o EVALUATION OF NITROGEN TARGETS AND WWTF LOAD REDUCTIONS o FOR THE PRO\TDENCE AND SEEKONK RIVERS o o o o o o O RHODE ISLAIID DEPARTMENT OF ENVIRONMENTAL MANAGEMENT OFFICE OF WATER RESOURCES DECEMBER 2OO4 a EVALUATION OF NITROGEN TARGETS ANID WWTF LOAD REDUCTIONS FOR THE PROVIDENCE A}ID SEEKONK RI\IERS. APPROACHESFOR ESTABLISHINGA NITROGEN LOAD REDUCTION PLAN The Providenceand SeekonkRivers are impacted by low dissolvedoxygon levels and high phytoplanktonconcentrations that are related to excessivenitrogen loadings. DEM has collected data and has been working with a contractorto develop a water quality model and a water quality lestoration plan (Total Maximum Daily Load (TMDL)) for the area.It has recently been determinedthat due to problemsencountered when modeling the interaction betweenthe deepcharurel and shallow flanks ofthese water bodies,the masstransporl componentof the modelsystem cannot be successfullycalibrated and validated. This problemhas been encountered in otherestuaries and has not beenresolved with stateofthe art numericalsolution techniques. Because water doesn't mix in the modelas it doesin the rivers,we areunable to simulatethe chemicaland biological behavior ofthe systemin the waterquality phase of the modelingeffort. Our inability to adequatelyvalidate the masstransport model also preventsus from applying the Massachusettsapproach to settingload allocationsthat usesambient total nitrogen concentrationas the indicator.which is describedbelow. Other elements ofthe Massachusettsapproach were found to be helpfulin the goal-settingphase ofthis discussion, however. When functioning properly, a water quality model predicts an accuratewater quality conditionthat results from a setofinputs (pollutantloadings) to the system.A computer- basednumerical model is typicallyused, however a physicalmodel can also serve as the analog for the river. In this case,information is available from the MERL (Marine EcosystemsResearch Laboratory) enrichment gradient experiment (Oviatt et al, 1986),DEM had initially usedrelationships observed during the MERL experimentto establishkinetic tems in the numericalmodel, however the experimentalresults themselves can also be used. THE MERL EXPERIMENTAND BEHA\'IOR OF KEY PROCESSES The enrichmentgftdient experimentwas conducted continuously between May 1981and September1983. A completeset of environmentalvariables and water column parameters weremeasured throughout the periodofthe study.These included weekly samplingof chlorophyll-a,DIN, DO from consecutivedawn-dusk-dawn measurements, daily production, and monthly benthic uptake. The experimentwas conducted in nine 13,000L tanksat URI. Threecontrol tanks, consistingof lower Bay waterwith no enrichmentcomprised the first group.The next group consistedofthe low treatrnents:1{ 2X, and4X tanks,where the 1x additionrate represented the meanaddition rate fuer unit area)of inorganicnitrogen, phosphorus, and siliconto NarragansettBay. The final groupconsisted ofthe high treatrnents:8X, 16X, and32X additions,where the 8X casewas considered representative of the ProvidenceRiver, and the 32X caserepresented the HudsonRiver regionofNew York Harbor.In general,the 1X- 32X Page1 of 32 o loadinggradient was selectedto reproducethe rangeof enrichmentlevels seen in real a estuaries.The loadinggradients and conesponding DIN loadingrates are listed in Table 1. Table l. MERL Loading Gradients and DIN Loading Rates Loading DIN LoadingRate o Gradient mmoleim2/day ke/m?ldav I 2.88 4.032E-05 z ). /t) 8.064E-05 4 11.52 1.6i3E-04 c 23.04 3226E-04 o 16 46.08 6.451E-04 )z 92.16 1.2908-03 As an initial phaseof herproject to refinethe WASP kineticsfor DEM, Dr. Aimee Keller evaluatedthe MERL results to extract information that would be usedin the proposed o applicationof the WASP modelfor the ProvidenceRiver. Fromt}re mesocosm data, Dr. Keller documenteda number of characteristicsof the MERL tanks that have sienificance for the Providenceand SeekonkRivers: Dissolvedoxygen: The DO observationsfor the three groupsoftanks are shownbelow. o The most significant featuresof DO behavior are that as the nutrient addition level increases,DO minimum levelsdrop precipitously and the variabilityincreases. Figures 1, 2, and 3 summarizeDO for the control tanks (mouth of bay water with no nutrient addition),low treatments(1X, 2X, and4X) andhigh treafrnents(8X, l6X, and 32X), respectively.The three figures show a distinct increasein deviation from the saturation o concentrationshown by the solid red line in eachfigure with increasingaddition rate. Another significant featureis an exponentialdrop in minimum summervalues with increasingnutrient loadings (Figure 4). Summerminimum values for the threehighest enrichmentsare less than 2 mg/I. The meanof obsewedDO concentrationsalso increases somewhat with increasins loading.