stems die back and accumulate as detritus on the
surface of the substrate. If the plants are not harvested and removed from the wetland, the
accumulated nutrients will be released back into the sediment as the plant matter decays.
(Tanner 2004, Mayo and Bigambo 2005)
The conversion of organic nitrogen to nitrogen gas is a microbially mediated process which
requires both an aerobic and anaerobic environment. Through the process of nitrification, the
organic compounds are converted to ammonia, nitrite, and nitrate by microbial respiration in
the presence of oxygen. Denitrification, the conversion of nitrate to nitrogen gas, occurs only
in the absence of oxygen. The subsurface gravel wetland provides these functions in the open,
aerated forebay, the surface of each treatment cell, and the subsurface, fully saturated gravel
layer (Figure 1). The full conversion of organic nitrogen requires adequate time in each of the
aerobic/anaerobic environments, and the rate of these processes controls the bioavailability of
nitrogen. Nitrification occurs very rapidly (on the order of 2 mmol NH4/min) and dissolved
oxygen is quickly consumed with approximately 4.3 mg O2 per mg of ammonia required
(Vymazal 2007). Based on hydraulic loading and influent nitrogen concentrations it does not
appear that the system exceeds the rate of nitrification. Denitrication rates are limited with
respect to easily degradable sources of organic carbon in the wetland (Kozub and Liehr, 1999).
Denitrification rates in the field vary widely and are affected by the amount of labile carbon,
nitrate concentration and temperature. Sirivedhin and Gray (2006) showed that denitrification
rates varied from 0.0021 to 0.8100 kgN2O/kg sediment per day as temperature was increased
from 4-25 C.
Page 4
Table 2: Laboratory analytical methods and expected performance data for each analyte.
Analyte
Merriam et al. (1996)
Method Detection Limit
(mg/L)
0.07 mg N/l
Particulate Nitrogen
EPA 440.0
15 ug N/l
Ammonia
EPA 350.1
5 ug N/L
Nitrate
EPA 353.2 (colormetric)
5 ug N/l
Nitrate
EPA 300.1
(ion chromatography)
EPA 353.2 (colormetric)
5 ug N/l
Variable
Total Phosphorus
EPA 300.1
(ion chromatography)
EPA 365.3
20 ug P/l
Total Dissolved Phosphorus
Ortho-Phosphate
EPA 365.3
EPA 365.1
20 ug P/l
5 ug P/l
Sulfate
Total Suspended Solids
Total Petroleum Hydrocarbons – Diesel
Range Organics
Metals (zinc)
EPA 300.1
SM2540D
EPA 8015.ME DRO
0.1 ug S/l
4 mg/l
25 ug/l
EPA 6010
0.014 mg/l
Total Dissolved Nitrogen
Nitrite
Nitrite
Analytical Method
3 ug N/l
Phosphorus
Phosphorous is comprised of particulate, dissolved organic, and soluble reactive forms.
Phosphorous removal mechanisms include plant uptake, sedimentation and filtration of
particulate phosphorous, and sorption to gravel or soil substrates (USEPA 2000, Shenker et al
2005). Co-precipitation with calcium, iron, or aluminum creates filterable particulates. The
phosphate in runoff likely forms a highly insoluble phosphate complex, specifically an alkali
iron-aluminum phosphate mineral ((K,Na)3(Al,Fe3+)5(PO4)2(HPO4)6•18H2O) (Anthony et al.
1995). Waters in New Hampshire, both runoff and groundwater, are generally rich in iron.
Free orthophosphate is the only form of phosphorus that appears to be utilized directly by
algae and macrophytes and is the nutrient of most concern in many surface waters. Iron (III)
depletion coincides with the maximum sediment oxygen demand (Bartoli et al., 2001).
Page 5
Influent
● Forebay 1
● Forebay 2
¤ GW1a
¤ GW2a
¤ GW2b
Effluent
Figure 2: (a, left) Sampling locations at the UNHSC field facility. Samples were collected from the influent, effluent,
and 3 locations within the wetland; (b, right) Locations of water quality measurements within the subsurface gravel
wetland, Durham, NH
SAMPLING METHODS
Field sampling methods are summarized here, and are discussed in more detail in the Project
Quality Assurance Project Plan (QAPP). The study site is located at the UNHSC field facility.
The UNHSC is located on the perimeter of a 3.6 hectare (900 parking space) commuter
parking lot at the University of New Hampshire (UNH) in Durham, NH. The parking lot,
installed in 1996, is standard dense mix asphalt, completely curbed, and is used to near
capacity throughout the academic year. Activity is a combination of passenger vehicles and
routine bus traffic. The runoff time of concentration for the lot is 22 minutes, with slopes
ranging from 1.5-2.5%. The area is subject to frequent plowing, salting, and sanding during
the winter months (typically November through April). Literature review comparisons of site
runoff water quality indicate that contaminant concentrations are above or equal to national
norms for parking lot runoff. The climatology of the area is characterized as coastal, cool
temperate. Average annual precipitation is 122 cm uniformly distributed throughout the year,
with average monthly precipitation of 10.2 cm +/-1.3 cm. The mean annual air temperature is
9°C, with the average low in January at -9°C, and the average high in July at 28°C.
Page 6
Sample Collection and Analysis.
Water samples were collected during eight storm events between October 2007 and September
2009. Half of the monitoring occurred during the coldest months (November-April) and the
remaining half during the warmest (May-October). An overview of the gravel wetland is
shown in Figure 2, and the sampling locations are shown in Figure 3. Flow is conveyed to and
from the treatment system by 12” diameter HDPE pipe with smooth wall. Influent samples
were collected at a concrete distribution box, located at the head of the research facility (GW
influent). The distribution box is located 180 feet from the gravel wetland forebay. Effluent
samples were collected at a 10-inch diameter outlet pipe which drains the unit approximately
115 feet downstream of the treatment unit (GW effluent). Typically all influent and effluent
samples are flow weighted composites. In-system stormwater samples were collected from
Forebay 2, a vegetated pretreatment forebay, and from a 1 inch diameter stainless steel well
driven into the gravel in the second cell (GW2b).
Samples during one storm were also collected from a well in the first gravel cell (GW1a). Insystem samples are all grab samples from wells and thus not flow-weighted due to the
limitations of monitoring flow rate through the subsurface. The sample locations were selected
to evaluate overall influent and effluent characteristics, and to identify removal processes
occurring in the aerated forebay, and the subsurface cells.
The samples were collected with ISCO 6712 automated samplers equipped with a stainless
steel strainer, 3/8 inch ID vinyl collection tubing and 24 discrete 1 liter low density
polypropylene (LDPE) sample bags. The automated sampling units were weather proof or
sheltered. A more detailed discussion of sampling methods can be found in the QAPP.
Precipitation (measured using an ISCO Model 674 tipping bucket rain sensor/gauge) and flow
measurement records were maintained for all events that occurred during the study period. If
an event failed to meet the criteria for a qualified sampling event, the collected influent and
effluent samples were not analyzed. Only data from qualified sampling events were used in
the calculation of pollutant removal efficiencies.
One liter disposable LDPE sample bags were used to assure clean, non-contaminated sample
containers. Prior to a sampling event, each bag was labeled with a unique, waterproof,
adhesive bar code that corresponded with a field identification number containing information
relating to the stormwater treatment unit, the sample number (1-24) and the date of sampling.
Records were kept that correlated sample number with sample time, date, flow, and other real
time water quality parameters. A chain-of-custody record accompanied each sample to track
all personnel handling and transporting each sample throughout the sampling process.
All samples, whether sent for analysis or frozen for later use, were processed within 24 hours
from the time of taking the sample and always maintained at temperatures of at most 4° C.
After each rainfall and sampling event, the sampler carousel was replaced with a new set of 24
bottles which were again labeled with unique identifying bar codes and the sampling process
began again.
Page 7
Flow weighted composite samples were collected for influent and effluent three times during
the warm months and three times during the cold months. Representative discrete samples
were selected based on the hydrograph and immediately composited using a Teflon cone
splitter. Composite samples were removed from the cone splitter, heat sealed, labeled and
delivered for laboratory analysis. Discrete samples were collected during two storms from
within the treatment cells. Discrete grab samples were selected after review of the storm event
hydrograph to represent early, mid- and late storm conditions. Real time water quality
parameters (pH, DO, temp, SC, flow) were also monitored continuously at the influent and
effluent locations, and in the gravel cells during selected storms. All samples were analyzed
for phosphorus and nitrogen compounds, total suspended solids, and cations/anions as listed in
Table 2. Analyses were performed by the Water Quality Analysis Laboratory at the
University of New Hampshire. Additional routine analyses collected under the UNHSC
general monitoring program were analyzed by Resource Laboratories Inc., in Portsmouth,
New Hampshire.
Storm event criteria and sampling methodologies followed guidance from the NPDES Storm
Water Sampling Guidance document (EPA 833-B-92-001) and Urban Stormwater BMP
Performance Monitoring (USEPA, 2002). Event criteria for inclusion in the study required
that: the depth of the storm must be greater than 0.1 inch accumulation and that the storm
must be preceded by at least 72 hours of dry weather. All measurements, sampling procedures,
and analyses followed a project specific QAPP, approved for NEIWPCC by the EPA.
Procedures other than those specifically described in the QAPP for field monitoring and
sample collection are outlined in specific manufacturers’ operation and maintenance manuals
(ISCO, YSI). These include procedures for field instrument calibration, and field equipment
cleaning, operation, and maintenance.
Water Quality Measurements
Water quality measurements for temperature (temp, °C), specific conductivity (SC, uS/cm),
dissolved oxygen (DO, mg/L) and pH were taken in the first cell of the gravel wetland (GW1a,
GW1b) and two wells in the second cell (GW2a, GW2b) from April 2007 August 18th, 2008
through October, 2008. Forebay sampling locations 1 and 2 were located in the vegetated
pretreatment sedimentation forebay. GW1a, Gw1b, GW2a and GW2b were 1 inch diameter
stainless steel pipes, hand-driven such that the screen intersected a gravel portion of the
wetland.
