7

Click here to load reader

Marsh-pond-marsh constructed wetland design analysis for swine lagoon wastewater treatment

Embed Size (px)

Citation preview

Page 1: Marsh-pond-marsh constructed wetland design analysis for swine lagoon wastewater treatment

Ecological Engineering 23 (2004) 127–133

Marsh-pond-marsh constructed wetland design analysisfor swine lagoon wastewater treatment

K.C. Stonea,∗, M.E. Poacha, P.G. Hunta, G.B. Reddyb

a USDA-ARS, 2611 West Lucas Street, Florence, SC 29501, USAb North Carolina A&T State University, Greensboro, NC, USA.

Received 29 March 2004; received in revised form 23 July 2004; accepted 26 July 2004

Abstract

Constructed wetlands have been identified as a potentially important component of animal wastewater treatment systems.Continuous marsh constructed wetlands have been shown to be effective in treating swine lagoon effluent and reducing theland needed for terminal application. Constructed wetlands have also been used widely in polishing wastewater from municipalsystems. Constructed wetland design for animal wastewater treatment has largely been based on that of municipal systems. Theobjective of this research was to determine if a marsh-pond-marsh wetland system could be described using existing designapproaches used for constructed wetland design. The marsh-pond-marsh wetlands investigated in this study were constructedin 1995 at the North Carolina A&T University research farm near Greensboro, NC. There were six wetland systems (11 m×40 m). The first 10-m was a marsh followed by a 20-m pond section followed by a 10-m marsh planted with bulrushes andc ductionso ions werer botht -N ature andcP

K

1

e

f

tely6haterd ordlyhas

ment

0

attails. The wetlands were effective in treating nitrogen with mean total nitrogen and ammonia-N concentration ref approximately 30%; however, they were not as effective in the treatment of phosphorus (8%). Outflow concentrateasonably correlated (r2 ≥ 0.86 andr2 ≥ 0.83, respectively) to inflow concentrations and hydraulic loading rates forotal N and ammonia-N. The calculated first-order plug-flow kinetics model rate constants (K20) for total N and ammonia

(3.7–4.5 m/day and 4.2–4.5 m/day, respectively) were considerably lower than those reported in the limited literurrently recommended for use in constructed wetland design for animal wastewater treatment.ublished by Elsevier B.V.

eywords:Wetland design analysis; Wastewater treatment; Plug-flow kinetics

. Introduction

Animal production has expanded rapidly during thearly 1990’s in the eastern US. In North Carolina, the

∗ Corresponding author. Tel.: +1 843 669 5203x111;ax: +1 843 669 6970.

E-mail address:[email protected] (K.C. Stone).

number of swine has increased from approxima2.8 million in 1990 to more than 9 million by 199(USDA-NASS, 2004). This rapid expansion of higpopulation animal production has resulted in greamounts of concentrated animal waste to be utilizedisposed of in an efficient and environmentally frienmanner. This rapid animal production expansionexceeded the pace at which new innovative treat

925-8574/$ – see front matter. Published by Elsevier B.V.doi:10.1016/j.ecoleng.2004.07.008

Page 2: Marsh-pond-marsh constructed wetland design analysis for swine lagoon wastewater treatment

128 K.C. Stone et al. / Ecological Engineering 23 (2004) 127–133

systems have been developed, and has resulted in theanimal production industry investigating the adapta-tion of some municipal wastewater treatment tech-nologies. In this study, we evaluated the effectivenessof a marsh-pond-marsh constructed wetland that hasbeen used to treat swine wastewater since 1995. De-sign parameters from these wetlands were calculatedto compare their effectiveness to other wetland sys-tems and aid in the design of future constructed wetlandsystems.

Constructed wetlands have been used for manyyears in municipal wastewater treatment. Since 1989,many constructed wetlands have been installed to in-vestigate their effectiveness in treating animal waste(Payne and Knight, 1997). The technical requirementsfor these wetlands treating animal waste were basedmainly on municipal systems and limited data on an-imal waste systems. Although, constructed wetlandstreating animal wastewater were originally thought tobe able to produce an effluent that could be discharged,concern for the environment and discharge regulationshas mostly precluded this approach. Constructed wet-lands treating animal waste are typically used to reducewastewater spray field nutrient loading. This is an im-portant concern where land for application is limited(Barker and Zublena, 1995).

