You are on page 1of 10

Performance Analysis of a Wet-Retention Pond

in a Small Agricultural Catchment


François Chrétien 1; Patrick Gagnon 2; Georges Thériault 3; and Mikael Guillou 4
Downloaded from ascelibrary.org by University of California, San Diego on 01/14/16. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Urban wet-retention pond performance has been extensively studied for many water contaminants. However, very few studies
have focused on wet-retention pond efficiency under agricultural production systems. This study presents the results drawn from a 5-year data
set collected on a retention pond constructed in Saint-Samuel (Canada) at the outlet of a 23.1-hectare watershed under cash crop production. A
total of 62 runoff events were analyzed for hydrological performance, 20 events for total suspended solids (TSS), 11 events for total nitrogen
(TN), and 14 events for total phosphorus (TP). Peak flows were reduced by 38% on average and the retention pond reduced the frequency and
duration of exceedance of the erosion threshold. Drawdown times for 50, 75, 90, and 100% of the runoff volume were 7, 18, 28, and 42 h,
respectively. The mean detention time was 3.2 h compared to the watershed lag time of 3.4 h. Mean removal efficiency ratios based on event-
mean concentrations (EMC) and event-total loads (ETL) were similar to those reported for urban areas, with values of 50–56%,
42–52%, and 48–59% for TSS, TN, and TP, respectively. Effluent water quality analyses have shown that environmental criteria were often
exceeded and threshold concentrations below which treatment is unlikely to occur were similar to irreducible concentrations reported for
urban areas with estimated values of 30, 1, and 0.2 mg=L for TSS, TN, and TP, respectively. The primary treatment mechanism of the
retention pond is most likely the sedimentation of sand and silt particles and particulate nutrients. DOI: 10.1061/(ASCE)EE.1943-
7870.0001081. © 2016 American Society of Civil Engineers.

Introduction and technical publications of which only 10 dealt specifically with


retention structures removal efficiency in agricultural settings.
The link between degradation of freshwater aquatic ecosystems and Moreover, and in opposition to well-established urban design cri-
stormwater runoff has already been established. Runoff events have teria, no consensus on design parameters exists for ag-retention
been shown to transport a multitude of contaminants to water ponds (Tiessen et al. 2011).
bodies, including suspended solids, nutrients, pesticides, bacteria, Most efficiency studies published to date for agricultural water-
oils, and heavy metals (Carpenter et al. 1998; U.S. EPA 1996; sheds have focused on dams. Multipurpose dams were historically
Puckett 1995). Source management, in-field conservation practices, built on agricultural watersheds for flood control, irrigation, livestock
and edge-of-field best management practices (BMPs) can all be needs, and/or recreation. As reported by Rausch and Schreiber (1981),
used to mitigate those water contamination issues. Wet-retention considerable research on trap efficiency (TE) has been conducted
ponds are edge-of-field BMPs that are widely used in urban area since the mid-1940s. However, TE is mainly a by-product service
to temporarily store stormwater and reduce pollutant levels in to the dam main purpose and these retention structures are relatively
downstream water bodies (Mallin et al. 2002). Urban wet-retention large compared to ag-retention ponds implemented as an edge-of-field
ponds have been extensively studied and within the last 25 years at stormwater BMP. This study was conducted to evaluate the environ-
least 10 sets of removal efficiency ratios have been published for mental and hydrological performance of a small ag-retention pond
suspended solids, nutrients, and other pollutants (Schueler 2011). designed to store stormwater runoff from an agricultural watershed.
However, very few studies have focused on wet-retention ponds The specific objectives of this project were to do the following:
efficiency under agricultural production systems (hereafter referred 1. Evaluate the ag-retention pond hydrological impact, that is,
to as ag-retention ponds). In an attempt to build an agricultural evaluate its capacity at reducing peak flow, detaining and retain-
BMP database, Clary et al. (2012) reviewed over 800 references ing storm volumes and mitigating erosion issues in the receiving
water body.
1
Research Professional, Agriculture and Agri-Food Canada, 2560 2. Determine the ag-retention pond efficiency at reducing suspended
Hochelaga Blvd., Québec City, QC, Canada G1V 2J3 (corresponding solids, total phosphorus, and total nitrogen using two different me-
author). E-mail: francois.chretien@agr.gc.ca trics: effluent water quality and removal efficiency ratios, based on
2
Postdoctoral Research Fellow, Agriculture and Agri-Food Canada, the event-total loads (ETL) (also referred to sum-of-all loads) and
2560 Hochelaga Blvd., Québec City, QC, Canada G1V 2J3. E-mail: the event-mean concentrations (EMC) methodologies.
patrick.gagnon@agr.gc.ca
3 3. Compare contaminant removal efficiency obtained under agri-
Research Professional, Agriculture and Agri-Food Canada, 2560
Hochelaga Blvd., Québec City, QC, Canada G1V 2J3. E-mail: georges
cultural production to the latest set of removal efficiency ratios
.theriault@agr.gc.ca published for urban wet-retention ponds by Schueler (2011).
4
Research Professional, Ministère de l’Agriculture, des Pêcheries et de
l’Alimentation du Québec, 1355, Johnson St., Office #3300, St-Hyacinthe,
QC, Canada J2S 8W7. E-mail: mikael.guillou@mapaq.gouv.qc.ca Material and Methods
Note. This manuscript was submitted on June 10, 2015; approved on
October 28, 2015; published online on January 8, 2016. Discussion period Site Description
open until June 8, 2016; separate discussions must be submitted for individual
papers. This paper is part of the Journal of Environmental Engineering, An ag-retention pond was constructed in July 2008 at the outlet
© ASCE, ISSN 0733-9372. of a 23.1-ha watershed under cash-crop production located in

© ASCE 04016005-1 J. Environ. Eng.