(Figure 5). Seasonalmaximum values also increase. o r Ternperatureappears to be the principal factor that affectsDO at lower loading rates (Figure6). At higherloadings, DO conelatessignificantly with ambientsilica andnikate concentrations. Figure4 is probablythe mosttelling graphicof this $oup. Daily minimumDO valuesdrop a precipitously,to lessthan 2 mg/I, at the 8X, 16X, and32X tanks.The regressionequation derivedfrom the tank datapredicts DO minima of 3.7 mgll,3.0 mg/I, and 1.98mg/l for the 2X, 4X, and8X cases,respectively, I Page2 of 32 a DissolvedOxygen - Controls MJJASSONDJ FMAMJJAASONDJ FMAMJJJAS 19E1-1983 Figure 1: DO for the control tanks, May 1981- September1983. DissolvedOxygen - LowTreatments 25 20 15 10 0 M J J A S S O N D J F M A M J J A A S O N D J F I,IAM J J J A S 1981-1983 Figure 2: DO for the low level treatment tanks, May 1981- September1983, Page 3 of 32 a o Dissolved Oxygen - High Treatments a mg'- 10 o 0 MJJASSON FMAMJJAASONOJ FMAMJ J o 1981-1983 Figure 3: DO for the high leyel treatment tanks, May 1981- September1983, a OxygenMinimum 1981-1983 o 4 E'3 o I 1 0 40 60 a Nitrogen Load (mmol/m?/d) Figure 4: The relationshipbetween the inorganic nitrogen loading rate and minimum observedoxygen concentration, which showsthat the oxygenminimum drops rapidly with increasingloadings. a Page4 of 32 Avg. Oxygenvs N-Load 11 y=0.27Ln(x)+8.5 10 O_f = 0.+S El O I 8 20 40 60 80 N-Load lmmol m'2d-1) Figure 5: Graph of mean MERL tank DO as a function of loading, showingthat the averageDO goesup slighfly as nutrient loadingsare increased. OxygenConcentration vs Temperatute t0 Tank 0 ' Control t4 t0 a at 1t l.l-' 2 Y34,3ox+11.90 R: = 0.79 0 Figure 6: The relationshipbetween D0 and temperaturein a control tank. Page5 of 32 a Phvtoolankton: o . Figures 7 - 9 show observedchlorophyll levels in the three setsof tanks during the experiment.As with DO, the variabilityas well asthe meanlevels of chlorophyllincrease asnutriurt additionlevels are increased. o The 4X and abovetreatments experience peak chlorophyll levels of 100 ug/l or greater. The variability ofchlorophyll in the 8X andhigher tanks could best be describedas o chaotic(Figure 9). Adjacentweekly measulernents in thesetanks appear to haveswung fiom one extremeto anotheron a weekly basis. The 8X is somewhatdisparate because the tank was inadvertently colonizedby a disparatelyhigh number of filter feedersthat uffealisticallydepressed both the variationsand the meanlevels ofphytoplankton. This result indicatesthat filter feederscan rnitigate the effects of increasednutrient loadings by reducingphytoplankton blooms, however filter feedersexert an oxygendemand o through their respfuationthat may exacerbateh;poxia under someconditions. r The meanlevel ofphytoplanktonincreases linearly with increasingnitrogen loading rate (Figure 10). o Chlorophyll a - Controls o 20 aDt5 10 o 0 o Figure 7: Phytoplankton biomassin the control tanks, May 1981- September1983. a a Page6 of 32 o Chlorophylla - Low Treatments 120 100 80 rD 60 40 20 0 M J J A S S O N D J F M A M J J A A S O N D J F M AM J J J A S 1981.1983 Figure 8: Phytoplanktonbiomass iu the low treatment tanks, May 1981- September 1983. Chlorophylla - HighTreatments 180 r60 --+- i6 x l-,.'r, x 140 t20 _ 100 l'r 580 40 0 M J J A S S O N D J F M AM J J A A S O N D J F M A M J 1981-1983 Figure 9: Phytoplankton biomassin the high treatment tanks, May 1981- September 1983. Page7 of 32 a Avergage Chlorophyll vs N Load o 70 60 a G40 oru o ozo o 10 0 Nitrogen Load (mmol,/m'/d) a Figure l0: Mean phytoplankton biomassas a function of loading from the experiment. At higherloading rates (> 8x), light penetrationdepth decreases in the upperwater columnas a resultofshading by phytoplanktonto thepoint wherephytoplankton growth becomeslimited by light, not nitrogen.In essence,nutrients are over-abundant at these o enrichments. Ambient nitroeen concentrations o The MERL tanksshow a linearincrease in the meanconcentration ofinorganic nitrogen in the tanksas a functionof the loadingrate (Figure 11). I MeanMERL tank DINvs, enrichmentlevel 5 4 o Er 5J = o Et E 1 I 0 o Figure 11r Relationshipbetween tank DIN concentrationand NJoading rate. Page8 of J2 o HOW DOES THE PROVIDENCEAND SEEKONKRIVER SYSTEMCOMPARE WITH THE MERL EXPERIMENT? The presentcondition of the Providenceand SeekonkRivers and sourcesis basedon the 1995- 1996study by DEM WaterResources. The studyconsisted of measurementsof loading from the principal sourcesto the area,which included the three WWTFs and the five major tributaries. The frequencyof samplingof the sourceswas proportional to the magnitudeof the source.The four principalsources: Bucklin andFields WWTFs, and the Blackstoneand SeekonkRivers, were sampledmore than 50 times