Preliminary constructed wetland design guidelinesfor animal waste treatment proposed by theUSDANatural Resources Conservation Service (1991)werebased on BOD5 loading to the wetlands (presumptivem uml t-l st 12d wet-l aseda sC h-o eld-t on ap

)p esigna mentw inet-ip y, at out-fl ap-

proach is a function of depth and porosity of the wet-lands.

Kadlec and Knight (1996)refer to their model asthek–C* model. The model incorporates the hydraulicloading rate, concentrations into and out of the wet-lands, and a temperature-based rate constant. They alsoinclude a background concentration parameter (C* ).Their rate constant differs fromReed et al. (1995)inthat depth and porosity are not included in the calcula-tion.

Payne and Knight (1997)compared both theReed etal. (1995)andKadlec and Knight (1996)design meth-ods. They found that theKadlec and Knight (1996)method typically required a greater surface area forthe constructed wetland than theReed et al. (1995)method. The main difference was based on the designdepth of the wetland in theReed et al. (1995)model.Payne and Knight (1997)suggested that if theReed etal. (1995)model were to be used, an initial minimumdepth should be used in order to maximize the surfacearea of the wetland.

The wetlands discussed in this paper were con-structed to treat swine lagoon effluent in 1995.Aspects of their performance have been discussed byReddy et al. (2001)andPoach et al. (2004a, 2004b).The objective of this paper was to determine if themarsh-pond-marsh wetland system could be describedusing existing design approaches used for constructedwetlands.

2

2

oonw linaA rthC intoa otherm sec-t ds rep dS z& d-d ra-te ere

ethod). These guidelines were based on minimevels of BOD5 and ammonia-N exiting in the weand and a recommended residence time of at leaays. Updated design guidelines for constructed

ands based on research findings and a physically bpproach were released byUSDA Natural Resourceonservation Service (2002). They proposed two metds: a modified presumptive method and a new fi

est method. The new field-test method was basedhysical approach byKadlec and Knight (1996).

Kadlec and Knight (1996)andReed et al. (1995resented physically based constructed wetland dpproaches based on municipal wastewater treatetlands. Both models are based on a first-order k

cs area-based uptake model.Reed et al. (1995)incor-orated flow rate, wetland depth, wetland porosit

emperature-based rate constant, and inflow andow concentrations. The rate constant used in this

. Methods

.1. Site description and operation

In 1995, six wetland systems to treat swine lagastewater were constructed at the North Caro&T State University farm near Greensboro, Noarolina. The wetland systems were configuredmarsh section, a central pond section, and anarsh section (marsh-pond-marsh). The marsh

ions were approximately 10 m× 10 m and the ponection was 10 m× 20 m. The marsh sections welanted withTypha latifoliaL. (broadleaf cattail) anchoenoplectus americanus(Pers.) Volkart ex SchinR. Keller (American bulrush) in March 1996. A

itional details on construction, initial wetland opeion and initial data analysis were reported byReddyt al. (2001). In September 2000, the wetlands w

Page 3: Marsh-pond-marsh constructed wetland design analysis for swine lagoon wastewater treatment

K.C. Stone et al. / Ecological Engineering 23 (2004) 127–133 129

reconfigured to allow each wetland system to be loadedwith a specific total-N (TN) loading rate from 10 to50 kg/(ha day) (5, 14, 23, 32, 41, and 50 kg/(ha day)).For approximately 1 year, the wetland hydraulic load-ing rates were held approximately equal, with only theTN loading rate varying. The operating depths of theconstructed wetlands were 15 cm for marsh sectionsand 75 cm for pond sections.

2.2. Statistical analyses

Statistical analyses on the constructed wetland datawere performed using the Statistical Analysis Systemsoftware (SAS, 1990). Nutrient concentration reduc-tions were calculated as:

Creduction= Cin − Cout (1)

whereCout is the outflow concentration (mg/L) andCinis the inflow concentration (mg/L).

Percentage nutrient concentration reductions werecalculated as:

%Creduction= Cin − Cout

Cin× 100 (2)

2.3. Regression analysis

A regression analysis was performed to determineif significant relationships existed between inflow andoutflow concentrations to the wetlands. The regressione tra-t ulicl

C

wc

re-g oce-d

l

T ver-a allyc d tom nal-y ewa-t odels

published byKnight et al. (2000). Their study includedsummary data from various constructed wetlands treat-ing dairy, cattle, swine, poultry, catfish pond water, andrunoff from cattle feeding operations. Their data, al-though extremely important and on a wide-ranging va-riety of systems, did not provide an extensive study norquantification of wetlands performance and operationthat are provided in this study. This comparison wasuseful because it combined the strength of both scalesof study.