J. Environ. Eng., 04016005


Saint-Samuel, in the administrative region of Centre-du-Québec, These soils are characterized by slow external and internal drainage
Québec, Canada. The ag-retention pond flows out into an intermit- and imperfect drainage, which promote surface sealing and runoff
tent creek supplied solely by the pond, which discharges in turn (Choinière and Laplante 1948). The Mehlich-3 P/Al saturation indi-
into the Nicolet River about 50 m downstream. The pond was ces at depths of 0–5 cm and 5–20 cm are 5.25% and 4.32%, respec-
designed to store a maximum volume of 82 m3 ha−1 , comprising tively, which are both below the critical environmental threshold
a permanent pool of water of 11 m3 ha−1 (Fig. 1). value of 13.1% for mineral soils containing less than 30% clay in
The storage capacity was actually lower than the recommended this region (Pellerin et al. 2006).
value of 146 m3 ha−1 as defined by Rivard (2011) for an equivalent From the start of the study, the following succession of crops
urban wet-retention pond. In Québec (Canada), urban ponds are was grown under reduced tillage (only one pass of the harrow per
designed to treat runoff resulting from 90% of the annual rainfall year): soybeans (2008), oats (2009), soybeans (2010), oats (2011),
events (Chicago type, 6 h) and to mitigate erosion issues, in the soybeans (2012), and corn (2013). Ground leveling work was
receiving stream, associated with a rainfall event (NRCS type II, undertaken in fall 2011 and was completed in spring 2012, in order
Downloaded from ascelibrary.org by University of California, San Diego on 01/14/16. Copyright ASCE. For personal use only; all rights reserved.

24 h) (Rivard 2011). According to data from the Environment to smooth out the existing leveled areas.
Canada station No. 7022160 located 25.3 km from the experimen- Environment Canada station No. 7022160 shows for the
tal site, these events corresponded to approximately 23 and 37 mm, experimental site a historical annual temperature average of
respectively. 6.4°C, varying from −10.2°C in January to 20.9°C in July, on
The soils are mainly loamy sand (gleyed humo-ferric podzol) con- average. The mean growing season length from 1979 to 2008 is
taining 55 to 65% fine sand to very fine sand in the B and C horizons. 204 days based on a temperature above 5.5°C. The average annual

Fig. 1. Saint-Samuel ag-retention pond with (a) water level at permanent pool maximum elevation (image by François Chrétien); (b) water level at
riser maximum elevation (image by François Chrétien); (c) schematic of the pond; (d) pond storage capacity

© ASCE 04016005-2 J. Environ. Eng.

J. Environ. Eng., 04016005


precipitation is 1,114 mm, out of which 242 cm fall as snow phosphorus–persulphate, and dissolved phosphorus (<0.45 μm).
and 598 mm fall as rain during the growing season (Agrométéo Only detailed analyses of the total suspended sediment (TSS), total
Québec 2015). nitrogen (TN), and total phosphorus (TP) are shown in this paper.
TSS was determined after filtration using a 0.45-μm glass fiber
filter (Type A/E, Pall Corporation, Cortland, New York) and by
Hydrological Measurements weighing the oven-dried sediments (105°C, 24 h). Following
Precipitation monitoring was done using a TR-525M tipping Corriveau et al. (2008), water samples for nitrogen analysis were
bucket rain gauge (Texas Electronics, Dallas, Texas) located on preserved in the refrigerator at 4°C in the dark for a maximum
the experimental site. The time step used to record data was 5 min. period of 48 h. TN was determined by automated colorimetry using
Hydrological monitoring was carried out during the snow-free the 4500-org persulfate method (APHA 1995). For total phospho-
period from 2009 to 2013. Since there was a multitude of entry rus (TP) analysis, samples were kept at 4°C immediately after being
points, flows/volumes entering the ag-retention pond could not transferred back to the laboratory and filtered using a 0.45-μm ny-
Downloaded from ascelibrary.org by University of California, San Diego on 01/14/16. Copyright ASCE. For personal use only; all rights reserved.

be measured directly and were determined through a two-step in- lon membrane filter (Nylaflo, Pall Corporation, Cortland, New
verse routing process. The first step consisted in determining the York). The filtrates and the unfiltered samples were stored in a
outflow rate based on the characteristics of the water evacuation freezer (−18°C) prior to P analysis. After a persulfate autoclave
structure, that is, a perforated riser connected to a horizontal pipe digestion, TP was determined with a Perkin Elmer Lambda 35
at its bottom. Flows leaving the pond followed three distinct stages UV/vis spectrometer (Perkin Elmer, Waltham, Massachusetts) us-
of evacuation: the outflow from the orifices of the perforated riser, ing the Murphy and Riley (1962) method. Detection limits were
the outflow controlled by the hydraulics of flow in the riser, and the <1 mg=L for TSS, <0.09 mg=L for TN, and <0.008 mg=L for
outflow controlled by the hydraulics of flow in the horizontal pipe. TP. Table S2 (Supplemental Data) summarizes the data for each
The second step consisted in establishing, through land surveying, of the water quality events.
the relationship between pond depth and storage area to obtain the
temporal variation of volume by tracking changes in water depth Data Analysis
over time. Land surveying was carried out using a total station
theodolite with an elevation precision of 5 mm and isolines were
developed at each 10-cm intervals using ArcGIS software (ESRI Impact on Hydrology
2012). Water depth monitoring in the ag-retention pond was The variable used to evaluate the impact of the ag-retention pond
performed using a Hobo water level logger (U20-001-01, Onset on peak flow for a given event was the reduction rate (R), calculated
Computer Corporation, Bourne, Massachusetts) operating with a as follows:
time step of 15 min. For a given time step, the inflow volume equals  
Qout
the sum of the outflow volume and the pond volume variation. The R ¼ 100% 1 − ð1Þ
Qin
inflow volume calculated this way comprises surface runoff, but
also includes interflow and groundwater volumes. However, since where Qin and Qout are respectively the highest instantaneous
the pond bottom and sidewalls were lined with an impervious clay inflow and outflow values observed at a 15-min time step during
layer (Champlain clay), it is assumed that interflow and ground- the event. Since R varies between events, correlations were com-
water contributions are marginal compare to surface runoff. A puted between peak flow reduction rates and some potential
2.5H flume was also installed in 2011 at the pond outlet, with explanatory variables. Due to the asymmetric distribution of the
Hobo water level loggers mounted in a stilling well, in order to reduction rates, the Kendall’s tau (τ ) (Lehmann 1998), a nonpara-
validate the outflow rate. Table S1 (Supplemental Data) shows metric correlation coefficient, was preferred to the more commonly
the hydrological properties of each of the 62 monitored events. used Pearson correlation coefficient.
An automated sampling campaign was carried out during the The ag-retention pond volume reduction capacity was also
snow-free period from 2010 to 2013. The sampling station con- assessed using Eq. (1) by replacing Qin and Qout by events total
sisted of two portable autosamplers (6712, Teledyne ISCO, Lin- inflow and outflow volumes (V in and V out ). The drawdown time
coln, Nebraska), which were triggered when the water level is defined as the interval between the time when the maximum
increased by 0.40 m relative to the level of the permanent pool. volume was attained in the ag-retention pond during a rainfall event
Samples were collected at field edge of the main inlet and at and the time when the total inflow volume has been released. To
the bottom of the outlet riser over a 24-h period after the trigger assess storage dynamics, drawdown times were subdivided into
signal. Even though water quality may vary for each entry point, four categories corresponding to the intervals between the time
water characteristics from the main inlet were assumed to be at which the maximum volume was attained and the times when
representative of inlet concentrations, since they all drain water 50, 75, 90, and 100% of the event total volume have been released.
from the same field with homogenous soils and cropping practices. In order to relativize the importance of the water holding capac-
Six 1-L composite samples, individually made of eight 125-mL ity induced by the ag-retention pond, the detention time (T d ) was
subsamples collected every 30 min during a period of 4 h and col- also computed and compared to the mean watershed lag time (T l ).
lectively covering a period of 24 h, were collected at each of the two T d is defined as the time interval between the center-of-mass of the
sampling points. inflow hydrograph and the center-of-mass of the outflow hydro-
graph while T l is defined as the time interval between the
center-of-mass of rainfall and the center-of-mass of the resulting
Analytical Methods
runoff (inflow hydrograph).
The collected water samples were immediately put on ice and The combined effect of peak flow reduction and detention
sent to the Québec government’s Centre d’expertise en analyse en- capacity of the pond on the receiving stream morphology and
vironnementale from 2010 to 2012 and to the Agriculture and Agri- erosion processes was assessed by computing the specific stream
Food Canada water quality laboratory in 2013. The samples were power (i.e., the rate of energy dissipation against the bed and
analyzed to determine the concentrations of suspended solids banks of a stream per unit channel width) for inflows and outflows
(>0.45 μm), total nitrogen, nitrite-nitrate, ammonia nitrogen, total (Bagnold 1966)