Page 8
ƒ
Flow
ƒ
Gravel Wetland Effluent DO
Figure:3: Oxygen concentrations in the subsurface gravel wetland effluent
Water quality measurements in the forebay were collected by immersing the probe in standing
water. Measurements were collected from the wells using a low-flow peristaltic pump to
purge water from the wells. The tubing was placed directly down into the well and the pump
was turned on. Water was examined for odor and clarity, with the results noted in the field
book. The water was pumped into a 10 liter Nalgene carboy to about half way, shaken to
completely rinse the carboy and then discarded. This process was repeated two more times
and then the water was allowed to fill the carboy to the point of overflowing. The YSI 556
probe module was placed in the carboy, ensuring complete immersion in the sample. The
readings were allowed to stabilize for at least ten minutes, at which time the data was recorded
in the field book. The probe module was rinsed with de-ionized water in between each
sampling site.
Summary and Performance Statistics
Event mean concentrations (EMC) were calculated for all analyzed storms by flow-weighted
composite samples. EMC is a parameter used to represent the flow proportional average
concentration of a given parameter during a storm event. It is defined as the total constituent
mass divided by the total flow volume. Performance measurements are best viewed in the
context of a range of metrics. Individual metrics viewed outside the larger context can be
misleading. Exploratory data analyses and statistics used here include removal efficiencies,
time series analyses, and regressions with analyses of variance (Burton and Pitt 2001, Pitt
2002, Geosyntec 2009). Removal efficiency is a commonly used metric in stormwater studies,
but the results should be interpreted with caution and in the context of influent concentration.
Percent removals are widely used because of their established regulatory presence. However,
small differences in low concentrations can lead to large differences in percent removals.
Averages are appropriately used to represent the central tendency of normally distributed data
sets. However, stormwater data sets are commonly non-normally distributed with a high
variability, and are difficult to describe using simple statistics. The median, which is exceeded
half the time, is often more helpful than the mean which has no set probability associated with
it.
Page 9
HISTORY AND MAINTENANCE
Aircraft Deicer Experiment
In 2007, the UNHSC subsurface gravel wetland was used to evaluate the ability of engineered
subsurface horizontal flow wetlands to remove propylene glycol (PG) (used as an aircraft
deicer and often co-mingled with airport stormwater runoff) from stormwater. This
experiment impacted the subsurface gravel wetland treatment performance during its
performance. Afterwards, during the period of time the nutrient cycling experiment was being
conducted, the effects of the propylene glycol study were still exhibited. However the
propylene glycol study did lead to some important discoveries regarding forebay design,
functionality, and maintenance.
Propylene glycol is readily biodegradable and produces a very high BOD, which can result in
impairment of receiving waters. After two months into the propylene glycol study, the
wetland effluent developed a brown biofloc generated from the abundant microbial activity in
the subsurface which persisted well after the experiment was concluded. The impacts upon
treatment performance are evident in several instances:
•
First, the biofloc development coincided with a reduction in TSS removal and an
increase in vegetative growth in the forebay. The TSS performance did recover
following summer maintenance (discussed below).
•
In the spring and summer of 2007 the PG stimulated excessive biomass production in
the form of plant growth in the forebay. During the fall, following the 2007 growing
season, a large release of nitrate and a drop in removal efficiency for TSS and nitrogen
was observed. The nitrate flux during this period would presumably be due to
senescence of the vegetation and the decomposition of organic matter. The nitrate
performance improved by the third summer following the PG experiment indicating
potential recovery. However recovery of winter performance has yet to be observed.
•
Third, a short term decline, over the first summer after the PG experiment, was
observed for phosphorous removal.
As a result, the performance data used in this report is from the events preceding the PG
experiment, prior to January 2007. The nutrient cycling mechanism data was collected from
the period from October 2007 through fall of 2009.
System Maintenance
In the summer of 2007, prior to the beginning of the nutrient study and several months after
the conclusion of the ADF experiment, the hydraulic outlet control was removed, allowing
high flows to move through the wetland for several storms and flushing the biofloc. The TSS
and TP removal performance improved after the flushing was performed, but nitrate removal
did not.
Forebay Maintenance
The forebay to the gravel wetland and other stormwater systems is designed to remove larger
sediments which settle onto the forebay floor. This provides pretreatment to runoff water,
Page 10
Table 3: Summary Performance Statistics for Gravel Wetland for September 2004- January 2007
Total Suspended Solids
(mg/l)
Statistic
Total Phosphorus
(mg/l)
Summers 1-3
Influent
Effluent
Winters 1-3
Influent
Effluent
Mean
59.6
0.77
58.7
0.54
60.37
0.96
Median
45.3
0
40.9
0.13
47.41
0
N
18
18
8
8
10
10
SD
57.9
1.67
57.3
1.13
61.4
2.03
Cv
0.97
2.14
0.98
2.08
1.02
2.12
Mean % RE
98.8
99.1
98.5
Median % RE
99
99.7
99
Years 1-3
Influent
Effluent
Summers 1-3
Influent
Effluent
Winters 1-3
Influent
Effluent
Statistic
Mean
0.1
0.04
0.13
0.06
0.09
0.02
Median
0.09
0.03
0.09
0.06
0.08
0.02
N
12
12
7
7
5
5
SD
0.09
0.03
0.1
0.03
0.07
0.01
Cv
0.74
0.77
0.73
0.48
0.79
0.47
Mean % RE
49.5
33.3
72.1
Median % RE
56.4
52.8
70.1
Years 1-3
Influent
Effluent
Summers 1-3
Influent
Effluent
Winters 1-3
Influent
Effluent
Statistic
Nitrate
(mg/l)
Years 1-3
Influent
Effluent
Mean
0.59
0.16
0.76
0.08
0.39
0.26
Median
0.37
0.03
0.37
0
0.36
0.1
N
21
21
11
11
10
10
SD
0.89
0.25
1.20
0.14
0.26
0.32
Cv
1.51
1.51
1.57
1.69
0.65
1.24
Mean % RE
Median % RE
38.9
79.5
-5.7
98
99
29.4
Page 11
which then moves into the wetland cells. Over time, as the forebay accumulates sediment and
plant debris, there is a risk that material may become resuspended during storm events, and
move into the gravel substrate. Prior to this nutrient study, the UNHSC had collected data
only from the influent and effluent sampling locations so that total removal could be assessed,
but; yellow shaded area indicates period of ADF (deicer) experiment; and vertical red lines
indicate period of forebay maintenance. No information was provided on nutrient removal
mechanisms exhibited in each of the different subsurface gravel wetland system components.
Initial discrete sampling in the fall of 2007 identified an increase in nitrogen, phosphorus, TSS
and other contaminants in the forebay compared to influent levels (Appendices B and C). At
this point the forebay supported a thick growth of plants, primarily cattails (typha latifolia),
and a layer of decayed plant debris. It appears that this material was becoming resuspended
during storm events, and serving as an in-system source of nutrients and solids. During the
summer of 2008, the plant material was removed and the forebay sediment was seeded with
mixed grass species. Plant removal was performed manually and included substantial root
mass. In September of 2008, sampling resumed, but forebay samples continued to contain
elevated concentrations of nutrients and TSS. In the summer of 2009, additional plant
biomass was removed, this time by cutting, to minimize soil disturbances, followed by reestablishment of mixed species grasses to stabilize the forebay. After each period of
maintenance, the system was taken off line to allow the new vegetation time to establish prior
to exposure to erosive velocities. These results have identified forebay maintenance issues
related to maintenance consisting of the harvesting of plant biomass in the system, and of
improved forebay design.
RESULTS
Long Term and Seasonal Results
The UNHSC has been collecting total phosphorus, nitrate, and total suspended solids data
since 2004. Figure 4 through Figure 6 show influent and effluent concentrations, and removal
efficiencies for these compounds from September 2004 through July 2009. However,
performance data discussed below is limited to September 2004 through January 2007, just
prior to the start of the propylene glycol deicer experiment. Appendix A--- and Appendix B--contain the nitrogen and phosphorus concentrations collected during the detailed nutrient
study from 2007-2009.
Page 12
influent TSS
gw effluent TSS
ADF
Maintenance
GW TSS Removal Efficiency
TSS EMC (mg/L)
450
400
350
300
250
200
150
100
50
0
100%
50%
0%
‐50%
‐100%
‐150%
‐200%
‐250%
‐300%
Figure 4: TSS performance for the subsurface gravel wetland September 2004 –July 2009
Nitrogen
The long term UNHSC monitoring data shows annual median nitrate removals of >95% for
the first three years. Figure 5 shows the influent and effluent concentration, and percent
removal for nitrate over the period sampled. Table 3 provides summary performance metrics
and statistics for nitrate, total phosphorus, and total suspended solids. The full statistical
analysis summarized in Table 3 is included in Appendix D--- through Appendix F---. Removal
rates are consistently above 95% for the first year of operation, and remain high during the
warm season (summer). In the cold season (winter) of 2005-2006 the removal dropped to
14% driven largely by one event with very high negative removal. In subsequent years the
winter removal efficiencies remained low.
This seasonal component is influenced by the plant growth cycle and by temperature changes.
In the spring, as plants start to grow, nutrients are removed from the sediment and stored in
plant material. In the fall, with the onset of colder weather, plants senesce and nutrients may
be translocated back into root systems and re-released to the environment. Eventually, the
wetland plant biomass accumulates on the surface of the wetland as plant litter. Decaying
litter then releases nutrients into stormwater when it passes through the system in the winter.
This seasonal plant cycle has a double impact; increased removal via uptake in the spring and
Page 13
1.5
influent NO3
gw effluent NO3
ADF Maintenance
NO3 EMC (mg/L)
1.3
1.1
0.9
0.7
0.5
0.3
GW NO3 Removal Efficiency
0.1
‐0.1
100%
50%
0%
‐50%
‐100%
‐150%
‐200%
‐250%
‐300%
‐350%
Figure 5: Nitrate performance for the subsurface gravel wetland September 2004 –July 2009; yellow shaded area
indicates period of ADF (deicer) experiment; vertical red lines indicate period of forebay maintenance.
summer, and re-release in the fall and winter. Cutting and removing the above ground plant
material effectively removes some of the nutrient mass from the system.
The amount of nutrient removed by harvesting depends upon the type and volume of plant
material, but a typical freshwater wetland carries 3-29 g of Nitrogen/m2 in above ground
biomass (Mitsch and Gosselink, 2000). The UNH wetland has a surface area of 18m2, which
would yield approximately 50-500 grams of nitrogen annually.
Another factor governing the seasonal removal of nitrogen is the seasonal change in water
temperature, and associated increase in dissolved oxygen concentrations. The mean
temperature of runoff at the UNHSC site is 59ºF during the warm season and 45ºF in the cold
season. Dissolved oxygen is closely correlated with temperature, with a mean of 7.8mg/l in
the warm season, and 10.2mg/l in the cold season. The increased oxygen entering the system,
combined with lower microbial activity will reduce the ability of the system to develop the
anaerobic conditions required for nitrate conversion.
