ponsored

Research You Can Use

Managing Runoff withConstructed Wetlands

Constructed WetlandsPurdue University research demonstrates howconstructed wetlands can be used to limit runoff.BY Z . J . R E 1 C H E R , E. A. KOHLER,V. L. P O O L E , A N D R. F .TURCO

Runoff from urban areas and golfcourses is automatically presumed

to contribute significantly tonon-point (NPS) water pollutionoriginating from urban areas. Generally,golf course drainage tile lines dischargeto surface water systems, whereas urbanstormwater is managed using direct dis-charge to surface water or temporarystorage in retention basins that eventuallydischarge to surface water.

To better define the role of golfcourses in urban stormwater manage-ment, the 1998 renovation of PurdueUniversity's North Golf Course incor-porated a series of created wetlands thatserve as both water hazards and waterquality management tools. The wetlandsystem was designed to allow golfcourse tile drainage and local urbansurface drainage water to mix and betreated in a series of wetland cells, test-ing the hypothesis that sustainable watermanagement in the urban environmentis possible using managed wetlands on agolf course as a treatment system.

GOLF ANDCOMMUNITY DEVELOPMENTThe popularity of golf has led to theuse of golf courses as centerpieces formany new home development projects.At the same time, these developingurban areas struggle with stormwatermanagement because urbanizationdecreases the amount of permeable sur-face available for absorption and infil-tration of rainwater and snowmelt. Thisincreased runoff can potentially containurban pollution from roofs, roads, and

parking lots that is often carried directlyto surface water. Increasing stormwaterrunoff and velocity magnifies problemsof moving water from one area to an-other, increases storage volume requiredto reduce flooding, and raises theimpact of potential contaminants suchas oils, sediment, and heavy metals.Thus, stormwater management andcleanup have become increasinglyimportant in urbanized areas, but theseefforts are still largely based on the useof retention ponds.

Using created wetlands on golfcourses as water-receiving locationsoffers a unique management and clean-up strategy for both the golf course andthe urban stormwater. Golf courses arehighly managed locations, and the turf-grass receives regular applications ofirrigation water during the growing

season. Even though best-water-man-agement practices may be utilized,some of this water is passed to thedrainage system, and this water is idealfor maintaining wet conditions for basalplant populations in the wetland cells.

Unlike stormwater retention basins,a wetland cell with active plants andanaerobic sediments will have a signifi-cant retention and degradation capacityfor introduced materials. Created wet-lands are able to remove significantamounts of suspended solids, organicmatter, nutrients, heavy metals, traceelements, pesticides, and pathogensthrough chemical, physical, and biologi-cal processes. In other settings, naturaland created wetlands have improved thequality of the water emanating frommunicipal sources, coal mine drainage,aquaculture, and agricultural drainage.

Unlike stormwater retention basins, a wetland cell with active plants and anaerobic sediments will have

a significant retention and degradation capacity for introduced materials.The created wetland system

in the Purdue University research study was efficient at improving water quality.

19

Table 1Water samples were analyzed for the presence of

nutrients, metals, petroleum products, pesticides, and PCBs.

Potential Contaminants Tested

2,4-D Dieldrin PCB aroclor 10162,4-DB Dissolved solids PCB aroclor 1 22 14,4'-DDD Endosulfan PCB aroclor 1 2324,4'-DDE Endosulfan II PCB aroclor 1 2424,4'-DDT Endosulfan sulfate PCB aroclor 1 2482,4,5-TP Endrin PCB aroclor 1 254Aldrin Endrin naldehyde PCB aroclor 1 260Alpha-BHC Ethoprop PendimethalinAlpha-chlordane Fenarimol PhosphorusAluminum Gamma-BHC PotassiumAmmonia Gamma-chlordane ProdiamineAntimony Heptachlor SeleniumArsenic Heptachlor epoxide SilverAtrazine Iron SiliconBarium Lead SimazineBenfluralin Lithium StrontiumBeryllium Magnesium SulfateBeta-BHC Manganese Suspended solidsBoron Malathion ThalliumCalcium MCPA TinChloride MCPP TitaniumChloropyrifos Mercury Total organic carbonChromium Methoxychlor ToxapheneCobalt Metolachlor TriadimefonCopper Molybdenum TrifluralinDelta-BHC Nickel VanadiumDiazinon Nitrate/nitrite ZincDicamba Oil and grease Zirconium