2.4. Wetland design analysis

Design of surface flow wetlands for animal wastetreatment was originally derived from municipal treat-ment wetlands (Kadlec and Knight, 1996). Surface flowtreatment wetlands typically have nutrient concentra-tion profiles that decrease exponentially with distancefrom the inlet (Knight et al., 2000). This exponentialdecrease in nutrient concentration through the wetlandis generally modeled as a simple first-order reaction.The first-order reaction model is typically integratedwith a plug-flow assumption (Kadlec and Knight, 1996;Reed et al., 1995). Although the flow in constructedwetlands is generally intermediate between plug-flowand completely mixed, the first-order model with plug-flow assumptions provides a conservative design esti-mate (Knight et al., 2000). Kadlec and Knight (1996)presented the area-based first-order plug-flow designmodel as:

l

w /L)a ture(

K

Ks e(

q

ws

quation was modeled to predict outflow concenion as a function of inflow concentration and hydraoading rate and took the form of:

out = aCbinqc (3)

hereq is the hydraulic loading rate (m/day) anda, b,are the regression coefficients.Eq. (3) was transformed in order to perform the

ression in the SAS system with the Proc Reg prure and was analyzed as:

n Cout = ln a + b ln Cin + c ln q (4)

he regression models provide information on the oll performance of the wetlands, but they are typiconsidered valid only for the range of data useodel them. To determine how our regression a

sis relates to other wetlands treating animal waster, we compared our results to those regression m

n

[Cout − C∗

Cin − C∗

]= −KT

q(5)

here C* is the background concentration (mgndKT is the rate constant adjusted for temperam/day).

T = K20θ(T−200) (6)

20 is the rate constant at 20◦C (m/day),θ is dimen-ionless temperature coefficient, andT is temperatur◦C).

The hydraulic loading rate (q) is defined as

= Qin

A(7)

hereQin is the inflow (m3/day) andA is the wetlandurface area (m2).

Page 4: Marsh-pond-marsh constructed wetland design analysis for swine lagoon wastewater treatment

130 K.C. Stone et al. / Ecological Engineering 23 (2004) 127–133

We rearranged Eq.(5) to solve for the temperature-related rate constant for TN, ammonia-N (NH4-N), andtotal phosphorus (TP) from the wetland data as:

KT = Q

Aln

[Cin − C∗

Cout − C∗

](8)

Eq. (6) was then rearranged in order to calculatetheK20 rate constant at 20◦C and the dimensionlesstemperature coefficient.

ln KT = ln K20 + (T − 20) lnθ (9)

where the lnKT would be regressed against the tem-perature term (T− 20).

In addition to solving for rate constants (K20, θ,andC* ) using regression analysis in SAS, we used theSolver spreadsheet function in Microsoft Excel 2000(Microsoft Corporation, Redmond, WA) to simultane-ously solve Eqs.(5) and(6) for K20, θ, andC* . Thisrequired an Excel spreadsheet to be constructed withcolumns ofCin, Cout, q, mean monthly temperature,initial estimates ofK20, C* , θ, estimatedCout, and thesum square error (SSE) term for the difference betweenobserved and estimatedCout. The Solver routine thenminimized the total SSE term by changing the esti-matedK20, C* , and θ values. This simultaneous so-lution method minimizes the sum of squares betweenthe measured and predicted outflow nutrient concen-trations (Kadlec, 2000, personal communication). Weused this Excel spreadsheet procedure for TN, NH4-N,a= anduc

3

toS s hada withaT ay)a 7t riedf uip-m as1 di-

Table 1Means of flow, residence time, and hydraulic loading rate for theconstructed wetland systems

Flow (m3/day) Nominal residencetime (day)

Hydraulic loadingrate (m/day)

Mean S.D. Mean S.D. Mean S.D.

6.38 1.83 16.40 12.70 0.02 0.01

vidual wetland systems (Table 2). Corresponding out-flow TN concentration was 70 mg/L and varied from∼20 to∼112 mg/L across the wetland systems. Over-all concentration reduction was 35% and ranged from28 to 43%.