Page 14
1.50
influent TP
gw effluent TP
ADF
Maintenance
TP EMC (mg/L)
1.30
1.10
0.90
0.70
0.50
0.30
0.10
GW TP Removal Efficiency
‐0.10
100%
0%
‐100%
‐200%
‐300%
‐400%
‐500%
‐600%
‐700%
Figure 6: Total phosphorous performance for the subsurface gravel wetland September 2004 –July 2009; yellow
shaded area indicates period of ADF (deicer) experiment; vertical red lines indicate period of forebay maintenance.
Phosphorus
Phosphorus is removed from stormwater by several mechanisms including uptake in plants,
sorption to soil media, precipitation, and microbial uptake. Plants remove orthophosphate and
release it as organic phosphorus in deposited plant litter, where it is transformed back to
orthophosphate as the material decays. Sorption binds phosphorus to subsurface particles, a
process that is regulated by particle size, sediment composition, and water chemistry including
phosphorus concentration and ionic strength. Calcium, iron, and aluminum can form
precipitates that store phosphorus in insoluble forms. Precipitation is governed by availability
of co-precipitating compounds (Fe, Al, dissolved oxygen), and by water chemistry,
particularly pH. Microbes uptake and store phosphorus, but release it again as they die. None
of these processes actually remove phosphorus from the wetland; they transform and/or store
phosphorus in materials and compounds which may then re-release the phosphorus when
conditions change.
Figure 6 shows the influent and effluent phosphorus concentrations and percent removal.
Median annual phosphorus removal in the UNH subsurface gravel wetland has been 55% with
summer removal rate >50% and a winter removal rate >70% for the first three years of
Page 15
Figure 7: Total phosphorus concentration in effluent as a function of pH
monitoring with no identifiable long term trend. We hypothesize that improved winter
performance is due to the observed seasonal increase in dissolved oxygen and a resultant
increase in the formation of insoluble phosphate complexes with iron and aluminum. We have
not yet measured the concentration of the alkali iron-aluminum phosphate mineral
((K,Na)3(Al,Fe3+)5(PO4)2(HPO4)6•18H2O) within the gravel wetland. This would be the next
logical step in evaluation of phosphorus removal mechanisms. Release of phosphorus during
the winter when high chloride levels are present suggests that sorption capacity is not
regulated by the observed variations in ionic strength. Five water quality parameters were
measured in the influent and effluent of each system: temperature, pH, dissolved oxygen,
specific conductivity, and turbidity. Review of the data shows no strong correlation between
phosphorus effluent concentrations and any of the measured parameters, except temperature
and pH. pH is expected to influence precipitation; as pH increases phosphorus tends to form
precipitates with calcium and magnesium carbonates, while under acidic conditions
phosphorus binds to aluminum and iron oxides. The highest effluent phosphorus
concentrations occur when the pH is between 5.5 and 6 (Figure 7).
Nutrient Cycling and Removal Mechanisms
Nitrogen
The nitrogen concentrations during the early to late portions of a storm sampled in November
2007 are presented in Figure 8. Influent from the parking lot contains high concentrations of
total nitrogen, which is the sum of particulate and dissolved nitrogen, and encompasses both
organic and inorganic forms. Ammonia, nitrite and nitrate are also present, although nitrite
concentrations are low throughout the system, reflecting the rapid transformation of nitrite to
nitrate.
Page 16
11/26/2007 - early storm (ug/l)
TN
PN
TDN
1000
800
600
400
200
0
DON
NH4
NO2
NO3
..
Inf luent
Forebay
GW1
GW2
Ef f luent
GW2
Ef f luent
GW2
Ef f luent
GW2
Ef f luent
11/26/2007 - mid storm
1000
800
600
400
200
0
..
Inf luent
Forebay
GW1
11/26/2007 - mid storm
1000
800
600
400
200
0
..
..
Inf luent
Forebay
GW1
11/26/2007 - late storm
1000
800
600
400
200
0
..
..
Inf luent
Forebay
GW1
Figure 8: Examination of nitrogen speciation (ug N/l) as it travels through gravel wetland throughout storm. GW1
was not sampled during this event.
Page 17
Figure 9: Mass loading for DRO, Zn, NO3, TSS as a function of normalized storm volume for two storms: (a) a large
60 mm rainfall over 1685 minutes; (b) a smaller 15 mm storm depth over 490 minute. DRO=diesel range organics, Zn=
zinc, NO3= nitrate, TSS= total suspended solids
Water moves from the forebay, onto the surface of the 1st wetland cell, then into the
subsurface gravel. Samples collected from GW Cell 2 and from the effluent during the early
part of the storm show a significant reduction in total nitrogen and very low levels of nitrate,
suggesting that the nitrate has been successfully converted to a nitrogen gas. Samples collected
later in the storm show lower nitrate removal. The discrete storm sampled in October 2009,
shows a similar trend. This progression supports our hypothesis that the aerobic/anaerobic
treatment system is supporting the full nitrogen transformation process, although nitrogen
removal is most efficient early in the storm when the first flush storm washoff mixes with the
oxygen-depleted water already present in the system. Oxygen measurements within the cells
both during the storm and between events confirm that anaerobic conditions are present in the
substrate, with many measurements below 2 mg/l. The complete nitrogen concentrations
measured during this study are shown in the graphs in Appendix B.
As discussed above, transformation of organic nitrogen to gaseous N2O or N2 requires two
phases: aerobic transformation from organic to nitrate forms, and anaerobic reduction from
nitrate to nitrogen gas. The process requires an aerobic environment followed by an anaerobic
setting. Stormwater runoff entering the gravel wetland cells are aerobic, but the subsurface
cells are designed to remain saturated, and isolated from surface oxygen. Microbial
respiration uses available oxygen, until the stored water is depleted, and anaerobic conditions
develop. Figure 3 shows the oxygen concentrations in the gravel wetland during a typical
storm event as measured at the outlet. Initially, as flow increases, the dissolved oxygen in the
effluent drops rapidly as the resident water moves out of the system. As the storm progresses,
oxygen levels increase as aerated stormwater fills the wetland cells. The resident volume in
the subsurface gravel is 22m3 (770 ft3) which is the volume of a 6mm (0.25 in) rainfall on the
Page 18
1-acre watershed. Some mixing of the residual and influent water occurs, providing an
interface between the fresh, nitrogen laden influent, and the oxygen depleted resident water.
This is important with respect to system sizing.
Figure 9 shows the first flush tendency for a range of contaminants (Roseen et al 2006). The
entire load of nitrate carried in the stormwater from the two events depicted was washed off by
the first 6mm of rain. That study examined runoff characteristics and found that the majority
of nitrate is concentrated in the early storm volume. This “first flush” coincides with the
optimal conditions for nitrogen removal, early in the storm when the aerobic/anaerobic
interface is present in the subsurface cells. Figure 9 shows nitrate wash-off for two distinct
storms at the UNHSC site. Figure 9(a) is for a large 60 mm (2.3 in) rainfall, where the total
mass of nitrate all washed off in the first 1% of the storm depth, a volume approximately equal
to 22 m3 (770 ft3). Figure 9(b) shows a smaller 15 mm storm depth with 75% of the mass of
nitrate being washed off during the first 20% of the storm, a volume approximately equal to 98
m3 (3300 ft3). The total volume for this system is 114 m3 (water quality volume =92 m3 plus
resident subsurface volume =22 m3) (4070 ft3). Each storm has different loading
characteristics, but nitrate mass is consistently highest early in the storm and corresponds to a
runoff volume less than the total system volume. The high nitrogen removal achieved by
gravel wetland systems reflects the optimal removal conditions occurring when most of the
nitrogen mass is in the system.
The excessive biomass production from the propylene glycol (PG) experiment is believed to
have resulted in a short-circuiting of the subsurface gravel wetland system processes. As
mentioned previously, the gravel wetland process relies on a sequence of chemical
transformations occurring first in an aerobic forebay (nitrification) where any particulate
organic nitrogen (typically an OM-NH4) or ammonium would be converted to nitrate. This
process is followed by the transformation or uptake of nitrate (denitrification). The excessive
biomass and high BOD load from the previous PG experiments resulted in a forebay that
became anaerobic, interrupting the nitrification process yielding high concentrations of
particulate organic nitrogen. The significance of this finding extends broadly to subsurface
gravel wetlands and is two part: 1) careful forebay design is needed to prevent anaerobic
conditions from developing, 2) maintenance and removal of the biomass is essential to prevent
rerelease of nitrogen and development of anaerobic conditions from accumulation of decaying
organic matter.
In particular, the recognition of need for improved forebay design is an important outcome
from this study. Forebay design should seek to provide treatment by settling, and to maintain
aerobic conditions with minimal maintenance. Shallow forebay design, while good for
settling, requires potentially higher maintenance than other designs in order to prevent a thick
matt of cattails and the subsequent development of anaerobic conditions. Improved designs
would include a deeper pool of water in excess of one meter (three feet), a deep sump catch
basin or proprietary treatment device for removal of solids, or other designs that could
maintain aerobic conditions for the long term.
Page 19
100
11/26/07 ‐ early storm
TP (ug P/L)
50
TDP (ug P/L)
0
PO4 (ug P/L)
Influent
100
Forebay
GW 2b
Effluent
11/26/07 ‐ mid storm
TP (ug P/L)
50
TDP (ug P/L)
0
PO4 (ug P/L)
Influent
50
Forebay
GW 2b
Effluent
11/26/07 ‐ mid storm
TP (ug P/L)
TDP (ug P/L)
0
PO4 (ug P/L)
Influent
100
Forebay
GW 2b
Effluent
11/26/07 ‐ late storm
TP (ug P/L)
50
TDP (ug P/L)
0
PO4 (ug P/L)
Influent
Forebay
GW 2b
Effluent
Figure 10: Examination of phosphorous speciation (ug/l) as it travels through gravel wetland throughout storm
Page 20
Phosphorous
The phosphorus concentrations in the gravel wetland in the November 2007 storm discussed
previously are displayed in Figure 10. Total phosphorus, total dissolved phosphorus, and
orthophosphate are all present in the influent. The orthophosphate is removed in the forebay
in this storm, but this behavior is not consistent; in the September 2009 storm for instance,
orthophosphate is present in samples collected throughout the system. See Appendix B for
more storm data. Concentrations of total and dissolved phosphorus increase in the forebay,
are absent in the gravel substrate (GW2b), but are present again in the effluent. Total and
dissolved phosphorus are consistently present in the influent at this site, while orthophosphate
is generally present at lower concentrations, or not detected. The increases in concentration in
the forebay noted for nitrogen is also present for phosphorus in many, but not all storms
sampled during this study. The sources of phosphorus in the forebay are similar to the sources
of nitrogen; decaying plant material, and re-suspended sediments. It is interesting to note,
however, that increased forebay concentrations of phosphorus and nitrogen are not always
consistent; on September 6, 2008 concentrations of total nitrogen are higher in the forebay
than in the influent, while concentrations of total phosphorus are lower in the forebay, The
amount of nutrient released in the forebay may be a function of several factors, including
storm characteristics, seasonality, and recent maintenance or disturbance.