This study was initiated to determinethe chemical characteristics of watermoving through created wetlandsassociated with a commercial 18-holegolf course and a residential area, andtrack changes in water quality throughthe wetland system during storm andnon-storm events. The site for ourstudy was the newly redesigned andrenovated Kampen Course on thecampus of Purdue University. In rede-signing the course, there was consider-able emphasis on minimizing the inputsof potential, but unknown, golf-course-related non-point source pollution toCelery Bog, a highly valued park andrecreational area adjacent to the newcourse. Additionally, with the plannedredesign came the opportunity toaddress untreated runoff from theadjacent urban area that was previouslytiled under the golf course directly intoCelery Bog.

MATERIALS AND METHODSPurdue University's Kampen GolfCourse is part of the Birck BoilermakerGolf Complex located in West Lafayette,Indiana. The course is situated near theheadwaters of the 968-acre Cuppy-McClure watershed, a rapidly urbaniz-ing area of West Lafayette. The golfcourse comprises 69 acres, of which 25acres drain directly into the createdwetlands used in this five-year study.Following treatment in the wetlandcells, the water either flows into CeleryBog or is pumped back into irrigationponds used on the course. The areaadjacent to the northeast side of thegolf course is urbanized and includestwo residential highways, a motel andparking lot, gas station, and approxi-mately 200 homes.

Initial construction of the redesignedcourse and wetlands was completed inearly 1998, wetland plants were installed,

and the course opened in June 1998.The cells were mechanically cleared ofall existing vegetation, packed, andre-vegetated with 10,800 plants thatincluded (scientific name and numberused):1. Arrowhead (Sagittaria spp., 300)2. Banded lake sedge (Carex lacustris,

100)3. Burreed (Sparganium americanum,

200)4. Creeping spikerush (Eleocharis fallax

Weatherby 100)5. Crested sedge (Carex cristatella, 500)6. Harlequin blueflag (Iris versicolor L.,

500)7. Lake sedge (Carex lacustris Willd.,

500)8. Lurid sedge (Carex lurida, 500)9. Pickerelweed (Pontederia cordata L.,

750)10. Prairie cordgrass (Spartina pectinata

Bosc ex Link, 300)11. Piver bulrush (Sdrpus fluviatilis

[Torrey] Gray, 250)12. Soft rush (Juncus effusus L. 1,500)13. Softstem bulrush (Schoenoplectus

tabernaemontani [K. C. Gmel.]Palla,3,100)

14. Sweet flag (Acorus gramineus Sol. exAlton grassleaf, 2,000)

15. Three square bulrush (Sdrpuspungens Vahl, 100)

16. White water lilies (Nymphaea alba,100)

17. Woolgrass (Sdrpus cyperinus (L.)Kunth, 200)

18. Yellow pond lilies (Nupharpolysepalem, 100)

Water flowing into the wetlandsystem comes from a number ofsources. During golf operations, Aprilto November, water enters the wetlandas part of the irrigation recovery system.Water also enters the wetlands as urbanrunoff from the adjacent areas. Themixed water then enters the constructedwetlands series of cells.The wetlandcells contain water all year, but theconstructed wetland will not dischargewater to the adjacent natural systemexcept under high flow (storm) condi-tions. During the golf season, wetland

20 G R E E N S E C T I O N R E C O R D

water is returned to the course from theirrigation pond using the irrigationsystem.

Four sites on the golf course and thewatershed outlet were chosen for watersampling in this work. Sampling loca-tions were selected to track the water asit progressed through the system, enteredthe eastern edge of the course, movedthrough the wetland system, and exitedthe northwestern edge of the course toCelery Bog or the south side to theirrigation pond. Site 1 (urban input orUI) characterizes urban runoff. Site 2(after wetland one or AWO) character-izes water exiting the first wetland cell.Site 3 (golf course tile or GCT) charac-terizes golf course tile drainage justprior to entering the wetland system.Site 4 (golf course output or GCO)characterizes wTater exiting the con-structed golf course wetlands andentering Celery Bog. Site 5 (watershedoutput or WO) is located at the mouthof the Cuppy-McClure watershed andcharacterizes the overall watershedwater quality and provides a basis ofcomparison of water quality betweenthe watershed and golf/urban discharge.