The mean NH4-N loading rate was 17 kg/(ha day)and ranged from 3 to 30 kg/(ha day) for the individualwetland systems. Mean inflow NH4-N concentrationwas 86 mg/L and ranged from 18 at the lower load-ing rate to 140 mg/L at the higher loading rates. Out-flow NH4-N concentration was 53 mg/L and rangedfrom 10 to 94 mg/L. The overall NH4-N concentrationreduction was 25%. Ammonia volatilization from thewetlands was identified to be a potential significant re-moval pathway at higher loading rates (Poach et al.,2002, 2004a, 2004b) particularly in the pond sectionof the marsh-pond-marsh wetlands.

The mean TP loading rate was 12 kg/(ha day) andranged from∼8 to ∼15 kg/(ha day) at the higher TNloading rates. The mean TP concentration entering thewetlands was 56 mg/L and ranged from∼40 at thelower loading rate to∼68 mg/L at the higher load-ing rates. Mean outlet TP concentration (48 mg/L) wasclose to the inlet concentration for most wetland sys-tems ranging from 34 to 62 mg/L with concentrationreductions ranging from−1 to 16% with a mean re-duction of 8%. To accomplish more efficient P removalin the wetland systems, pre- and/or post-treatment willlikely be required.

TM con-s

TNT

nd TP. Additionally, we calculatedK20 andθ usingC*

0 to compare with our SAS regression results,singC* values estimated fromKnight et al. (2000)toompare with their results.

. Results

During the study period from September 2000eptember 2001, the marsh-pond-marsh wetlandmean residence time of approximately 16 dayshydraulic loading rate of∼0.02 m/day (Table 1).

he overall mean TN loading rate was 23 kg/(ha dnd the individual wetland systems ranged from

o 39 kg/(ha day). The actual TN loading rates varom target rates due to rainfall and occasional eqent malfunctions. Mean inflow TN concentration w16 mg/L and ranged from 42 to 173 mg/L for the in

able 2eans of inflow, outflow, removal, and percent removal for the

tructed wetland systems

Inflow (mg/L) Outflow (mg/L) Percent reduction(mg/L)

Mean S.D. Mean S.D. Mean

N 116 63 70 40 35H4-N 86 56 53 37 25P 56 28 48 15 8

Page 5: Marsh-pond-marsh constructed wetland design analysis for swine lagoon wastewater treatment

K.C. Stone et al. / Ecological Engineering 23 (2004) 127–133 131

Fig. 1. Relationship between total nitrogen mass loading and out-let concentration. Equations plotted with mean loading rate ofq =0.02 m/day for comparison.

3.1. Regression analysis

The coefficient of determination (r2) for the re-gression of TN outlet concentration as a function ofinlet concentration and flow was 0.86 for the marsh-pond-marsh wetland systems. (Fig. 1). For comparison,we plotted our calculated regression equation with themean hydraulic loading rate of 0.02 m/day along withthose fromKnight et al. (2000)(Fig. 1). Our regres-sion results predicted less treatment over the range ofloading rates than those ofKnight et al. (2000).

The regression for NH4-N hadr2 values was 0.83(Fig. 2). Our NH4-N regression results were very simi-lar to those from TN, which was expected since most ofthe TN is in the NH4-N form. Our wetlands predictedlower NH4-N treatment thanKnight et al. (2000).Slopes of both the TN and NH4-N regression lines weresimilar to those ofKnight et al. (2000)indicating sim-ilar treatment characteristics, however at a lower treat-ment efficiency.

The TP regression analysis for the inlet concentra-tion and hydraulic loading rate versus the outlet TP con-centration hadr2 values of approximately 0.64 (Fig. 3).For TP, our regression equation predicted less treatmentthanKnight et al. (2000), particularly at lower loadingrates.

3.2. Wetland design analysis

The wetland data for the entire study period werea

Fig. 2. Relationship between ammonia-N mass loading and out-let concentration. Equations plotted with mean loading rate ofq =0.02 m/day for comparison.

and TP for the six marsh-pond-marsh wetland systems.The temperature-based rate constants were calculatedusing Eq.(8) and then regressed against the temper-ature to determine theK20 rate constant andθ fromEq. (9) and using Excel solver. InTables 3 and 4, K20andθ are shown, for TN and NH4-N for the wetlandsystems studied. These results are lower than thosefrom continuous marsh systems reported byReed etal. (1995), Kadlec and Knight (1996)and Knight etal. (2000). The NRCS field-test method (Payne and

Fig. 3. Relationship between total phosphorus mass loading and out-let concentration. Equations plotted with mean loading rate ofq =0.02 m/day for comparison.

nalyzed to calculate the rate constants of TN, NH4-N,
Page 6: Marsh-pond-marsh constructed wetland design analysis for swine lagoon wastewater treatment