Real Time Water Quality Parameters
Real-time water quality measurements taken in the UNHSC subsurface gravel wetland from
April 2007 to October 2008 are shown in Figure 11. Temperature ranges from 5-7 ºC, with
higher temperatures in the warmer months. Conductivity ranges from 0.2-6 us/cm3, with
higher concentrations in the winter, when road salt is applied to the parking lot. pH ranges
from 5-7, and is generally lower in the wetland cells. Dissolved oxygen ranges from 0.614.3mg/l in the forebay, but does not rise above 3mg/l in the subsurface cells. These
measurements were collected once a week, during non-rain periods. These data confirm the
presence of low oxygen reducing conditions in the gravel substrate but also show that it varies
inter-storm.
Regression Analyses
Performance analyses by regression of EMCs were performed for the three contaminants of
interest and are presented in Appendix C. Analyses of variances for the model parameters
examined the significance of performance. It can be observed that TSS removal is strong for
all three seasons with a slope indicating >99% treatment for all three seasons. Although fit
parameters are weak for TSS, indicating a non-linear function, this is primarily due to the
presence of values falling beneath the detection limit of the equipment, which create an
artificial cluster at 0. The same is true for NO3 and TP. Larger seasonal data sets will be
required to resolve significance.
Page 21
20
Temp (C)
15
10
5
0
-5
Dissolved Oxygen
% saturation
16
12
8
4
0
8
pH
7
6
5
Conductivity (us/cm3)
4
7
6
5
4
3
2
1
0
GW_in
Foreb 1
Foreb 2
GW_1a
GW_1b
GW_2a
GW_2b
GW_eff
4/6/2007
4/9/2007
4/27/2007
4/30/2007
5/2/2007
5/4/2007
5/10/2007
5/14/2007
10/17/2007
10/24/2007
10/31/2007
11/7/2007
11/14/2007
11/20/2007
11/28/2007
12/4/2007
12/12/2007
12/18/2007
2/15/2008
2/25/2008
3/24/2008
4/4/2008
4/10/2008
4/16/2008
4/23/2008
5/1/2008
5/15/2008
5/22/2008
5/29/2008
8/18/2008
8/27/2008
9/10/2008
9/18/2008
9/23/2008
10/13/2008
10/27/2008
Figure 11: Real-time water quality measurements in the gravel wetland, collected weekly April 2007-October 2008
Page 22
CONCLUSIONS
This study explored the performance of an engineered subsurface horizontal flow gravel
wetland for removing TSS and phosphorus and nitrogen nutrient species from stormwater.
The results of this study, and long term monitoring at the UNHSC indicate that subsurface
gravel wetlands are highly effective at removing nitrogen, and moderately effective at
removing phosphorous, with seasonal variations apparent for both nutrients, and the
requirement of maintenance for optimal nitrate removal.
Because of the disruption of the system caused by the propylene glycol study, additional
monitoring at other sites will be needed to evaluate long-term functionality beyond the three
year period observed at the UNHSC. This information will be available as soon as 2010 from
other gravel wetland sites that have been constructed and are currently being monitored for
permit compliance. One such site is a commercial development named Greenland Meadows,
in Greenland, New Hampshire which is currently on its the second winter (2009-2010). The
UNHSC is responsible for monitoring and annual reporting to New Hampshire Department of
Environmental Services for this site from 2007-2013.
Specifically, this study found that:
•
The “first flush” coincides with the optimal conditions for nitrogen removal, early in the
storm when anaerobic conditions exist in the subsurface. Studies at the UNHSC have
shown that the majority of the mass of nitrate wash-off is within a first flush volume nearly
equal to 22 m3 (770 ft3), the volume of resident water within the gravel wetland. This has
important implications with respect to system sizing.
•
The forebay to the gravel wetland and probably all stormwater systems may become a
source of contamination as the system ages. Sediments and plant debris stored in the
forebay may be re-suspended and released in subsequent storms. Routine maintenance is
an important component in maintaining performance. The maintenance interval for which
a decline in performance can be observed is three growing seasons, or the end of the
summer of the third year.
•
The median removal efficiency for nitrate measured at the UNHSC since 2004 is 80%.
Removal is higher in the summer than in the winter probably due to the combined effect of
reduced uptake and microbial activity in the winter, coupled with organic nitrogen released
from plant detritus in the fall and winter. The long term monitoring shows that the annual
median nitrate removal rate is >95% for the first three years, prior to system maintenance.
Summertime performance > 98% and reduced winter performance of >30%.
•
Nitrate removal declined after three growing seasons with observed increase in the forebay
indicating rerelease due to decomposition of organic matter.
•
Phosphorus removal may be effected by pH, with lower removal occurring when the
influent pH is between 5.5 and 6. Long term phosphorus data shows an annual median
removal rate of 55% for the first three growing seasons with increased performance in the
winter > 70%, likely related to increased dissolved oxygen concentrations in winter runoff
Page 23
and to influent pH. Summertime total phosphorus removal drops to >50%. A decline in
phosphorous removal related to maintenance has not been observed.
•
Forebay design is an important component of effective treatment. Forebay design should
provide treatment by settling, and maintain aerobic conditions with minimal maintenance.
Shallow forebay design, while good for settling, requires potentially higher maintenance
than other designs to prevent the development of a thick matt of cattails and anaerobic
conditions. Improved designs would include a deeper pool of water in excess of a meter, or
a deep sump catch basin or proprietary treatment device for removal of solids.
Page 24
REFERENCES
Avanzino, R. J., and Kennedy, V. C. (1993). "Long-Term Frozen Storage of Stream Water
Samples for Dissolved Orthophosphate, Nitrate Plus Nitrite, and Ammonia Analysis."
29(10), 3357-3362.
Anthony, J. W., Bideaux, R. A., Bladh, K. W., and Nichols, M. C. (1995). Handbook of
Mineralogy: Arsenates, Phosphates, Vanadates. IV., Mineral Data Publishing, Tucson,
Arizona.
Bartoli M., G. Castaldelli, D. Nizzoli, L. G. Gatti, P. Viaroli, (2001) Benthic fluxes of oxygen,
ammonium and nitrate and coupled-uncoupled denitrification rates within communities of
three different primary producer growth forms. In Faranda F.M., Guglielmo L., G. Spezie
(eds), Mediterranean Ecosystems: Structures and Processes. Springer Verlag,Milano, 29, 225–
233.
Burton, G. A. J., and Pitt., R. (2001). Stormwater Effects Handbook: A Tool Box for
Watershed Managers, Scientists, and Engineers, CRC Press, Inc., Boca Raton, FL.
García, J., Aguirre, P., Mujeriego, R., Huang, Y., Ortiz, L., Bayona, J.M., (2004). Initial
contaminant removal performance factors in horizontal flow reed beds used for treating urban
wastewater. Water Res. 38, 1669–1678.
GeoSyntec, Wright Water Engineers, I., USEPA, WERF, FHWA, and ASCE/EWRI. (2009).
"Urban Stormwater BMP Performance Monitoring."
Kozub, D.D., Liehr, S.K., (1999) Assessing denitrification rate limiting factors in a
constructed wetland receiving landfill leachate. Water Sci. Technol. 40, 75–82.
Mayo A.W. and Bigambo, T (2005) Nitrogen transformation in horizontal subsurface flow
constructed wetlands I: Model Development Phys. And Chemistry of the Earth 30,658-667
Mitsch, W. J. and J. G. Gosselink. Wetlands 3rd edition. John Wiley and Sons. 2000.
Pitt, R. (2002). "Statistical Considerations." Technology Assessment Protocol- Ecology's
(TAPE) by the State of Washington, Department of Ecology, WADOE, State of Washington,
Department of Ecology.
Roseen, R.M., T. P. Ballestero, J.J. Houle, P. Avelleneda, R.Wildey, and J.Briggs Storm
Water Low-Impact Development, Conventional Structural, and Manufactured Treatment
Strategies for Parking Lot Runoff Performance Evaluations Under Varied Mass Loading
Conditions.
Rousseau, D., Vanrolleghem, P., De Pauw, N. (2004). Model-based design of horizontal
subsurface flow constructed treatment wetlands: a review. Water Res. 38 (6), 1484–1493.diffusion limitations in biofilm
Page 25
Shenker, M., Seitelbach, S., Brand S., et al. (2005) Redox Reactions and phosphorous release
in re-flooded soils of an altered wetland. European Journal of Soil Science 56, 515-525.
Sirivedhin, T and Gray, K. A. (2006) Factors affecting denitrification rates in experimental
wetlands: Field and laboratory studies Ecological Engineering 26, 167–181
Tanner C. (2004) Nitrogen Removal Processes. In Wong M. H. (ed), Constructed Wetlands in
Developments of Ecosystems, Elsevier, United Kingdom, 8,331-346.
University of New Hampshire Stormwater Center (UNHSC), Roseen, R. M., Ballestero, T. P.,
and Houle, J. J. (2008). "UNHSC Subsurface Gravel Wetland Design Specifications."
University of New Hampshire Stormwater Center, Durham, NH.
http://www.unh.edu/erg/cstev/pubs_specs_info.htm#specs
University of New Hampshire Stormwater Center (UNHSC), Watts, A., R. Roseen, Houle, J.,
and T. Ballestero. (2007). "Quality Assurance Project Plan-for Investigation of Nutrient
Removal Mechanisms of a Constructed Gravel Wetland." University of New Hampshire
Stormwater Center, Durham, NH.
USEPA (2000). Constructed Wetlands Treatment of Municipal Wastewater. United States
Environmental Protection Agency, Office of Research and Development,
Cincinnati, OH.
USEPA, ASCE (2002). "Urban Stormwater BMP Performance Monitoring." EPA-821-B-02001, Washington DC.