RESULTS AND DISCUSSION:STORM EVENTSUrban input (UI) was the main sourceof N-NO3/NO2 and N-NH, (Table 2)into the created wetland. Even thoughmore than 16,000 Ibs. of N was appliedto the golf course area that drains intothe wetland (25 acres) during the five-year period when storm events weresampled, discharge of N-NO3/NO2

and N-NH3 from the golf course tilewas minimal (1.10 and 0.25 ppm,respectively) (Table 2). The wetlandefficiently removed an estimated 97%of N-NO3/NO2 and 100% of N-NH,.

The area of the golf course thatdrains into the wetland received 2,033Ib. P during the five-year storm-eventsampling years. Despite this, low levels(<0.5 ppm) of P were detected duringstorm events at all sites (Table 2). Massloading removal of P was 74% duringstorm events.

During storm events, K concen-tration in drainage water increased aswater moved through the wetland(Table 2). Water at the GCO had ahigher K concentration than water ateither the GCT or the UI (Table 2),

Table 2Mean concentration of nutrients and major elements

in the wetland measured during six storm events.

Site*

Parameter UI AWO GCT GCO WOppm

N-NO3/NO2 1.38 0.29 1.10 0.18 0.67N-NH3 2.70 0.42 0.25 0.30 0.60Phosphorus (P) 0.31 0.1 I 0.44 0.44 0.45Potassium (K) 3.35 5.43 5.80 6.41 2.77Chemical O2 demand 294 39 50 34 61Total Organic Carbon (TOC) 106.2 12.5 12.5 9.6 17.3Dissolved solids 335 350 478 280 362Suspended solids 33 47 155 92 228Aluminum (Al) 2.04 1.07 4.09 2.54 1.82Calcium (Ca) 47.83 62.83 92.60 48.40 74.00Chlorine (Cl) 44.77 60.83 100.20 23.60 37.85Iron(Fe) 1.49 1.74 6.15 2.13 3.43Magnesium (Mg) 13.13 24.17 32.20 27.00 19.67Manganese (Mn) 0.37 0.20 0.26 0.22± 0.45Sodium (Na) 20.75 28.83 42.40 8.14 19.77Silicon (Si) 3.53 4.10 13.26 8.06 6.25SO, 26.67 38.67 70.80 50.60 62.33

*UI, urban input; AWO, after wetland one; GCT, golf course tile; GCO, golf course output;WO, watershed outlet. Data are averaged over six storm events.

Aerial photo of Purdue University's KampenGolf Course with numbers indicating watersampling sites.

resulting in an overall mass removalefficiency of 12%. This is similar toother work that found potassium con-centration increases as water passesthrough a wetland and that naturalwetlands often export potassium.

Chemical oxygen demand (COD)and total organic carbon (TOC) werehighest at the UI, which would beexpected with the first flush of a stormpushing organic matter from a residen-tial area (including roads and parkinglots) into the created wetland system(Table 2). However, COD and TOCwere reduced by wetlands during stormevents. Reductions from the UI to theGCO were 90% for COD and 91%for TOC.

During storm events, GCT had thehighest concentration of dissolved andsuspended solids, while the UI had thelowest concentration. Mass loadingremoval of dissolved solids was 59%,indicating that the wetlands were effec-tive at removing dissolved solids duringstorm events. However, mass loadingremoval of suspended solids was 0%in this study. Other researchers foundhigher removal efficiencies of suspendedsolids during storm events. This appar-ent difference could be due to the addi-tional tile lines feeding water directlyinto the third long wetland cell border-

21

Research at Purdue

University demonstrates

that golf course wetlands

are an excellent way of

reducing surface water

pollution from nutrients,

sediments, and many

potential chemical

pollutants.

ing Celery Bog. The tile water camepartially from sand bunkers on thegolf course and tends to be high insuspended solids.