© ASCE 04016005-3 J. Environ. Eng.

J. Environ. Eng., 04016005


pgQS efficiency ratios were computed in order to compare scarce ag-
ω¼ ð2Þ
w performance results to broadly available urban retention pond stud-
ies. While easily interpretable, mean removal efficiency ratios alone
where ω is the specific stream power (W=m2 ), p is the density of are poor metrics to evaluate the efficiency of a BMP (Strecker et al.
water (kg=m3 ), g is the acceleration due to gravity (9.8 m=s2 ), Q is 2001). Among their weaknesses, removal efficiency ratios alone do
the flow (m3 =s), S is the slope of the stream (0.032%), and w is the not permit to isolate the contribution of water volume reduction
surface channel width (about 2.2 m for high flows). and pollutant reduction. Also, mean removal efficiency ratios often
give erroneous indications about the effluent quality, since removal
Impact on Water Quality efficiency ratios tend to be higher when the inflow water is dirty
Effluent water quality and removal efficiency ratios metrics were (Stormwater Assessment Monitoring and Performance 2005), and
used to assess the ag-retention pond impact on water quality. they do not indicate whether or not the pollutant reduction is
Removal efficiency ratios may differ significantly whether they statistically significant or not. To circumvent these issues, the
Downloaded from ascelibrary.org by University of California, San Diego on 01/14/16. Copyright ASCE. For personal use only; all rights reserved.

are calculated with concentrations or total loads (Strecker Wilcoxon signed-rank test (Lehmann 1998) was applied on
et al. 2001; Lenhart and Hunt 2011). In this study, the event-total EMCin and EMCout for each pollutant to determine whether the
load (ETL) and the event-mean concentration (EMC) were impact of the pond on the reduction of exported concentrations
calculated as follows for a given event and a given pollutant was statistically significant or not. Data from each event are also
(TSS, TN, or TP) presented in different ways: boxplots of efficiency ratios, graphs of
X
n the inflow versus outflow event concentrations, and empirical
ETL ¼ V i Ci and ð3Þ distribution of concentrations.
i¼1 This latter analysis of effluent water quality allows evaluating
whether the ag-retention pond is able to put a cap on outflow con-
ETL centration or whether outflow concentration continuously increases
EMC ¼ Pn ð4Þ
i¼1 V i
with inflow concentration. It also helps to easily identify the
proportion of events for which inflow and outflow concentrations
where V i is the volume of water that flows during the 4-h sampling exceeded the recommended ambient water quality criteria. Estima-
period i, Ci is the concentration of the pollutant in the composite tion of the concentration for each pollutant below which the studied
sample of period i, and n is the number of composite samples dur- ag-retention pond has no impact, was also computed using the
ing the event (up to six for a maximum of 24 h). concentrations distribution.
The removal efficiency ratio (EETL ) for a given event and a given
pollutant was calculated from the ETL metric
  Results and Discussion
ETLout
EETL ¼ 100% 1 − ð5Þ
ETLin Impact on Hydrology
where the subscripts in and out designate whether the metric was During the monitoring period, 90% of the 268 rainfall events gen-
calculated from inflow or outflow data. erated less than 25 mm of rainfall, which was very close to the
The mean removal efficiency ratio (EETL ) is not the arithmetic historical 90th percentile. Indeed, between 1967 and 2000, 90%
average of the efficiency ratios from all the events; it is rather the of the 1,341 monitored events (minimum rainfall depth of 1 mm
efficiency ratio determined from the mean of the ETLs calculated and 6-h interevent period) at the Environment Canada reference
from all events station No. 7022160 consisted of rainfall events of 23 mm or less.
 
ETLout Peak Flow and Volume Reduction
EETL ¼ 100% 1 − ð6Þ
ETLin Fig. 2(a) illustrates the reduction in peak flow estimated as a
function of precipitation depth for the 62 rainfall events greater
Removal efficiency ratios were applied to the EMC metric than 10 mm. The percentage of reduction in peak flow showed a
as well by replacing ETL by EMC in Eqs. (5) and (6). Removal negative correlation with precipitation depth (Kendall τ of −0.27;

Fig. 2. Reduction in peak flow as a function of (a) precipitation depth; (b) inflow depth (inflow volume divided by the watershed area) (62 events)

© ASCE 04016005-4 J. Environ. Eng.