Vymazal J. (2007) Removal of nutrients in various types of constructed wetlands Science of
the total Environment. 380,48-65
Page 26
Appendix A--- Nitrogen speciation within gravel wetland influent,
forebay, cells, and effluent
Nitrogen concentrations (ug N/l) measured in the UNHSC gravel wetland influent, forebay,
cells, and effluent. See for Figure 2 sampling locations. Nitrogen species sampled were: total
nitrogen (TN), particulate nitrogen (PN), total dissolved nitrogen (TDN), dissolved organic
nitrogen (DON), ammonium (NH4), nitrite (NO2) and nitrate (NO3).
11/26/2007 - early storm (ug/l)
1000
TN
800
PN
TDN
600
DON
400
NH4
200
NO2
NO3
0
Inf luent
Forebay
GW1
GW2
Ef f luent
% Removal
11/26/2007 - mid storm
TN
PN
1000
800
600
400
200
0
TDN
..
DON
NH4
NO2
Inf luent
GW1
Forebay
GW2
Ef f luent
% Removal
11/26/2007 - mid storm
1000
800
600
400
200
0
NO3
TN
PN
TDN
..
DON
NH4
..
NO2
Inf luent
Forebay
GW1
GW2
Ef f luent
% Removal
11/26/2007 - late storm
1000
800
600
400
200
0
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
% Removal
NO3
TN
PN
TDN
DON
NH4
NO2
NO3
Nitrogen concentrations in discrete storm samples collected in November, 2007. All results in
ug N/l.
Page 27
10/16/2007 - early storm
3500
3000
2500
2000
1500
1000
500
0
% Removal
TN
PN
TDN
DON
NH4
NO2
NO3
% Removal
TN
PN
TDN
DON
NH4
NO2
NO3
% Removal
TN
PN
TDN
DON
NH4
NO2
NO3
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
10/16/2007 - mid storm
3500
3000
2500
2000
1500
1000
500
0
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
10/16/2007 - mid storm
3500
3000
2500
2000
1500
1000
500
0
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
10/16/2007 - late storm
3500
3000
2500
2000
1500
1000
500
0
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
% Removal
TN
PN
TDN
DON
NH4
NO2
NO3
Nitrogen concentrations in discrete storm samples collected in October, 2007. All results in ug
N/l.
Page 28
1/11/2008
2500
2000
1500
1000
500
0
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
% Removal
3/5/2008
2500
2000
1500
1000
500
0
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
% Removal
9/6/2008
2500
2000
1500
1000
500
0
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
% Removal
9/26/2008
2500
2000
1500
1000
500
0
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
% Removal
3/26/2009
2500
2000
1500
1000
500
0
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
% Removal
9/27/2009
2500
2000
1500
1000
500
0
..
..
Inf luent
Forebay
GW1
GW2
Ef f luent
% Removal
TN
PN
TDN
DON
NH4
NO2
NO3
TN
PN
TDN
DON
NH4
NO2
NO3
TN
PN
TDN
DON
NH4
NO2
NO3
TN
PN
TDN
DON
NH4
NO2
NO3
TN
PN
TDN
DON
NH4
NO2
NO3
TN
PN
TDN
DON
NH4
NO2
NO3
Nitrogen concentrations in composite storm samples collected in January, 2008 through
September 2009. All results in ug N/l.
Page 29
Appendix B--- Phosphorus speciation in gravel wetland influent, forebay,
cells, and effluent
Phosphorus concentrations measured in the UNHSC gravel wetland influent, forebay, cells,
and effluent. See Figure 2 for sampling locations. Phosphorus species sampled were: total
phosphorus (TP), total dissolved phosphorus (TDP), and orthophosphate (PO4).
100
11/26/07 ‐ early storm
TP (ug P/L)
50
TDP (ug P/L)
0
PO4 (ug P/L)
Influent
100
Forebay
GW 2b
Effluent
11/26/07 ‐ mid storm
TP (ug P/L)
50
TDP (ug P/L)
0
PO4 (ug P/L)
Influent
50
Forebay
GW 2b
Effluent
11/26/07 ‐ mid storm
TP (ug P/L)
TDP (ug P/L)
0
PO4 (ug P/L)
Influent
100
Forebay
GW 2b
Effluent
11/26/07 ‐ late storm
TP (ug P/L)
50
TDP (ug P/L)
0
PO4 (ug P/L)
Influent
Forebay
GW 2b
Effluent
Phosphorus concentrations in discrete storm samples collected in November, 2007. All results
in ug P/l.
Page 30
150
10/16/2008 Early
100
TP (ug P/L)
50
TDP (ug P/L)
0
PO4 (ug P/L)
Influent Forebay
GW 1a
GW 2b
Effluent % removal
600
10/16/2008 Mid
400
TP (ug P/L)
200
TDP (ug P/L)
0
PO4 (ug P/L)
Influent Forebay
200
GW 1a
GW 2b
Effluent % removal
10/16/2008 Mid
TP (ug P/L)
100
TDP (ug P/L)
PO4 (ug P/L)
0
Influent Forebay
80
GW 1a
GW 2b
Effluent % removal
10/16/2008 Late
60
TP (ug P/L)
40
TDP (ug P/L)
20
PO4 (ug P/L)
0
Influent Forebay
GW 1a
GW 2b
Effluent % removal
Phosphorus concentrations in discrete storm samples collected in October, 2008. All results in
ug P/l.
Page 31
60
1/11/08
40
TP (ug P/L)
20
TDP (ug P/L)
0
PO4 (ug P/L)
Influent
Forebay
GW 2b
150
100
50
0
‐50
Effluent
% removal
3/5/08
TP (ug P/L)
TDP (ug P/L)
PO4 (ug P/L)
Influent
Forebay
GW 2b
500
Effluent
% removal
9/6/08
TP (ug P/L)
300
TDP (ug P/L)
100
PO4 (ug P/L)
‐100
Influent
Forebay
GW 2b
Effluent
% removal
100
9/26/08
TP (ug P/L)
50
TDP (ug P/L)
0
‐50
80
60
40
20
0
‐20
Influent Forebay
GW 1a
GW 2b
Effluent
% removal
PO4 (ug P/L)
3/26/09
TP (ug P/L)
TDP (ug P/L)
Influent
Forebay
80
GW 2b
Effluent
% removal
PO4 (ug P/L)
9/27/09
60
TP (ug P/L)
40
TDP (ug P/L)
20
PO4 (ug P/L)
0
Influent
Forebay
GW 2b
Effluent
% removal
Page 32
Appendix C--- Regression of EMC and Analysis of Variance
Parameter Estimates y= 0.14 + 0.01x
R2=0.156
Term
Estimate Std Error t Ratio Prob>|t|
Intercept
0.143
0.560
0.26
0.8019
M
0.011
0.007
1.69
0.1108
(a) Annual
Parameter Estimates y = -0.381 + 0.017x
R2=0.804
Term
Estimate Std Error t Ratio Prob>|t|
Intercept
-0.381844 0.312455
-1.22 0.2761
M
0.0174048 0.003841
4.53 0.0062*
(b) Summer
Parameter Estimates y= 0.51 + 0.007x
R2=0.048
Term
Estimate Std Error t Ratio Prob>|t|
Intercept
0.5194734 0.960931
0.54
0.6035
M
0.0073402 0.011457
0.64
0.5396
(c) Winter
Figure 12: TSS Regression of Event Mean Concentrations with ANOVA for Annual (a), Summer (b), and Winter (c) from September 2004- January
2007
Page 33
Parameter Estimates Y= 0.0589 + 0.305X
R2=0.0912
Term
Estimate Std Error t Ratio Prob>|t|
Intercept
0.0589938 0.11069
0.53 0.6010
M
0.3049289 0.233351
1.31 0.2087
(a) Annual
Parameter Estimates Y= -0.0917 + 0.454X
R2=0.661
Term
Estimate Std Error t Ratio Prob>|t|
Intercept
-0.091783 0.06019
-1.52 0.1711
M
0.4540437 0.12282
3.70 0.0077*
(b) Summer
Parameter Estimates Y= 0.190 + 0.174X R2=0.019
Term
Estimate Std Error t Ratio Prob>|t|
Intercept
M
0.1901924 0.201491
0.174757 0.438264
0.94
0.40
0.3728
0.7005
(c) Winter
Figure 13: NO3 Regression of Event Mean Concentrations with ANOVA for Annual, Summer, and Winter from September 2004- January 2007
Page 34
Parameter Estimates Y= 0.0349 + 0.030
R2=0.006
Term
Estimate Std Error t Ratio Prob>|t|
Intercept
0.0349243 0.014786
2.36 0.0425*
M
0.029713 0.12249
0.24 0.8138
(a) Annual
Parameter Estimates Y= 0.0480 + 0.0761X
R2=0.071
Term
Estimate Std Error t Ratio Prob>|t|
Intercept
0.0480074 0.017048
2.82 0.0480*
M
0.0761197 0.137172
0.55 0.6085
(b) Summer
Parameter Estimates Y= 0.023 - 0.072X
R2=0.522
Term
Estimate Std Error t Ratio Prob>|t|
Intercept
0.0231275 0.004679
M
-0.072921 0.040232
4.94 0.0159*
-1.81
0.1676
(c) Winter
Figure 14: TP Regression of Event Mean Concentrations with ANOVA for Annual, Summer, and Winter from September 2004- January 2007
Page 35
Appendix D--- TSS Distributions for the first three years of operation
September 2004-January 2007
Distributions
TSS (%R)
TSS EMC Effluent (mg/l)
101
7
100
6
99
5
98
4
97
3
96
2
95
1
94
0
93
-1
Quantiles
Quantiles
100.0% maximum
99.5%
97.5%
90.0%
75.0%
quartile
50.0%
median
25.0%
quartile
10.0%
2.5%
0.5%
0.0%
minimum
100
100
100
100
100
100
97.85
94.09
94
94
94
Moments
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
200
150
100
50
0
Quantiles
100.0% maximum
6.455
99.5%
6.455
97.5%
6.455
90.0%
3.6299
75.0%
quartile 0.71575
50.0%
median
0
25.0%
quartile
0
10.0%
0
2.5%
0
0.5%
0
0.0%
minimum
0
Moments
98.772222
2.0504822
0.4833033
99.791903
97.752541
18
18
1777.9
4.2044771
-1.621682
1.4481842
2.0759705
12
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
TSS EMC Influent (mg/l)
100.0% maximum 214.415
99.5%
214.415
97.5%
214.415
90.0%
190.253
75.0%
quartile 68.8703
50.0%
median 45.2865
25.0%
quartile 16.9928
10.0%
11.5848
2.5%
10.188
0.5%
10.188
0.0%
minimum
10.188
Moments
0.7771667
1.6662433
0.3927373
1.60577
-0.051437
18
18
13.989
2.7763667
2.790132
8.1951792
214.39974
12
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
59.6115
57.875891
13.641478
88.392504
30.830496
18
18
1073.007
3349.6188
1.8322929
2.9188719
97.088466
12
Page 36
May-November
Where(:Season 2 == "Warm")
Distributions
TSS (%R)
TSS EMC Effluent (mg/l)
100.5
100
99.5
99
98.5
98
97.5
97
96.5
96
3.5
3
150
2
1.5
100
1
0.5
50
0
-0.5
0
Quantiles
100
100
100
100
100
99.7
98.4
96.4
96.4
96.4
96.4
Moments
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