132 K.C. Stone et al. / Ecological Engineering 23 (2004) 127–133

Table 3Regression parameters for the calculation of rate constants for thefirst-order area-based treatment design model

Intercept K20

(m/year)Slope θ r2

TN 1.308 3.69 0.024 1.02 0.138NH4-N 1.486 4.20 0.056 1.05 0.349TP 0.084 1.09 −0.003 0.99 0.001

Knight, 1997) suggests using aK20 of 14 m/year forTN and 10 m/year for NH4-N. We calculated TNK20values of 3.7–4.5 m/year and NH4-N K20 values of4.2–4.5 m/year. The TN and NH4-N K20 values aresimilar for our system because most of the TN in oursystem was in the NH4-N form. Our lower values forthe rate constants compared with the NRCS field-testmethod were calculated assuming bothC* = 0 andC* equal to those suggested in the literature (USDANatural Resources Conservation Service, 2002; Knightet al., 2000). The calculated lowerK20 values wouldresult in a more conservative prediction for treatmentin the wetland systems and would result in a largerwetland area. In our regression analysis, we had verylow coefficients of determination, which suggest thatthe rate constants in our systems were not strongly re-lated to temperature. However, theser2 values werehigher than those calculated for a continuous marshwetland system located in the coastal plain of NorthCarolina (Stone et al., 2002). Mean weekly air tem-peratures ranged from−3 to 21◦C during the 1-yearstudy.

The rate constants (K20) for TP ranged from 1.1 to1.7 m/year for the marsh-pond-marsh wetland systemstudied (Tables 3 and 4). These rate constant valueswere much lower than those reported inKadlec and

Table 4Rate constant (K20) and dimensionless temperature coefficient (θ)calculated simultaneously using the excel solver routine to min-i tflowc

T 4N 6T 3

Knight (1996)andReed et al. (1995). Their values fromthe analyzed databases ranged from 2 to 24 m/year witha mean of 12 m/year, andReed et al. (1995)suggesteda value of 10 m/year. Our data were on the lower endof their range of values. The data from this project hada much higher loading rate of TP than many of thosereported in the references.Stone et al. (2002)calculatedsimilar rate constants for a continuous marsh wetland inthe NC coastal plain with similar high TP loading rates.This suggests that an alternative method of phosphorusremoval should be investigated.

The marsh-pond-marsh constructed wetlands wereless efficient in treating swine lagoon wastewater thancontinuous marsh wetlands (Hunt et al., 2002). Poachet al. (2004a)measured ammonia volatilization fromboth the marsh and pond sections of these wetlands.They found that the marsh sections had relatively lowvolatilization while at higher loading rates the pondsections had high rates of ammonia volatilization. Theoriginal intent of the pond section was to enhance ni-trification (Hammer, 1994; Reaves, 1996). Subsequentresearch on marsh-pond-marsh wetlands has found thatmarsh-pond-marsh wetlands did not improve N re-moval compared to continuous marsh wetlands (Poachet al., 2004b).

4. Conclusions

Marsh-pond-marsh constructed wetlands at theN arG eat-m on-s genf HN pec-t

e int duc-t l ofP ouldb

dicto ndh thel rsh-p tured

mize sum of squares between observed and predicted ouoncentrations

C* assumeda C* = 0

K20

(m/year)θ C* (mg/L) K20

(m/year)θ

N 4.45 1.04 10 4.02 1.0H4-N 4.53 1.06 3 4.35 1.0P 1.67 1.03 2 1.61 1.0

a C* assumed fromKnight et al. (2000).

orth Carolina A&T University swine farm nereensboro, North Carolina were evaluated for trent of swine lagoon effluent. Overall, these c

tructed wetlands were effective in treating nitrorom swine lagoon wastewater. Mean TN and N4-

concentration reductions were 35 and 25%, resively.

The constructed wetlands were not very effectivreating phosphorus. Overall TP concentration reion was 8%. To accomplish more efficient remova

in the wetland systems, additional treatment we required either pre- or post-wetland.

The calculated regression equations to preutflow concentration from inflow concentration aydraulic loading rate were lower than those in

iterature. They predict less treatment in the maond-marsh wetlands than those from the literaatabase.