J. Environ. Eng., 04016005


p-value ¼ 0.002). Fig. 2(b) shows that the percentage of reduc- and duration of 25.2 mm, 6.3 h, and 25.6 mm, 1.9 h, respectively,
tion in peak flow had an even stronger negative correlation with were representative of the design rainfall for water quality improve-
inflow volume (Kendall τ of −0.55; p-value <0.001). The reduc- ment. The mean drawdown time for these two storms is 12.3 h,
tion rates covered almost the entire range of potential values, from which meets the 12-h criterion set for water quality improvement
near 0% for the larger events, for which water flows from the top in Quebec, Canada (Rivard 2011). It should be noted that several
of the perforated riser [Fig. 1(b)], to almost 100% for the smallest studies related to construction sites recommended a minimum
inflow values, for which water barely reaches the lower holes of drawdown time of 24 h for enhanced control of effluent quality
the perforated riser. Moreover, Fig. 3 demonstrates that on aver- (TRCA 2006). The event of June 2, 2012 (37.6 mm depth, 33 h
age, peak flows are significantly reduced for the entire range of duration), representative of the design rainfall event for erosion
storm-to-pond ratios, defined as the ratio of inflow volumes on control, had a 90% drawdown time of 14.25 h. Hence, the ag-
the pond maximum storage capacity (82 m3 ha−1 ). Peak flows are retention pond did not meet the Quebec 24-h criterion for erosion
reduced by more than 50% for small events with storm-to-pond control (Rivard 2011).
Downloaded from ascelibrary.org by University of California, San Diego on 01/14/16. Copyright ASCE. For personal use only; all rights reserved.

ratios of less than 0.5, are relatively stable at about 30% for ratios The average watershed T l value was 3.4 h for storms with a
of 0.5 to 3, and diminish to 13% for larger events (ratios > 3). simple triangular-shaped hyetograph (n ¼ 7). Estimated runoff, de-
Recent research on low impact development (LID) has shown rived from the water level variations of the pond, was influenced by
the importance of volume reduction as a mean of improving interflow and groundwater flows, which are generally slower than
wet-retention pond pollutant removal rates on a mass basis surface runoff. Consequently, the center-of-mass values calculated
(Schueler 2011). As shown in Fig. 3, the ag-retention pond had for inflow hydrographs were shifted toward the end of the events
little or no capacity to reduce inflow volume except for the smallest and thus increased T l . Nevertheless, the observed mean value was
events with storm-to-pond ratios of 0.25 or less. The ag-retention very close to the theoretical T l values calculated for the drainage
pond is therefore more representative of first generation LID basin using the methods of Nash, Capece, and SCS, of 4.60, 3.23,
practices where pollutants removal occurs as the inflow volume and 1.94 h, respectively (Sheridan 1994). For these same events, the
passes through a practice. average T d was 2.7 h (average T d for all events was 3.2 h, n ¼ 62).
With the observed mean lag time (3.4 h) and mean detention time
Detention Capacity (2.7 h), it can thus be concluded that the ag-retention pond is
In addition to their effect on peak flows, an important characteristic capable of nearly doubling the mean time it takes for the runoff
of ag-retention ponds relates to their ability to delay the evacuation wave to reach the outlet.
of water. The mean drawdown times for the ag-retention pond to
release 50, 75, 90, and 100% of the total inflow volume, were 7, 18, Erosion Potential
28, and 42 h, respectively (n ¼ 62). Since the outlet hydrographs Morphogenetic flows also referred to as bankfull discharge are
showed very elongated tails (the last 10% of the total volume were generally considered to have the greatest influence on stream
released in 14 h on average), the category corresponding to the time morphology and erosion processes. These flows can be established
at which 90% of the maximum volume was released has been used by in stream measurements and are generally associated to a return
for comparison with urban design criteria. Two rainfall events, period of 1 to 1.5 years. According to the work of Cappuccitti and
which occurred on June 20, 2010 and July 13, 2010 with depth Page (2000), a T d of 24 h for the runoff volume generated

Fig. 3. Mean reduction (%) of peak flow, water volume and contaminant concentrations as a function of storm-to-pond ratio

© ASCE 04016005-5 J. Environ. Eng.

J. Environ. Eng., 04016005


during a 1-year return event can sufficiently attenuate release Impact on Water Quality
velocities below critical values causing erosion in receiving
Runoff associated with rainfall events is one of the most important
streams. The event of June 2, 2012 (37.6 mm depth, 33 h duration), sources of water contamination. The efficiency of the ag-retention
which presented characteristics approaching the design rainfall as- pond was evaluated for total suspended solids (TSS), total nitrogen
sociated with morphogenetic flow (36 mm depth, 24 h duration, (TN), and total phosphorus (TP) based on effluent water quality,
SCS type II distribution), had a T d value of 2.1 h, which is much event-mean concentrations, and load reductions.
lower than the recommended value of 24 h (Cappuccitti and
Page 2000). Effluent Water Quality
A more formal approach to evaluate the reduction of erosion Fig. 5 shows the presence of threshold values above which outlet
potential is to compute the specific stream power (ω). The exact concentrations are systematically lower than inlet concentrations.
ω threshold beyond which erosional adjustment occurs is unknown. Based on the graphical interpretation of effluent water quality,
According to Brookes (1987), we assumed that an erosional adjust-
Downloaded from ascelibrary.org by University of California, San Diego on 01/14/16. Copyright ASCE. For personal use only; all rights reserved.