200
2.5
Quantiles
100.0% maximum
99.5%
97.5%
90.0%
75.0%
quartile
50.0%
median
25.0%
quartile
10.0%
2.5%
0.5%
0.0%
minimum
TSS EMC Influent (mg/l)
Quantiles
100.0% maximum
3.316
99.5%
3.316
97.5%
3.316
90.0%
3.316
75.0%
quartile 0.41075
50.0%
median
0.132
25.0%
quartile
0
10.0%
0
2.5%
0
0.5%
0
0.0%
minimum
0
Moments
99.125
1.2781124
0.451881
100.19353
98.056471
8
8
793
1.6335714
-1.672025
2.5804757
1.2893947
8
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
100.0% maximum 187.568
99.5%
187.568
97.5%
187.568
90.0%
187.568
75.0%
quartile 76.4968
50.0%
median 40.9425
25.0%
quartile 16.9453
10.0%
10.188
2.5%
10.188
0.5%
10.188
0.0%
minimum
10.188
Moments
0.545375
1.1336156
0.4007936
1.4931013
-0.402351
8
8
4.363
1.2850843
2.6919613
7.4088781
207.85983
8
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
58.663625
57.352332
20.277112
106.61137
10.715875
8
8
469.309
3289.29
1.9225999
4.2083107
97.764726
8
Page 37
November-April
Where(:Season 2 == "Cold")
Distributions
TSS (%R)
TSS EMC Effluent (mg/l)
101
7
100
6
99
5
98
4
97
3
96
2
95
1
94
0
93
-1
Quantiles
Quantiles
100.0% maximum
99.5%
97.5%
90.0%
quartile
75.0%
median
50.0%
quartile
25.0%
10.0%
2.5%
0.5%
0.0%
minimum
100
100
100
100
100
100
96.125
94.01
94
94
94
Moments
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
200
150
100
50
0
Quantiles
6.455
100.0% maximum
6.455
99.5%
6.455
97.5%
5.9693
90.0%
quartile 1.57925
75.0%
0
median
50.0%
0
quartile
25.0%
0
10.0%
0
2.5%
0
0.5%
0
0.0%
minimum
Moments
98.49
2.5440344
0.8044943
100.30989
96.670107
10
10
984.9
6.4721111
-1.314631
-0.054299
2.5830383
4
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
TSS EMC Influent (mg/l)
100.0% maximum 214.415
214.415
99.5%
214.415
97.5%
203.96
90.0%
quartile 72.9735
75.0%
47.412
median
50.0%
quartile 16.9928
25.0%
13.3059
10.0%
13.264
2.5%
13.264
0.5%
13.264
0.0%
minimum
Moments
0.9626
2.0393074
0.6448856
2.4214326
-0.496233
10
10
9.626
4.1587745
2.6190281
7.224018
211.85408
4
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
60.3698
61.3782
19.409491
104.27712
16.462481
10
10
603.698
3767.2834
2.0733685
4.6310886
101.67037
4
Page 38
Appendix E--- Total Phosphorus distributions for the first three years of
operation
September 2004-January 2007
Page 39
May-November
Page 40
November-April
Distributions Season 2=Cold
TP (%R)
TP EMC Effluent (mg/l)
100
95
90
85
80
75
70
65
60
55
0.2
0.02
0.1
0.01
0.05
0.005
Quantiles
98
98
98
98
88.9
70.1
56.4
55.2
55.2
55.2
55.2
Moments
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
0.15
0.015
Quantiles
100.0% maximum
99.5%
97.5%
90.0%
75.0%
quartile
50.0%
median
25.0%
quartile
10.0%
2.5%
0.5%
0.0%
minimum
TP EMC Influent (mg/l)
0.025
72.14
17.535336
7.8420406
93.912995
50.367005
5
5
360.7
307.488
0.7572681
-0.362207
24.307369
9
100.0% maximum
99.5%
97.5%
90.0%
quartile
75.0%
median
50.0%
quartile
25.0%
10.0%
2.5%
0.5%
0.0%
minimum
Quantiles
0.024
0.024
0.024
0.024
0.022
0.018
0.0095
0.004
0.004
0.004
0.004
Moments
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
100.0% maximum
99.5%
97.5%
90.0%
quartile
75.0%
median
50.0%
quartile
25.0%
10.0%
2.5%
0.5%
0.0%
minimum
0.22
0.22
0.22
0.22
0.1595
0.081
0.0375
0.033
0.033
0.033
0.033
Moments
0.0162
0.0075631
0.0033823
0.0255908
0.0068092
5
5
0.081
0.0000572
-1.235069
1.9622598
46.685606
9
Mean
Std Dev
Std Err Mean
Upper 95% Mean
Lower 95% Mean
N
Sum Wgt
Sum
Variance
Skewness
Kurtosis
CV
N Missing
0.095
0.0749833
0.0335336
0.1881041
0.0018959
5
5
0.475
0.0056225
1.5449706
2.5514457
78.929823
9
Page 41
Appendix F--- Nitrate distributions for the first three years of operation
September 2004-January 2007
Page 42
May-November
Page 43
November-April
Page 44
Appendix G--- UNHSC Subsurface Gravel Wetland Design
Specifications
Page 45
UNHSC
Subsurface Gravel Wetland
Design Specifications
June 2009
University of New Hampshire Stormwater Center (UNHSC)
Gregg Hall ● 35 Colovos Road ● Durham, New Hampshire 03824-3534 ● http://www.unh.edu/erg/cstev
UNHSC SUBSURFACE GRAVEL WETLAND
DESIGN SPECIFICATIONS
JUNE 2009
NOTICE
The specifications listed herein were developed by the UNHSC for UNHSC related projects and
represent the author’s best professional judgment. No assurances are given for projects other than the
intended application. The design specifications provided herein are not a substitute for licensed,
qualified engineering oversight and should be reviewed, and adapted as necessary and on full
recognition of site specific conditions.
ACKNOWLEDGEMENTS
These specifications were developed at the University of New Hampshire Stormwater Center
(UNHSC), of Durham, New Hampshire. The principal UNH authors are Robert M. Roseen, PE, Ph.D.,
and Thomas P. Ballestero, PE, Ph.D., PH, CGWP, PG. The contractor who constructed the original
system in 2004 at the University of New Hampshire field site was Mike Alesse of East Coast
Excavating. Other contributions to the project were made by James Houle, UNHSC Program Manager.
Special recognition goes to Mr. Richard Claytor of the Horsley Witten Group (and formerly the Center
for Watershed Protection) for his immense influence on improved stormwater management and for
largely starting the use of gravel wetlands for stormwater applications.
The UNH Stormwater Center is housed within the Environmental Research Group (ERG) at the
University of New Hampshire (UNH) in Durham, New Hampshire. Funding for the program was and
continues to be provided by the Cooperative Institute for Coastal and Estuarine Environmental
Technology (CICEET) and the National Oceanic and Atmospheric Administration (NOAA).
Robert M. Roseen, PE., PhD.
Director, UNHSC
Phone: 603-862-4024
robert.roseen@unh.edu
Thomas P. Ballestero, PE, PhD
Senior Scientist, Principal
Investigator
Phone: 603-862-4024
tom.ballestero@unh.edu
James J. Houle, CPSWQ
Program Manager, Outreach
Coordinator
Phone: 603-767-7091
james.houle@unh.edu
University of New Hampshire Stormwater Center (UNHSC)
Gregg Hall ● 35 Colovos Road ● Durham, New Hampshire 03824-3534 ● http://www.unh.edu/erg/cstev
UNH STORMWATER CENTER
SUBSURFACE GRAVEL WETLAND
JULY 2008
DESIGN SUMMARY
Category Type: Filtration system and stormwater wetland
Layout Description
The subsurface gravel wetland (SGW) is designed as a series of horizontal flow-through
treatment cells, preceded by a sedimentation basin (forebay) (Figure 1). The device is designed
to retain and filter the entire Water Quality Volume (WQV): 10% in the forebay, and 45% in
each of the respective treatment cells. For small, frequent storms, each treatment cell filters
100% of the WQV. The SGW is designed as flow through treatment, where the stormwater
passes through a gravel substrate that is a microbe rich environment. The device can be designed
with a multi-staged outlet control to detain the Channel Protection Volume (CPV) and allow
overflow for larger storms. By design, the WQV is contained and filtered then drains to
stormwater conveyance or receiving waters. All surface basin (and forebay) side slopes are 3:1
or flatter for maintenance. Standing water of significant depth is not expected other than during
large rainfall events. The wetlands cell soils are to be continuously saturated below a depth of
four inches (10 cm) from the ground surface in order to both promote water quality treatment
conditions and support wetland vegetation. To force this near-surface ground water condition,
the system primary outlet has an invert four inches (10 cm) below the wetland ground surface
(see Figure 1). Vertical perforated or slotted risers deliver the forebay outflow to the gravel
below. Within the gravel layer, horizontal subdrains distribute the incoming flow, which then
passes through the gravel substrate to subdrains on the downstream end. These subdrains collect
the flow and deliver it into the next cell. Hydraulic control of the system occurs at the primary
outlet. For large precipitation events, this hydraulic control throttles the flow through the
system and forces stormwater inflow to be temporarily stored above the wetland surfaces. By
design, the ponded water slowly drains down into the gravel layer below and is filtered through
this layer prior to leaving the system. Precipitation events larger than the WQV will have some
portion that overflows to receiving waters through an emergency spillway. At a minimum this
“spilled” water will have received minor treatment much as in a traditional detention
pond/wetland system.