Both frequency and level of pesticidedetection at any sampling location werelow during storm events, and no PCBs(poly-chlorinated biphenyls) werefound. On June 11,1999, atrazine wasdetected at 0.01 ppb at UI and at 0.17ppb at AWO (Site 2), while simazinewas detected at 0.22 ppb at AWO. OnNovember 1,1999, MCPA at 0.56 ppbwas detected at AWO. No pesticideswere detected on the four othersampling dates. None of the threepesticides detected were found at siteslocated on the golf course. The fact thatatrazine was found at the UI, but not atGCO (Site 4), was likely due to thewetland removing atrazine, as previousresearch has shown that wetlands re-move atrazine during storm events.Although the golf course was not

treated with atrazine or simazinedirectly, these triazine herbicides arelikely used in corn production areasthat surround West Lafayette. Atrazinehas been detected in rainwater, so it isnot surprising that it was detected bothentering the golf course and in thewatershed. Despite common use of2,4-D and dicamba on home lawns forbroadleaf weed control in turfgrass,however, no 2,4-D or dicamba wasfound during storm events at any site,including the UI.

No metals such as arsenic (As),cadmium (Cd), chromium (Cr), copper(Cu), lead (Pb), mercury (Hg), orselenium (Se) were detected duringstorm events at any sampling location,despite the urban area having roads andparking lots where heavy metals arelikely to be found. Likewise, oil andgrease were not detected during stormevents at any sampling location, includ-ing the urban input with its close prox-

imity to roads, a gas station, and parkinglot, which potentially can be sources ofpetroleum product pollutants.

The golf course's impact on wetlandwater quality can be summarized bycomparing parameters at the UI andthe GCO. During storm events, 11 ofthe 17 measured parameters (NO3/NO2,NH3, P, COD,TOC, dissolved solids,Ca, Cl, Mg, Mn, and Na) had highermass loading entering the course atthe UI than leaving the golf course atthe GCO. Thus, during storm eventsthe mass of most of the parametersdecreased as water flowed through thewetland system. Furthermore, not allstorm events (June 11,1999, and August23, 2000) were great enough to causedischarge from the largest wetland cellinto Celery Bog, causing all water andany potential contaminants to remainwithin the closed wetland system.

Comparing data at GCO with datafrom the whole watershed outlet (WO)provides an estimate of the impact ofthe golf course within the entire water-shed. A lower concentration for 13 ofthe 17 parameters, except potassium(K), aluminum (Al), magnesium (Mg),and silicon (Si), was found at the GCOthan at WO (Table 2). Therefore, waterexiting the golf course during stormevents is not a major source of contami-nation to the Cuppy-McClure water-shed despite urban runoff inputs andsignificant fertilizer and pesticide inputsused on the golf course.

As for the exceptions, golf coursefertilization is not likely responsible forthe export of Al, Mg, or Si from the golfcourse, but may be the result of erosionor leaching through sand bunkers. Con-versely, 2,033 Ib. K was applied to the25-acre golf course area over five yearsand may have added to the K export.However, previous research has shownthat wetlands often export K, whichmay also have increased K levels in thesystem.

NON-STORM EVENTSConcentration of N-NO3/NO2 andN-NH3 discharged from the GCT

22 G R E E N S E C T I O N R E C O R D

(Site 3) was minimal (<1.0 ppm forN-NO3/NO2 and <0.5 ppm forN-NH3).The wetlands reduced theN-NO3/NO2 concentration by asmuch as 95%. In contrast to N-NCVNO2, there was little change in N-NH3

concentration through the wetlands,suggesting that denitrification waslikely responsible for N-NO3/NO2

reductions. It is reassuring to note thatwhile the golf course applied 5,800 Ib.N to the area that drains into the wet-land during the five-year period whennon-storm events were sampled, theaverage levels of N-NO3/NO2 andN-NH3 in the GCT were only 0.85and 0.31 ppm, respectively (Table 3).