these threshold values rounded to the nearest number are 30, 1, and
ment is susceptible to occur when ω exceeds 25 W=m2 . Fig. 4(a) 0.2 mg=L for TSS, TN, and TP respectively. These values are close
illustrates the ω at the outlet of the pond calculated from the peak to the irreducible concentration values of 20–40, 1.9, and
inflows and outflows. During the event of June 2, 2012, ω reached 0.15–0.2 mg=L reported by Schueler and Holland (2000) for
24 W m−2 at the inlet and 16 W m−2 at the outlet. Although not TSS, TN, and TP, respectively.
quite equivalent to the design rainfall, the event of September For TP, outlet concentrations are slightly higher than inlet
21, 2013 (31.5 mm depth, 20 h duration) showed ω values of concentrations in the first half of the distribution, suggesting that
31 W m−2 at the inlet and 18 W m−2 at the outlet. These points phosphorus is generated in the pond. Resuspension of accumulated
are highlighted as filled symbols on Fig. 4(a). The frequency of sediment-bound phosphorus is a well-established process and is
exceedance of the erosion threshold (25 W=m2 ) declines from likely to occur (Braskerud 2002). For the events with higher
27% (17=62) when peak inflows are considered to 11% (7=62) concentrations, outlet concentrations are smaller than inlet concen-
when peak outflows are considered. trations for the three pollutants. However, there is no visible thresh-
The total time when the flow exceeded the ω threshold for the 62 old at which the ag-retention pond is able to maintain outlet
events was about 22 h (88 instantaneous flows at a 15-min time step) concentrations. Hence, for the three pollutants, outflow concentra-
at the inlet and 14 h (56 instantaneous flows) at the outlet; a mean tions continue to increase with increasing influent concentrations.
reduction of 32% that may vary from year to year [Fig. 4(b)]. The At the outlet of the ag-retention pond, the very low risk criteria
ag-retention pond was therefore likely to reduce erosion in the receiv- for suspended solids (<25 mg=L) established by the European
ing stream by reducing the frequency and duration of exceedance of Inland Fisheries Advisory Commission (1965) are exceeded in
the erosion threshold and by decreasing ω below 25 W m−2 during 85% of the events [low risk (25–80 mg=L): 6 events; moderate risk
events representative of morphogenetic flow. (80–400 mg=L): 6 events; high risk (>400 mg=L): 5 events;
The methodology based on specific stream power seems there- Fig. 5(a)]. Effluent quality is therefore likely to have an adverse
fore to be more suited for the evaluation of the erosion reduction impact on fish and their habitats. The TN water quality
potential in the receiving stream compared to the simplified criteria of 0.38 mg=L, recommended for the U.S. ecoregion VIII
methodology based on the prolonged detention time. It should (U.S. EPA 2001), is exceeded in all events. However, most of
be noted that the outflow from the studied pond was the only source the events (7=11) fall between 0.38 and 1 mg=L [Fig. 5(b)], with
of inflow to the receiving stream. In the case of facilities installed the upper limit being considered as representative of over fertilized
in larger drainage basins, the entire flow downstream the structure watersheds in Quebec (Ministère du Développement durable, de
must be used to evaluate specific stream power and the erosion l’Environnement et de la Lutte contre les changements climatiques
reduction potential. 2015). In all cases, the TP concentration at the outlet of the

Fig. 4. (a) Specific stream power as a function of precipitation depth; (b) total time (sampled events only) when flow exceeded the erosion threshold
(25 W=m2 ) for each year

© ASCE 04016005-6 J. Environ. Eng.

J. Environ. Eng., 04016005


Downloaded from ascelibrary.org by University of California, San Diego on 01/14/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Empirical cumulative distribution for inlet and outlet concentrations for (a) TSS; (b) TN; (c) TP (n ¼ 20, 11, 14, respectively); dashed lines
delimit environmental criteria mentioned in the text

ag-retention pond exceeds 30 μg=L, which corresponds to Quebec inlet (4.71 and 6.45 mg=L) showed high EMC-based removal
TP criterion for the prevention of eutrophication (Gangbazo et al. efficiency ratios (i.e., 82% and 90%, respectively). For the other
2005). The outflows show concentrations similar to those found in events, excluding one event with an EEMC of −38% (mean TN
eutrophic (35–100 μg=L) to hypereutrophic waters (>100 μg=L) concentrations at the inlet and outlet of 2.20 to 3.03 mg=L, respec-
[Fig. 5(c)]. Note that the TP criterion for the prevention of eutrophi- tively), the EEMC had a value situated between þ25 and −25%. For
cation is lower than the estimated concentration of about 0.2 mg=L TP, the six events with the highest mean inflow concentrations,
below which the pond has no effect. It is therefore unlikely that the ranging from 0.28 to 1.49 mg=L, all gave EMC-based removal ef-
environmental criterion, which was established for rivers and ficiency ratios greater than 30%. The three events with mean inflow
streams, will be met at the outlet of the pond. With respect to concentrations ranging from 0.28 to 1 mg=L gave a mean EEMC of
the observed exceedances, no assumption was made to account 40%, whereas the three events with mean inflow concentrations
for duration of exposure and dilution caused by mixing with greater than 1 mg=L gave a mean EEMC of 78%. For seven of the
receiving water. eight events with mean inflow concentrations below 0.18 mg=L,
the EEMC was lower than 30% and was negative in most cases.
Removal Efficiencies Therefore, mean EMC-based removal efficiency ratios of 42%
Mean removal efficiency ratios (asterisks in Fig. 6) based on event- for TN and 48% for TP, both estimated from all the events taken
mean concentrations (EMC) and event-total loads (ETL) were 50 together, should be interpreted with caution.
and 56% for TSS (n ¼ 20), 42 and 52% for TN (n ¼ 11), and 48
and 59% for TP (n ¼ 14), respectively. Overall, computation of Treatment Mechanisms
total inflow and outflow loads from sampled events for the entire Fig. 3 shows the importance of storm-to-pond ratios on TSS and
study duration shows a reduction of 1.7 T=ha from a total of nutrients removal. For smaller storms with volume reduction, the
3.1 T=ha for TSS, 1.8 kg=ha from a total of 3.5 kg=ha for TN,
and 0.7 kg=ha from a total of 1.1 kg=ha for TP (Table S2, Supple-
mental Data). Removal efficiency ratios were generally higher
when evaluated from total loads compared to mean concentrations,
but the mean differences between the two approaches were
relatively small (Fig. 7). In opposition, Lenhart and Hunt (2011)
obtained strikingly different performance results with mean
variation between approaches of 79.2%, 86.7%, and 47.2%,
respectively, for TSS, TN, and TP (n ¼ 11). For a second-
generation wet-retention pond with high volume reduction capac-
ity, the removal efficiency based on loading seems to be more
representative of the pond performance. However, in this case
[i.e., for a small pond with a negligible volume reduction capacity
(Fig. 3)], the two metric performances were not strikingly different
except for events with small runoff volumes (Fig. 7).
For TSS, although the mean removal efficiency ratio based on
concentrations was 50%, it varied greatly among events, ranging
from 92% to negative values for 5 of the 20 events. Negative
Fig. 6. Tukey boxplots of event-mean concentrations (EMC) and
removal efficiencies occurred for small inflow concentrations
event-total loads (ETL) removal efficiency ratios [Eq. (5)] for TSS,
and/or when the inflow runoff volume is very small (Table S2,
TN and TP (n ¼ 20, 11, 14, respectively); asterisks are the mean
Supplemental Data).
removal efficiency ratios [Eq. (6)] and dashed lines delimit the
In most cases, the inflow concentrations of TN and TP were too
recommended removal efficiency ratios (ETL metric) based on several
small to detect a treatment effect, especially for TN (Fig. 6). That
urban studies from Schueler (2011)
being said, the two events with the highest TN concentrations at the

© ASCE 04016005-7 J. Environ. Eng.