System Functionality
System functionality has multiple components. From a unit process perspective, sedimentation,
filtration, physical and chemical sorption, microbially mediated transformation, and uptake and
storage predominate. There is pre-treatment by a sedimentation forebay. This is followed by
treatment above the wetland cells by sedimentation akin to a dry pond. In addition, there is water
quality treatment during the flow through the wetland plants as well as the minor infiltration
through the wetland soil to the gravel below. Finally and predominantly, within the gravel layer
there is treatment involving filtration, sorption, uptake and storage, and microbially mediated
transformation. The conversion and removal of nitrogen is dependent on two conditions: an
aerobic sedimentation forebay followed by subsurface anaerobic treatment cells. Aerobic
conditions exist in the forebay when it is designed and maintained as a dry area with temporary
Gravel Wetland Design and Maintenance
UNH Stormwater Center
June 2009
G-1
ponding conditions during storm events. The anaerobic condition in the treatment cells is created
by maintaining the high water table within the system as well as the slow flow through the gravel
layer. This saturated condition drives the dissolved oxygen level down and creates conditions in
which nitrate conversion to nitrogen gas occurs.
Retrofit Options
SGWs are well suited for retrofits within stormwater pond systems. These SGW systems are well
suited because: 1) there is a limited hydraulic head requirement, 2) the SGW can be lined and
does not require separation from groundwater, and 3) there is a straightforward placement of the
system within the footprint of existing stormwater ponds. Hydraulic head requirements for gravel
wetlands are approximately four inches (10 cm), whereas underdrained filtration systems may
require as much as three feet (one meter) or more. Because the SGW is a horizontal porous
media flow system it does require a hydraulic head to drive the water through the system. At a
minimum, the driving head is the difference between the vertical distance from the ponded water
level above the wetland surface and the invert of the primary outlet. To maintain the system in
its saturated condition, it must be situated in low hydraulic conductivity soils or lined below the
gravel layer. Because infiltration is not designed to occur, separation from groundwater is not
required and the SGWs are sited much like stormwater ponds. While SGWs have a relatively
large footprint for a stormwater quality treatment technology, they easily fit within the footprint
of existing stormwater ponds that were sized for flood control. When retrofitting a SGW system
into a stormwater pond, it is located towards the outlet of the pond. The area within the pond
preceding the SGW is used for pretreatment. A wet pond retrofit would require a conversion to a
dry pond by the elimination of the permanent pool.
SPECIFICATIONS SUMMARY
• May be preceded by pre-treatment: hydrodynamic separator, swale, forebay. Pre-treatment
should normally be capable of holding 10% of the WQV.
• Two treatment cells.
• A subsurface water level is maintained through the design of the outlet invert elevation
(invert just below the wetland soil surface).
• Retain and filter the entire Water Quality Volume (WQV), 10% in the forebay, and 45%
above each of the respective treatment cells.
• Option to retain the Channel Protection Volume (CPV) for 24-48 hrs.
• No-geotextile or geofabric layers are used within this system, but may be used to line walls.
• If a native low hydraulic conductivity soil is not present below the desired location of the
SGW, a low permeability liner or soil (hydraulic conductivity less than 10-5 cm/s = 0.03
ft/day) below the gravel layer should be used to minimize infiltration, preserve horizontal
flow in the gravel, and maintain the wetland plants (Figure 2).
• Gravel length to width ratio of 0.5 (L:W) or greater is needed for each treatment cell with a
minimum flow path (L) within the gravel substrate of 15 feet (4.6 m).
• 8 in. (20 cm) minimum thickness of a wetland soil as the top layer. (See description in
Surface Infiltration Rates section for details (Figure 2)). This layer is leveled (constructed
with a surface slope of zero).
Gravel Wetland Design and Maintenance
UNH Stormwater Center
June 2009
G-2
Figure 15: Gravel Wetland—Concept Design
WQV Bypass
3” min pea gravel
Gravel Wetland Design and Maintenance
UNH Stormwater Center
June 2009
Large flow
bypass
WQV release
by orifice
control
G-3
•
•
•
•
•
•
•
•
•
•
•
•
3 in. (8 cm) minimum thickness of an intermediate layer of a graded aggregate filter is needed to prevent the
wetland soil from moving down into the gravel sub-layer. Material compatibility between layers needs to be
evaluated.
24 in. (0.6 m) minimum thickness of ¾-in (2 cm) crushed-stone (gravel) sub-layer. This is the active zone
where treatment occurs.
The primary outlet invert shall be located 4” (10 cm) below the elevation of the wetland soil surface to
control groundwater elevation. Care should be taken to not design a siphon that would drain the wetland:
the primary outlet location must be open or vented. In contrast to Figure 1, the primary outlet can be a
simple pipe.
An optional high capacity outlet at equal elevation or lower to the primary outlet may be installed for
maintenance. Thus outlet would need to be plugged during regular operation. This optional outlet allows for
flushing of the treatment cells at higher flow rates. If it is located lower, it can be used to drain the system
for maintenance or repairs.
The Bypass outlet (emergency spillway, or secondary spillway) is sized to pass designs flows (10-year, 25year, etc.). This outlet is sized by using conventional routing calculations of the inflow hydrograph through
the surface storage provided by the subsurface gravel wetland system. Local criteria for peak flow
reductions are then employed to size this outlet to meet those criteria.
The primary outlet structure and its hydraulic rating curve are based on a calculated release rate by orifice
control to drain the WQV in 24-48 hrs. For orifice diameter calculations refer to the NY Stormwater Manual
(2001) or HDS 5 (FHWA, 2005) for details.
The minimum spacing between the subsurface perforated distribution line and the subsurface perforated
collection drain (see Figure 1) at either end of the gravel in each treatment cell is 15 ft (4.6 m): there should
be a minimum horizontal travel distance of 15 ft (4.6 m) within the gravel layer in each cell.
Vertical perforated or slotted riser pipes deliver water from the surface down to the subsurface, perforated or
slotted distribution lines. These risers shall have a maximum spacing of 15 feet (4.6 m) (Figure 1).
Oversizing of the perforated or slotted vertical risers is useful to allow a margin of safety against clogging
with a minimum recommended diameter of 12” (30 cm) for the central riser and 6” (15 cm) for end risers.
The vertical risers shall not be capped, but rather covered with an inlet grate to allow for an overflow when
the water level exceeds the WQV.
Vertical cleanouts connected to the distribution and collection subdrains, at each end, shall be perforated or
slotted only within the gravel layer, and solid within the wetland soil and storage area above. This is
important to prevent short-circuiting and soil piping.
Berms and weirs separating the forebay and treatment cells should be constructed with clay, or nonconductive soils, and/or a fine geotextile, or some combination thereof, to avoid water seepage and soil
piping through these earthen dividers.
The system should be planted to achieve a rigorous root mat with grasses, forbs, and shrubs with obligate
and facultative wetland species. In northern climates refer to the NY Stormwater Manual
(http://www.dec.state.ny.us/website/dow/toolbox/swmanual/) or approved equivalent local guidance for
details on local wetland plantings.
Standard design approach for stormwater ponds should be followed as per the NY Stormwater Manual
(2001) or approved equal with regards to forebay, spillways, bypass, side slopes, erosion control, use of rip
rap for stabilized regions at outlets and other locations of concentrated flow. etc.
Gravel Wetland Design and Maintenance
UNH Stormwater Center
June 2009
G-7
SURFACE INFILTRATION RATES AND HYDROGEOLOGIC MATERIALS
Wetland Soil
The surface infiltration rates of the gravel wetland soil should be similar to a low hydraulic conductivity
wetland soil (0.1-0.01 ft/day = 3.5 x 10-5 cm/sec to 3.5 x 10-6 cm/sec)). This soil can be manufactured using
compost, sand, and some fine soils to blend to a high % organic matter content soil (>15% organic matter).
Avoid using clay contents in excess of 15% because of potential migration of fines into subsurface gravel layer.
Do not use geotextiles between the horizontal layers of this system as they will clog due to fines and may
restrict root growth.
An intermediate layer of a graded aggregate filter (i.e., pea gravel) is needed to prevent the wetland soil from
migrating down into the crushed-stone (gravel) sub-layer. This is to prevent migration of the finer setting bed
(wetland soil) into the coarse sublayer. Material compatibility should be evaluated using FHWA criteria (see
Ferguson, 2005):
Criteria 1:
D15, COARSE
SUBLAYER
≤ 5 x D85 , SETTING
Criteria 2:
D50 , COARSE
SUBLAYER
≤ 25 x D50 ,
BED
SETTING BED
Below the wetland soil and pea gravel is a crushed stone (gravel) sublayer with a 24 in. (0.6 m) minimum
thickness. Angular crushed stone is needed with a minimum size ~3/4–in (2-cm). Large particle, angular coarse
to very coarse gravel is needed to maintain system longevity.
Figure 16: Gravel Wetland Materials Cross -Section
8” (20 cm) minimum
thickness of wetland soil
3” (8 cm) minimum thickness of
graded filter (i.e., pea gravel) if
needed
FLOW
24” (60 cm) minimum thickness
of ¾” (2 cm) crushed stone
(gravel)
Low permeability soil or
liner if underlying soils are
very permeable
Native Materials and Liner
If native a low hydraulic conductivity native soil is not present below the gravel layer, a low permeability liner
or soil should be used to: minimize infiltration, preserve horizontal flow in the gravel, and maintain the wetland
plants. If geotechnical tests confirm the need for a liner, acceptable options include: (a) 6 to 12 inches (15 – 30
Gravel Wetland Design and Maintenance
UNH Stormwater Center
June 2009
G-8
cm) of clay soil (minimum 15% passing the #200 sieve and a minimum permeability of 1 x 10-5 cm/sec), (b) a
30 ml HDPE liner, (c) bentonite, (d) use of chemical additives (see NRCS Agricultural Handbook No. 386,
dated 1961, or Engineering Field Manual), or (e) a design prepared by a Professional Engineer.
DESIGN SOURCES
The primary design sources for the development of the subsurface gravel wetland are listed below, in the order
of use.
•
•
•
•
University of New Hampshire Stormwater Center 2002, http://www.unh.edu/erg/cstev//
Claytor, R. A., and Schueler, T. R. (1996). Design of Stormwater Filtering Systems, Center for Watershed
Protection, Silver Spring, MD.
Georgia Stormwater Management Manual, Volume 2: Technical Handbook, August 2001, prepared by
AMEC Earth and Environmental, Center for Watershed Protection, Debo and Associates, Jordan Jones and
Goulding, Atlanta Regional Commission.