It should be noted that this golfcourse pumps water from the third longwetland cell into a storage pond andthen recycles it to the irrigation system.This irrigation water is applied to thecourse and drains back into the wetlandfor additional treatment, where it iseither redirected to the irrigation pondor to Celery Bog. This system of treat-ing irrigation return flow is ideal fornitrate removal. Our data suggest thereis no buildup of N levels in the wetlanddue to the recirculation of irrigationwater, as N levels detected at all siteson the course were < 1 ppm N-NO3/NO2/NH3, even with fertilizerapplications.

Low levels (<0.5 ppm ) of P weredetected during non-storm events(Table 3), even though the area of thegolf course that drains into the wetlandsduring non-storm sampling years wasfertilized with 450 Ib. P. However, theGCT contributed higher amounts of Pthat were not reduced before reachingthe GCO. Despite higher P concentra-tions at the GCO than at the UI, con-centrations were < 1 ppm at all sites(Table 3). Our results show that thesecreated wetlands were able to removeP from water effluents with < 1 ppmtotal P.

The golf course was not a majorsource of K, nor was the constructedwetland a sink for K (Table 3). Whilethere was a slight reduction in K con-

centration between the GCT andGCO, K concentration remained un-changed as water passed through thewetland system during non-stormevents (Table 3). Other researchers havealso found an increase in K as waterpasses through a wetland.

Trends in chemical oxygen demand(COD) were similar to the results fortotal organic carbon (TOC) trends(Table 3). COD and TOC levels werestable as water flowed through the con-structed wetland system. The GCT hadlower COD and TOC than the UI,which would be expected due to soilfiltering the water before entering thetile lines. However, COD and TOC atGCO increased to near UI levels(Table 3).

Passage through the wetlands reducedthe concentration of dissolved solids byas much as 53% (Table 3).The UI hadthe lowest suspended solid concentra-tion, while the GCT had the highestsuspended solid concentration (Table 3).This may be due to the fact that waterentering the urban area passes through agrassy ditch prior to reaching the UI,

and previous research clearly showsthat vegetative filters are important inremoving suspended solids. The GCTwater has no such type of bio-filter, asmuch of this water enters the tile linesdirectly from erodable sand bunkers.

There was only one instance ofpesticide detection during non-stormevents from any sampling location.During non-storm events, only thedinitroaniline herbicide trifluralin wasdetected at 0.22 ppb on September 28,2001, and was found on the golf courseat AWO. No trifluralin was applied tothe golf course anytime during thestudy, so it is unknown how the chemi-cal arrived on the golf course. It is notsurprising that so few pesticides weredetected in the wetland system, becauseprevious research has shown createdwetlands are able to reduce pesticideconcentrations. Furthermore, all thewetland cells are surrounded by turf,and any pesticides would have beenapplied directly to the turfgrass. Previousresearch on the leaching and runoff ofpesticides applied to turfgrass has shownminimal loss. Thus, vegetative strips

Table 3Mean concentration of nutrients and major elementsin the wetland measured during six non-storm events.

Parameter

N-NO3/NO2

N-NH3

P

K

UI

0.68

0.27

0.153.47

Chemical O2 demand 37TOCDissolved solidsSuspended solidsAlCa

ClFeMg

Mn

NaSiS04

*UI, urban input;

9.652016

0.2889.17

130.170.7523.500.41

81.504.7834.00

AWO, after wetland one;WO, watershed outlet. Data are averagec

AWO

0.36

0.310.09

3.40

437.9462

83

1.7880.3375.002.8828.00

0.4336.676.2043.17

Site*

GCT

ppm

0.85

0.310.174.6719

4.3697249

3.21132.00

138.502.27

54.00.47

66.0011.1375.50

GCT, golf course tile;

GCO

0.040.27

0.183.72

359.8330

142

2.7547.80

332.0325.00

0.2215.47

6.0040.00

WO

0.45

0.350.194.0339

12.5492

412.41

85.00

49.672.3822.000.25

28.507.62

53.33

GCO, golf course output;over six non-storm events.