J. Environ. Eng., 04016005


pond, which in turn result in short-circuiting and turbulent flow
instead of laminar conditions. Moreover, water temperature impact
was not considered in this project. Temperature may influence re-
moval efficiencies since it affects viscosity, governing the settling
rates of sediments.
The removal efficiencies for TP are of the same order of
magnitude as those for TSS and point to a strong dominance of
particulate phosphorus relative to dissolved phosphorus. Analyses
of water samples from the experimental site have shown that par-
ticulate phosphorus accounts for more than 80% of the measured
phosphorus concentrations, which support the hypothesis that TP is
mainly retained through settling of sediment-bound phosphorus.
Downloaded from ascelibrary.org by University of California, San Diego on 01/14/16. Copyright ASCE. For personal use only; all rights reserved.

Effluent water quality analyses have demonstrated that TN out-


flow concentrations are lower than inflow values for part of the
distribution (>1 mg=L). According to Tiessen et al. (2011), TN
retention is generally due to removal of nitrate nitrogen (N) through
denitrification and/or plant and algal uptake. These processes are in
turn affected by HRT and water temperature that influence the
Fig. 7. Differences between event-mean concentration (EMC) and speed and intensity of biological activities involved in nutrient
event-total load (ETL) metrics for TSS at various combinations of transformation processes. In opposition to these results, our study
TSS concentration and total inflow volume; the area of the circles has shown that nitrate-N removal was not significant and the two
is proportional to the difference value between the metrics with the strongest TN reductions, which followed land-leveling work in
exception of the two identified groups which would have been too 2012, mainly occurred through ammonia-N reduction. It is there-
small to be showed adequately fore likely that TN reduction during events have happened through
settling of particulate N and that HRT was too short to allow sig-
nificant N transformation and uptake by aquatic plants and algae.
Detailed analysis of nitrate-N, ammonia-N, and water temperature
removal efficiencies are negative, indicating that the sampling would be required to validate the N treatment mechanisms.
possibly started too late, missing part of the inflow loads, and/or
the outflow loads are more representative of water from earlier
storms being displaced from the permanent pool. For intermediate Agricultural versus Urban Removal Efficiencies
storms, with storm-to-pond ratios of 0.5 to 3, TSS treatment Compared to past studies in urban watersheds, higher TSS median
efficiency is relatively stable at 52% but rapidly decrease for larger concentrations and lower TN and TP median concentrations have
events (ratios > 3). For TP, the highest efficiency occurs at storm- been observed in the present study, in contradiction with what
to-pond ratios of 0.5 to 1, while the highest TN efficiency occurs at would be expected (Coulter et al. 2004; Leisenring et al. 2012;
ratios between 2 and 3, for which inflow concentrations where also Wong 2006). In all likelihood, the higher TSS concentrations in
the highest. More data would be required to validate the optimal this study are caused by the land-leveling work that occurred from
design range, but storm-to-pond ratios of 0.5 to 3 are likely to op- late 2011 to the beginning of 2012.
timize TSS and nutrients retention capacity for ag-retention ponds Fig. 6 shows the removal efficiency ratio distribution of the
with similar design, storage capacity, and sediment compositions. ag-retention pond for all events for TSS, TN, and TP and the
For ratios above 3, the high outflow rates through the top of the recommended removal efficiency ratios (ETL metric; dashed lines)
perforated riser limit the hydraulic residence time (HRT) and likely based on several urban studies (Schueler 2011). The removal
impair settling capacity and biological uptake of nutrients. efficiency ratios that were recommended for the Chesapeake
Taking into account the size of the soil particles at the study site, Bay Program were 60, 20, and 45%, respectively, for TSS, TN,
along with the geometry of the ag-retention pond, the surface areas and TP compared to mean removal efficiency ratios of 50 and
required, according to Eq. (7) (U.S. EPA 1976), are 8, 526, and 56% for TSS, 42 and 52% for TN, and 48 and 59% for TP for
8,467 m2 to settle medium sands, medium silts, and clays, whereas the ag-retention pond (asterisks in Fig. 6). The most important dif-
the pond areas at permanent pool, top of the outlet riser, and maxi- ference between the observed removal efficiency ratios from the
mum elevations are 326 m2 , 1,090 m2 , and 5,548 m2 , respectively present study and those of the urban reference was for TN. Recall
that the TN removal efficiency ratio in our study has been obtained
As ¼ 1.2Q=Vs ð7Þ from only 11 storm events and that the relatively high mean re-
moval efficiency ratio obtained was principally caused by only
where As is the pond surface area required to capture particles of two major events with high TN inflow concentrations. In addition,
a given size, V s is the settling velocity for particle size (medium due to the small number of high flow events, the differences
sand 20 × 10−3 m=s; medium silts 0.29 × 10−3 m=s; clay 0.018 × between in and out median concentrations were not statistically
10−3 m=s), Q is the flow generated by a 10-year rainfall event significant for TN and TP (Table 1). More data from agricultural
lasting 6 h (0.127 m3 =s, derived using the SWMM software; studies are needed to determine the level of significance of these
Rossman 2010). differences.
This suggests that the primary treatment mechanism for TSS is Other agricultural studies focusing on different retention
the sedimentation of sands and silts, which comprises the main pro- structures, such as small dams (Rausch and Schreiber 1981; Cooper
portion of the soils particle size distributions. However, coefficient and Knight 1990; Tiessen et al. 2011), stormwater wetlands
1.2 in theoretical Eq. (7) could be increased to account for nonper- (Lenhart and Hunt 2011), water and sediment control basins
manent flow caused by the dynamic nature of water storage during (Mielke 1985), detention ponds (Edwards et al. 1999; Fiener et al.
events (flow variation) and the multitude of entry points in the 2005), and retention ponds (Rushton 2003), have also shown high

© ASCE 04016005-8 J. Environ. Eng.