New York State Stormwater Management Design Manual, October 2001, prepared by Center for Watershed
Protection, 8391 Main Street, Ellicott City, MD 21043, for New York State, Department of Environmental
Conservation, 625 Broadway, Albany, NY 12233.
MAINTENANCE
Maintenance and Inspection Recommendations are largely adapted from CTDEP (2004) Stormwater Quality
Manual for filtration systems. Inspection schedules have two periods: i) 1st Year Post-Construction, ii) PostConstruction Routine Monitoring. Maintenance is critical for the proper operation of subsurface gravel wetland
systems. 1st Year Post-Construction monitoring differs primarily by its increased frequency to assure proper
vegetative establishment and system functioning. Post-Construction Routine Monitoring is based on USEPA
requirements for Good Housekeeping practices.
Unlike other filtration systems, a subsurface gravel wetland is a subsurface, horizontal filtration system and
does not rely upon the surface soils for treatment. As such, surface infiltration rates are expected to be low and
are not used for the criteria for cleaning/maintenance. Rather, stormwater access to the subsurface gravel layer
is the critical hydraulic performance measure.
Gravel Wetland Design and Maintenance
UNH Stormwater Center
June 2009
G-9
Inspection and Maintenance
❍ 1st Year Post-Construction: Inspection frequency should be after every major storm in the first year
following construction.
❑ Inspect to be certain system drains within 24-72 hrs (within the design period, but also not so quickly
as to minimize stormwater treatment)).
❑ Watering plants as necessary during the first growing season
❑ Re-vegetating poorly established areas as necessary
❑ Treating diseased vegetation as necessary
❑ Quarterly inspection of soil and repairing eroded areas, especially on slopes
❑ Checking inlets, outlets, and overflow spillway for blockage, structural integrity, and evidence of
erosion.
❍ Post-Construction: Inspection frequency should be at least every 6 months thereafter, as per USEPA Good
House-Keeping Requirements. Inspection frequency can be reduced to annual following 2 years of monitoring
that indicates the rate of sediment accumulation is less than the cleaning criteria listed below. Inspections
should focus on:
❑ Checking the filter surface for dense, complete, root mat establishment across the wetland surface.
Thorough revegetation with grasses, forbs, and shrubs is necessary. Unlike bioretention, where mulch is
commonly used, complete surface coverage with vegetation is needed.
❑ Checking the gravel wetland surface for standing water or other evidence of riser clogging, such as
discolored or accumulated sediments.
❑ Checking the sedimentation chamber or forebay for sediment accumulation, trash, and debris.
❑ Inspect to be certain the sedimentation forebay drains within 24 to 72 hrs.
❑ Checking inlets, outlets, and overflow spillway for blockage, structural integrity, and evidence of
erosion.
❑ Removal of decaying vegetation, litter, and debris.
❍ Cleaning Criteria for Sedimentation Forebay: Sediment should be removed from the sedimentation
chamber (forebay) when it accumulates to a depth of more than 12 inches (30 cm) or 10 percent of the
pretreatment volume. The sedimentation forebay should be cleaned of vegetation if persistent standing water
and wetland vegetation becomes dominant. The cleaning interval is approximately every 4 years. A dry
sedimentation forebay is the optimal condition while in practice this condition is rarely achieved. The
sedimentation chamber, forebay, and treatment cell outlet devices should be cleaned when drawdown times
exceed 60 to 72 hours. Materials can be removed with heavy construction equipment; however this equipment
should not track on the wetland surface. Revegetation of disturbed areas as necessary. Removed sediments
should be dewatered (if necessary) and disposed of in an acceptable manner.
❍ Cleaning Criteria for Gravel Wetland Treatment Cells: Sediment should be removed from the gravel
wetland surface when it accumulates to a depth of several inches (>10 cm) across the wetland surface. Materials
should be removed with rakes rather than heavy construction equipment to avoid compaction of the gravel
wetland surface. Heavy equipment could be used if the system is designed with dimensions that allow
equipment to be located outside the gravel wetland, while a backhoe shovel reaches inside the gravel wetland to
remove sediment. Removed sediments should be dewatered (if necessary) and disposed of in an acceptable
manner.
Gravel Wetland Design and Maintenance
UNH Stormwater Center
June 2009
G-10
❍ Draining and Flushing Gravel Wetland Treatment Cells: For maintenance it may be necessary to drain or
flush the treatment cells. The optional drains will permit simpler maintenance of the system if needed. The
drains need to be closed during standard operation. Flushing of the risers and horizontal subdrains is most
effective with the entire system drained. Flushed water and sediment should be collected and properly
disposed.
REFERENCES
• Center for Watershed Protection (CWP). 2001. The Vermont Stormwater Management Manual – Public
Review Draft. Prepared For Vermont Agency of Natural Resources.
• CTDEP (2004). Connecticut Stormwater Quality Manual. Hartford, CT.
• Ferguson, B. K. (2005). Porous Pavements, CRC Press.
• FHWA. (2005). "Hydraulic Design of Highway Culverts (HDS-5)." NHI-01-020, Federal Highway
Administration.
• Hepting, G. H., 1971, Diseases of Forest and Shade Trees of the United States, USDA Agricultural
Handbook No. 386, USDA, Washington, DC
• New York State Stormwater Management Design Manual, October 2001, prepared by Center for Watershed
Protection, 8391 Main Street, Ellicott City, MD 21043, for New York State, Department of Environmental
Conservation, 625 Broadway, Albany, NY 12233.
• UNHSC, Roseen, R., T. Ballestero, and Houle, J. (2007). "UNH Stormwater Center 2007 Annual Report."
University of New Hampshire, Cooperative Institute for Coastal and Estuarine Environmental Technology,
Durham, NH.
• USDA, 1979, Engineering field manual for conservation practices, U.S. Dept. of Agriculture, Soil
Conservation Service, Washington, D.C.
Gravel Wetland Design and Maintenance
UNH Stormwater Center
June 2009
G-11
Appendix H‐‐‐ UNHSC Gravel Wetlands Planting Assessment – October 2006
Page 57
Planting History. The site was constructed in 2003-2004. The gravel wetland was
planted in 2004 with a wetland seed mix and selected wetland plants including shrubs and
trees. In October 2006 the site was surveyed to determine what plant species were
present, and which of the planted species survived. The wetlands contained
predominantly emergent marsh/wet meadow species, and all areas with standing water
were populated by Typha (cattail). No Phragmites was found, a few Purple Loosestrife
plants were found and removed. Note that the assessment was conducted in October,
very late in the growing season.
Gravel wetland – mixed wetland grasses, reeds, herbaceous plants and shrubs growing
vigorously. 53% of the planted species are still present. Trees and shrubs had a high
survival, but the emergent obligate wetland species (e.g water lily, pickerelweed) survival
was very low. 100% cover, except for open water in forebay. Very few upland plants
were present, and the wetland system seemed generally healthy and diverse.
Listing of plant species planted at UNHSC or identified.
CO
Acer rubrum
FAC
Red maple
ACC
IV
AI
Acorus calamos
Alnus incana
Amelanchier
Canadensis
Asclepias incarnata
Aster novae-angiae
Aster puniceus
Betula nigra
Bidens connata
Caltha palustris
Carex comosa
Carex lurida
Clethra alnifolia
Eleocharis rostellata
Eurpatorium
maculatum
Eurpatorium
perfoliatum
Fraxinus
pennsylvanica
Ilex verticillata
Iris versicolor
Juncus Canadensis
Lindera benzoin
Lythrum salicaria
Myrica pensylvanica
Nuphar luteum
OBL
FAC
FAC
Sweetflag
Speckled alder
Shadblow
serviceberry
Swamp milkweed
New England aster
Swamp aster
River birch
Stick-tight
Marsh marigold
Bearded sedge
Lurid sedge
Sweet pepperbush
Spike rush
Joepieweed
ASI
ANA
ASP
BN
BC
CP
CC
CL
CA
ER
EM
EP
CA
IV
IVE
JC
LB
LS
MP
NL
OBL
FACW
OBL
FACW
FACW
OBL
OBL
OBL
FAC
OBL
FACW
FACW Boneset
FACW Green ash
FACW
OBL
OBL
FACW
FACW
FAC
OBL
Winterberry
Blue flag iris
Canada rush
Buttonbush
Purple loosestrife
Bayberry
Yellow water lily
H-1
NS
OC
PAC
PV
PHA
PA
PC
RF
RV
RP
SL
CS
SAC
SAT
SV
SC
SO
SpL
TO
TXR
TR
TL
VA
VH
VC
VT
OBL
FACW
FAC
FACU
UPL
?
Nyssa sylvatica
Osmunda
cinnamomea
Panicum agrostoides
Peltanda virginica
Phragmites australis
Polygonum
amphibium
Pontederia cordata
Rhamnus frangula
Rhododendron
viscosum
Rosa palustris
Sagittaria latifolia
Salix nigra
Scirpus acutus
Scirpus atrovirens
Scirpus validus
Scripus cyperinus
Solidago
Spirea latifolia
Taraxacumum
officinale
Toxicodendron
radicans
Trifolium
Typha latifolia
Valisneria
americana
Verbena hastate
Viburnum
cassinoides
Viburnum trilobum
FACW Tupelo
FACW Cinnamon fern
?
OBL
FACW
OBL
Panic grass
Arrow arum
Common reed
Water smartweed
OBL
FAC
OBL
Pickerelweed
Buckthorn
Swamp azalea
OBL
OBL
FACW
OBL
OBL
OBL
FACW
FAC
FACU
Swamp rose
Northern arrowhead
Black willow
Hard-stem bulrush
Green bulrush
Soft-stem bulrush
Wool grass
Goldenrod
Meadowsweet
Common dandelion
FAC
Poison ivy
FACU
OBL
OBL
Clover
Common cattail
Wild celery
FACW Blue vervain
FACW Wild raisin
FACW American cranberry
bush
Obligate wetland species
Usually in wetlands
Either wetlands or uplands
Usually in uplands
Obligate upland
Not listed
From US Fish and Wildlife, 1988 National List of Plant Species that Occur in Wetlands –
Region 1.
H-2
Photos taken 10/11/06 and 10/13/06 at UNHSC, by A. Watts.
Gravel wetland forebay
H-3
Gravel Wetland Middle
H-4
Gravel Wetland West
H-5
Gravel Wetland Planting Plan
H-6