J U L Y - A U G U S T 2 0 0 5 23

The Kampen Course on the Purdue University campus was the site used to investigate wetland

filtering capabilities.The study determined the chemical characteristics of water moving through

created wetlands associated with an 18-hole golf course and a residential area.

such as those that surround the wetlandcells (and drainage ditch prior to theUI) are effective filters for chemicals insurface runoff. This may explain "whyno 2,4-D or dicamba was detected atany site, including the UI, despite thecommon use of 2,4-D and dicamba onhome lawns for broadleaf weed control.

The heavy metals As, Cd, Cr, Cu, Pb,Hg, or Si were not detected duringnon-storm events from any samplinglocation despite the urban area havingroads and parking lots where heavymetals are likely to be found. Likewise,oil and grease were not detected duringnon-storm events at any sampling loca-tion, including the UI with its closeproximity to roads, a gas station, andparking lot, which can potentially be asource of petroleum-based pollutants.

Comparing measured parameters atthe UI and the GCO can estimate thegolf course's impact on wetland waterquality. During non-storm events, onlyseven of the 17 measured parameters(NO3/NO2, COD, dissolved solids, Ca,Cl, Mg, and Na) had a higher concen-tration in water at the UI than at GCO(Table 3).Thus, during non-stormevents, the concentrations of eight ofthe different parameters increased aswater flowed through the -wetlandsystem. However, the concentrations of

these parameters was well below drink-ing water standards and no dischargeoccurred at the GCO into Celery Bog.Thus, despite increasing concentrationof eight of the 17 parameters, all waterwas contained within the closed-loopedwetland system, resulting in 100% massremoval efficiencies for all parameters.Although there was an increase in con-centration for most parameters duringflow through the golf course wetlandsduring non-storm events, 14 of the 17parameters (except suspended solids, Al,and Mg) were at a lower concentrationat GCO than at WO (Table 3).There-fore, water on the golf course does notrepresent a major source of pollutantsto the Cuppy-McClure watershed.

BOTTOM LINE:CREATED WETLANDS WORKOur study showed that this golf coursedoes not reduce quality of its watercompared to water entering the golfcourse or water in the larger Cuppy-McClure -watershed. The created -wet-land system was efficient at improvingwater quality. Although mass removalefficiency ranged from -182% to 100%during storm events, 9 of 17 parametershad mass removal efficiencies >50% aswater flowed through the wetlandsystem. More importantly, mass removal

efficiencies were 100% during baselineflow conditions due to no flow off thecourse. Therefore, with the combinationof higher mass removal efficiencies andthe lack of flow into Celery Bog dur-ing non-storm events, introduction ofpotential pollutants into the greaterwatershed is highly unlikely duringnormal, day-to-day operation of thegolf course wetland.

Overall, this system demonstratedthat created wetlands on golf courses canbe used to filter golf course tile drains,as well as runoff from areas adjacent tothe course. With the increasing numberof golf courses in urbanized areas, cre-ated wetlands could be used to improveregional water quality while enhancingthe aesthetics of golf courses. However,to insure maximum water qualityimprovement, -wetlands should be sizedsufficiently to maximize water-holdingduring storm events and to minimizeoutputs during non-storm periods.

ACKNOWLEDGEMENTSThis research project would not havebeen possible -without the support ofJim Scott, superintendent of the BirckBoilermaker Golf Complex; the UnitedStates Golf Association; Pete Dye, Inc.;Kevin Tungesvick; Spencer RestorationNursery; and Heritage Environmental.

Editor's Note: For a full-text versionof this report, including nutrient andchemical mass loadings, use URLhttp://usgatero.msu.edu/vQ4/n02.pdfof USGA Turfgrass and Environ-mental Research Online(http://usgatero.msu.edu).

Z. J. REICHER, PH.D., associate professorand turfgrass extension specialist, Depart-ment of Agronomy; E. A. KOHLER, PH.D.,former graduate assistant, Department ofAgronomy; V. L. PooLE,field technician,Department of Forestry and NaturalResources; R. F. TURCO, PH.D.,professorof soil microbiology, Department ofAgronomy, Purdue University,West Lafayette, Indiana.

24 G R E E N S E C T I O N RECORD