J. Environ. Eng., 04016005


Table 1. Influent/Effluent Summary Statistics for Pollutant Concentrations (mg=L) in This Study Compared to the Literature
Number of
events 25th percentile Median 75th percentile
Statistical significant
Source Pollutant In Out In Out In Out In Out decrease?a
Leisenring et al. (2012) TSS 725 723 20.7 5.7 70.7 13.5 180.0 33.0 Yes
TN 259 272 1.05 0.86 1.83 1.28 2.87 1.78 Yes
TP 657 654 0.15 0.06 0.30 0.13 0.53 0.23 Yes
Present study TSS 20 20 80.2 38.8 270.4 129.2 573.4 307.7 Yes
TN 11 11 0.73 0.65 0.90 0.85 1.99 1.03 No
TP 14 14 0.12 0.13 0.15 0.16 0.70 0.32 No
a
For Leisenring et al. (2012) results, statistical significance was evaluated using the bootstrap method (Efron and Tibshirani 1993). See Leisenring et al. (2012)
Downloaded from ascelibrary.org by University of California, San Diego on 01/14/16. Copyright ASCE. For personal use only; all rights reserved.

for details. In the present study, the Wilcoxon signed-rank test was applied on the ranks (95% level).

removal efficiency ratios for water contaminants but individual ratios were relatively small, mainly because the small ag-retention
efficiency values varied greatly. Given the broad range of removal pond did not reduce significantly the total flow volume.
efficiency ratio values, a similar approach used by Leisenring et al. • Despite the fact that the storage capacity of the ag-retention
(2012) to determine categorical removal efficiency ratios from pond was smaller than recommended values for urban retention
multiple studies should be used for ag-retention ponds. Agricultural ponds, the mean removal efficiency ratios were similar.
efficiency studies should also include detailed analysis of effluent • The primary treatment mechanism of the ag-retention pond is
water quality. Terminology standardization should also be most likely the sedimentation of sand and silt particles and
conducted to better classify agricultural retention structures and particulate nutrients. Detailed analysis of nitrate-N, ammonia-N,
determine their categorical performance. The following categories and water temperature would be required to validate the N treat-
of agricultural retention structures are proposed: (1) small ag-dams ment mechanisms.
for in stream retention structures with continuous inflows; These findings indicate that ag-retention ponds are an efficient
(2) ag-wetland for stormwater structures with aquaphytic plants method for reducing peak flows at farm level and for improving
or artificial aquatic ecosystem; (3) ag-detention pond for storm- runoff quality. However, more studies are needed to precisely quan-
water ponds that dries completely, and (4) ag-retention pond for tify the environmental impact of ag-retention ponds. In order to
ponds with a permanent pool of water. improve the environmental performance of the agriculture sector,
such ponds should be established in conjunction with conservation
tillage practices that reduce soil erosion and should be implemented
Conclusions in an integrated manner at the watershed scale.
Ag-retention ponds are designed to store stormwaters temporarily
in order to reduce peak flows downstream and to permit settling out
Acknowledgments
of contaminants. This type of best management practice has been
used in urban settings for a number of years, but very few have This research project was funded by Agriculture and Agri-Food
been constructed in agricultural areas. This study focused on an Canada and site construction was co-funded by Québec minister
ag-retention pond constructed in Saint-Samuel (Québec, Canada) of Agriculture, des Pêcheries et de l’Alimentation, the municipality
in 2008. Detailed analyses during the snow-free period from of Saint-Samuel (Québec) and Ferme Bergeroy S.E.N.C. We would
2009 to 2013 showed the following: like to thank Nadia Goussard, Victor Savoie, Camille Desmarais,
• Peak flows from 62 precipitation events were reduced by 38% Claudia Doucet, Claude Bergeron, Guylaine Bergeron, René
on average. Bergeron, and Mayor René Mongrain for their valuable contribu-
• The mean drawdown time for 90% of the event volume was tions. The authors would also like to thank the two anonymous re-
28 h. Drawdown times evaluated for events representative of viewers for their constructive comments.
the design rainfalls suggested that the ag-retention pond meets
the recommendations for the protection of water quality,
but does not meet the criteria for the prevention of erosion in Supplemental Data
a receiving stream.
• The ag-retention pond was likely to reduce erosion in the Tables S1 and S2 are available online in the ASCE Library (http://
receiving stream by reducing the frequency and duration of www.ascelibrary.org).
exceedance of the erosion threshold and by decreasing specific
stream power below 25 W=m2 during events representative of
bankfull discharge. References
• Effluent water quality analysis revealed the presence of
threshold concentrations below which treatment is unlikely to Agrométéo Québec. (2015). “Atlas agroclimatique du Québec.” 〈http://
occur. These concentrations were about 30, 1,and 0.2 mg=L, www.agrometeo.org/index.php/atlas〉 (Feb. 16, 2015).
for TSS, TN, and TP respectively. Comparable irreducible APHA (American Public Health Association). (1995). “Water environment
concentrations values were found in the literature for urban federation.” Standard methods for the examination of water and
wastewater, Washington, DC, 19.
wet-retention ponds (Schueler and Holland 2000).
Bagnold, R. (1966). “An approach to the sediment transport problem from
• Mean removal efficiency ratios based on event-mean concentra- general physics.” U.S. Geological Survey, U.S. Government Printing
tions (EMC) and event-total loads (ETL) were 50–56%, 42–52%, Office, Washington, DC, 231–291.
and 48–59% for TSS, TN, and TP, respectively (n ¼ 20, 11, 14). Braskerud, B. C. (2002). “Design considerations for increased sedimenta-
Each metric varied greatly between events. Overall, the differ- tion in small wetlands treating agricultural runoff.” Water Sci. Technol.,
ences between EMC-based and ETL-based removal efficiency 45(9), 77–85.

© ASCE 04016005-9 J. Environ. Eng.

J. Environ. Eng., 04016005


Brookes, A. (1987). “River channel adjustments downstream from rivières et petits cours d’eau.” 〈http://www.mddelcc.gouv.qc.ca/eau/
channelization works in England and Wales.” Earth Surf. Processes eco_aqua/rivieres/annexes.htm〉 (Jun. 4, 2015) (in French).
Landforms, 12(4), 337–351. Murphy, J., and Riley, J. P. (1962). “A modified single solution method for
Cappuccitti, D. J., and Page, W. E. (2000). “Stream response to stormwater the determination of phosphate in natural waters.” Anal. Chim. Acta,
management best management practices in Maryland.” Maryland Dept. 27(1), 31–36.
of the Environment, U.S. Environmental Protection Agency, Baltimore, Pellerin, A., Parent, L.-É., Fortin, J., Tremblay, C., Khiari, L., and Giroux,
319. M. (2006). “Environmental Mehlich-III soil phosphorus saturation in-
Carpenter, S. R., Caraco, N. F., Correll, D. L., Howarth, R. W., Sharpley, dices for Quebec acid to near neutral mineral soils varying in texture
A. N., and Smith, V. H. (1998). “Nonpoint pollution of surface waters and genesis.” Can. J. Soil Sci., 86(4), 711–723.
with phosphorus and nitrogen.” Ecol. Appl., 8(3), 559–568. Puckett, L. J. (1995). “Identifying the major sources of nutrient water
Choinière, L., and Laplante, L. (1948). “Étude des sols du comté de Nicolet.” pollution.” Environ. Sci. Technol., 29(9), 408A–414A.
Ministère Provincial De l’Agriculture, Ste-Anne-De-La-Pocatière, QC, Rausch, D., and Schreiber, J. (1981). “Sediment and nutrient trap efficiency
Canada (in French).
of a small flood-detention reservoir.” J. Environ. Qual., 10(3), 288–293.
Downloaded from ascelibrary.org by University of California, San Diego on 01/14/16. Copyright ASCE. For personal use only; all rights reserved.