Page 7: Marsh-pond-marsh constructed wetland design analysis for swine lagoon wastewater treatment

K.C. Stone et al. / Ecological Engineering 23 (2004) 127–133 133

Rate constants for the first-order rate equation (k–C*

model) developed byKadlec and Knight (1996)weredetermined for nutrient treatment (TN, NH4-N, and TP)in the marsh-pond-marsh constructed wetlands treatingswine lagoon effluent in eastern North Carolina. Thecalculated rate constants were generally much lowerthan those reported in the limited literature. Use of ourcalculated rate constants and parameters would result ina more conservative design and require a larger wetlandarea.

References

Barker, J.C., Zublena, J.P., 1995. Livestock manure nutrient assess-ment in North Carolina. 98-106. In: Proceedings of the SeventhInternational Symposium on Agricultural Wastes, Chicago, IL,June 18–20, ASAE Pub. 7-95, St. Joseph, MI.

Hammer, D.A., 1994. Guidelines for design, construction, and oper-ation of constructed wetlands for livestock wastewater treatment.In: DuBowy, P.J., Reaves, R.P. (Eds.), Constructed Wetlands forAnimal Waste management. Purdue Research Foundation, WestLafayette, IN, pp. 155–181.

Hunt, P.G., Szogi, A.A., Humenik, F.J., Rice, J.M., Matheny, T.A.,Stone, K.C., 2002. Constructed wetlands for treatment of swinewastewater from an anaerobic lagoon. Trans. ASAE 45 (3),639–647.

Kadlec, R.H., Knight, R.L., 1996. Treatment Wetlands. Lewis Pub-lishers, Boca Raton, FL, p. 893.

Knight, R.L., Payne, V.W.E., Borer, R.E., Clarke Jr., R.A., Pries,J.H., 2000. Constructed wetlands for livestock wastewater man-agement. Ecol. Eng. 15, 41–55.

P imalOp-

nnis

Poach, M.E., Hunt, P.G., Sadler, E.J., Matheny, T.A., Johnson,M.H., Stone, K.C., Humenik, F.J., Rice, J.M., 2002. Ammoniavolatilization from constructed wetlands that treat swine wastew-ater. Trans. ASAE 45 (3), 619–927.

Poach, M.E., Hunt, P.G., Reddy, G.B., Stone, K.C., Matheny, T.A.,Johnson, M.H., Sadler, E.J., 2004a. Ammonia volatilization frommarsh-pond-marsh constructed wetlands treating liquid swinemanure. J. Environ. Qual. 33 (3), 844–851.

Poach M.E., Hunt P.G., Reddy G.B., Stone K.C., Johnson M.H.,Grubbs A., 2004b. Swine wastewater treatment by marsh-pond-marsh constructed wetlands under varying nitrogen loads, Ecol.Eng., in press.

Reaves, R.P., 1996. Constructed wetlands for swine waste treatment.In: Kenimer, A.L., Morton, A.D., Wendler, K. (Eds.), Proceed-ings of the Second National Workshop on Constructed Wetlandsfor Animal Waste Management. The Texas A&M University Sys-tem, College Station, TX, pp. 118–128.

Reddy, G.B., Hunt, P.G., Phillips, R., Stone, K., Grubbs, A.,2001. Treatment of swine wastewater in marsh-pond-marshconstructed wetlands. Water Sci. Technol 44 (11/12), 545–550.

Reed, S.C., Crites, R.W., Middlebrooks, E.J., 1995. Natural Systemsfor Waste Management and Treatment, second ed. McGraw-Hill,NY.

SAS, 1990. SAS version 6.07, SAS Institute, Cary, NC.Stone, K.C., Hunt, P.G., Szogi, A.A., Humenik, F.J., Rice, J.M.,

2002. Constructed wetland design and performance for swinelagoon wastewater treatment. Trans. ASAE 45 (3), 723–730.

USDA-NASS, 2004. Agricultural Statistics Database (http://www.nass.usda.gov:81/ipedb/). U.S. Department of Agriculture, Na-tional Agricultural Statistics Service, Washington, DC.

USDA Natural Resources Conservation Service, 1991. ConstructedWetlands for Agricultural Wastewater Treatment, Technical Re-quirements, Washington, DC.

U ucted, Na-on-

ayne, V.W.E, Knight R.L., 1997. Constructed wetlands for anwaste treatment, In: Manual on Performance, Design, anderation with Case Histories. Gulf of Mexico Program, SteSpace Center, MS.

SDA Natural Resources Conservation Service 2002. ConstrWetlands, Chapter 3. Part 637 Environmental Engineeringtional Engineering Handbook, USDA Natural Resources Cservation Service, Washington, DC, 73 pp.