Clary, J., Wildfire, L., and Lord, B. (2012). “Agricultural best management
Rivard, G. (2011). “Guide de gestion des eaux pluviales.” 〈http://www
practices (BMP) database phase 1 literature review.” Wright Water
.mddep.gouv.qc.ca/eau/pluviales/guide.htm〉 (Oct. 28, 2014) (in
Engineers and Geosyntec Consultants, Denver.
French).
Cooper, C. M., and Knight, S. S. (1990). “Nutrient trapping efficiency
of a small sediment detention reservoir.” Agric. Water Manage., 18(2), Rossman, L. A. (2010). “Storm water management model user’s manual
149–158. version 5.0.” EPA/600/R-05/040, National Risk Management Research
Corriveau, J., van Bochove, E., Bégin, G., and Cluis, D. (2008). “Effect of Laboratory, U.S. Environmental Protection Agency, Cincinnati.
preservation techniques on the determination of nitrite in freshwater Rushton, B. (2003). “Runoff characteristics from row crop farming in
samples.” Water Air Soil Pollut., 193(1–4), 335–342. Florida.” World Water and Environmental Resources Congress 2003,
Coulter, C. B., Kolka, R. K., and Thompson, J. A. (2004). “Water quality in ASCE, Reston, VA.
agricultural, urban, and mixed land use watersheds.” J. Am. Water Schueler, T. (2011). “Nutrient accounting methods to document local
Resour. Assoc., 40(6), 1593–1601. stormwater load reductions in the Chesapeake Bay watershed.” Tech-
Edwards, C., Shannon, R., and Jarrett, A. (1999). “Sedimentation basin nical Bulletin 9, Chesapeake Stormwater Network, Ellicott City, MD.
retention efficiencies for sediment, nitrogen, and phosphorus from Schueler, T., and Holland, H. (2000). “Article 65: Irreducible pollutant
simulated agricultural runoff.” Trans. ASAE, 42(2), 403–409. concentrations discharged from stormwater practices.” The practice of
Efron, B., and Tibshirani, R. J. (1993). An introduction to the bootstrap, watershed protection, Center for Watershed Protection, Ellicott City,
Vol. 57, Chapman & Hall/CRC, New York. MD, 377–380.
ESRI. (2012). “ArcGIS desktop: Release 10.1.” Redlands, CA. Sheridan, J. (1994). “Hydrograph time parameters for flatland watersheds.”
European Inland Fisheries Advisory Commission. (1965). “Water quality Trans. ASAE, 37(1), 103–113.
criteria for European freshwater fish: Report on finely divided solids Stormwater Assessment Monitoring and Performance. (2005). “Synthesis
and inland fisheries.” Int. J. Air Water Pollut., 9(3), 151–158. of monitoring studies conducted under the SWAMP program.” Toronto
Fiener, P., Auerswald, K., and Weigand, S. (2005). “Managing erosion and and Region Conservation Authority, Toronto.
water quality in agricultural watersheds by small detention ponds.” Strecker, E. W., Quigley, M. M., Urbonas, B. R., Jones, J. E., and Clary,
Agric. Ecosyst. Environ., 110(3), 132–142. J. K. (2001). “Determining urban storm water BMP effectiveness.”
Gangbazo, G., Roy, J. and Le Page, A. (2005). “Capacité de support des J. Water Resour. Plann. Manage., 10.1061/(ASCE)0733-9496(2001)
activités agricoles par les rivières: Le cas du phosphore total.” Ministère 127:3(144), 144–149.
du Développement durable, de l’Environnement et des Parcs, Direction Tiessen, K., Elliott, J., Stainton, M., Yarotski, J., Flaten, D., and Lobb, D.
des politiques en milieu terrestre, Québec, QC, Canada, 28 (in French).
(2011). “The effectiveness of small-scale headwater storage dams and
Lehmann, E. L. (1998). Nonparametrics: Statistical methods based on
reservoirs on stream water quality and quantity in the Canadian
ranks, Prentice Hall, Upper Saddle River, NJ.
Prairies.” J. Soil Water Conserv., 66(3), 158–171.
Leisenring, M., Clary, J., and Hobson, P. (2012). “International stormwater
TRCA (Toronto and Region Conservation Authority). (2006). “Evaluation
best management practices (BMP) database pollutant category sum-
of design criteria for construction sediment control ponds-Markham,
mary statistical addendum: TSS, bacteria, nutrients, and metals.” Wright
Water Engineers and Geosyntec Consultants, Denver. Ontario.” Vaughan, ON, Canada.
Lenhart, H. A., and Hunt, W. F., III. (2011). “Evaluating four storm-water U.S. EPA (Environmental Protection Agency). (1976). “Erosion and sedi-
performance metrics with a North Carolina coastal plain storm-water ment control: Surface mining in the eastern U.S.” EPA/625/3-76/006a.”
wetland.” J. Environ. Eng., 10.1061/(ASCE)EE.1943-7870.0000307, Washington, DC.
155–162. U.S. EPA (Environmental Protection Agency). (1996). “Nonpoint source
Mallin, M. A., Ensign, S. H., Wheeler, T. L., and Mayes, D. B. (2002). pollution: The nation’s largest water quality problem.” Washington, DC.
“Pollutant removal efficacy of three wet detention ponds.” J. Environ. U.S. EPA (Environmental Protection Agency). (2001). “Ambient water
Qual., 31(2), 654–660. quality criteria recommendations, information supporting the develop-
Mielke, L. (1985). “Performance of water and sediment control basins in ment of state and tribal nutrient criteria, rivers and streams in nutrient
northeastern Nebraska.” J. Soil Water Conserv., 40(6), 524–528. ecoregion VIII.” EPA-822-B-01-015, Washington, DC.
Ministère du Développement durable, de l’Environnement et de la Lutte Wong, T. (2006). “Australian runoff quality: A guide to water sensitive
contre les changements climatiques. (2015). “Suivi de la qualité des urban design.” Engineers Media, Crows Nest, NSW, Australia.

© ASCE 04016005-10 J. Environ. Eng.

J. Environ. Eng., 04016005

You might also like