Predictive Hydrodynamic and Eutrophication Modeling of Lake ...

outflow, rainfall and evaporation. In general,both the model results and observed data show adecreasing trend in elevation from January toJuly, then the elevation tends to rise from July toOctober, then the temporal trend level-offtowards the end of the year, which reflects theintra-annual precipitation distribution wheremajority of the rainfall occur during the summerin this area. The hydrodynamic model was alsoevaluated by its simulated circulation pattern inFigure 3. Apparently, the model results showthat the circulation in the lake is dominated by acounter-clockwise flow filed while local microcirculationpatterns do exist, which matches thepast knowledge about the circulation in LakeDianchi based on historically measured data.Both the elevation and circulation pattern resultssuggest that the hydrodynamic model likely hasrepresented the hydrologic balance and fluiddynamics well, serving a reasonable basis forfurther water quality modeling.After the hydrodynamic model was calibrated,the water quality simulation was turned on tosimulate the nutrient, phytoplankton, DO andwater column-sediment interaction dynamics.The simulated water quality was compared withobserved data, and key kinetic parameters wereadjusted until a reasonable match between modelresults and data were achieved. Figures 4 to 6plot simulated NH4, DO, TN, and TPconcentrations against observed data at threemonitoring stations in Lake Dianchi. As shown,the model has reasonably reproduced both thespatial and temporal patterns of the observeddata. As shown in Figures 4 and 5, NH4concentration drops sharply during the first threemonths, which was likely caused by the lack ofsufficient inflow loading to offset theconsumption by an early spring algal bloom inthe lake. Accompanying the sharp drop in NH4,we can observe in both the model results andmeasured data that DO tend to increase for thatperiod, which also suggests a likely spring algalbloom.Figures 4 to 6 demonstrate that during thesummer period, DO concentration can rise tovery high level well above the saturation level.This is caused by the intensive summer algalbloom in the lake. Comparing the magnitude ofsummer DO at the north lake stationHuiwanzhong and that at the south lake stationHaikouxi, it is obvious that the peak DO value atthe north lake is significantly higher than that atthe south lake, suggesting a more intensive algalbloom in the north lake. This pattern agrees withthe general observation in Lake Dianchi, and canbe explained by the high nutrient input loadingfrom Daqinhe River and Panlongjiang River.During the dry period, the loading signal fromthe tributaries is significant only at the north lakestation Huiwanzhong. However, during the wetperiod, we can observe spikes of increasednutrient concentration at all the stations. It isinteresting to note that after watershed loadingscaused a spike in NH4 concentration, the spikewould turn into a sharp downtrend until anexternal loading event produces another spike.The downturn of NH4 following a spike is anindicator of stimulated algae activities caused byincreased nutrient concentration, which quicklyuptake inorganic nutrient and depleteconcentration in the water column. Thesephenomenon clearly indicate that the waterquality in Lake Dianchi is significantly impactedby the watershed loading, and the temporaldistribution of incoming loading largelydetermine how the lake water quality vary overtime.In summary, the model in general catches thespatial and temporal distribution of water qualitywell, suggesting it likely to be a reasonablenumerical representation of the prototypesystem. The model-data comparison does showminor disparity between the model result anddata at certain locations and times, which can beattributed to the uncertainty in both the modeland data. Future research will be targeted tocollect more data to characterize the incomingloading as well as the in-lake water qualitypattern. Particularly, a more accurate watershedmodel is anticipated to be developed to providemore robust boundary condition than does theGWLF model.Preliminary Model ApplicationThe purpose of a hydrodynamic and waterquality model is to serve as a platform forevaluating various water quality restorationschemes. In this study, the model was applied toconduct a preliminary scenario analysis throughsimulating how the in-lake algal bloom intensityresponds to load reductions at all the tributariesexcept for Daqinghe River and PanlongjiangRiver. The purpose is to illustrate how the loadreductions at discrete tributaries would haveimpact on the algal bloom intensity in the lake.To execute such an analysis, a baseline scenariowas formulated which was based on the existingloading level as in the calibration model, butrecycling the boundary conditions for 20 years to

simulate the long term trend, and the results onthe 20 th year were used to be the basis ofcomparison (Figure 7). The reduction scenariothen was configured based on the baselinemodel, but removing the nutrient loading for allthe inflow tributaries except for Daqinghe Riverand Panlongjiang River, of which the loadinglevel are the same as in the baseline. Thereduction scenario model was also simulated for20 years and the results on the 20 th year wereplotted in Figure 8. Comparing Figures 7 and 8,it was observed that without reducing thenutrient from Daqinghe and Panlongjiang Rivers,the algal bloom in Lake Dianchi maintain thesimilar level of intensity as in the pre-reductioncondition even though all other tributaries areextremely treated. This result suggests thedifficulty in achieving significant depression ofalgal bloom intensity for Lake Dianchi. It mightrequire much more drastic load reductions andengineering measures to make it possible tosignificantly depress the algae bloom in LakeDianchi. From the modeling result we canunderstand why the tremendous pollutionmanagement efforts in the past decades onlyresult in no observable improvement in the waterquality in the lake. In future research, weanticipate that the model will be further refinedto serve as the core component of the decisionsupport tools to help design robust and efficientwater pollution control and water qualityrestoration schemes for Lake Dianchi. The modelwill be potentially used to evaluate different loadreduction schemes such as how much wouldneed to be reduced from each tributariesincluding Daqinghe and Panlongjiang, and howmuch need to be reduced for internal sources,and what will happen if we restore macrophyteReferences1. Blumberg, A.F. and Mellor, G.L.(1987). “A description of a threedimensionalcoastal ocean circulationmodel”. In: Three-dimensional coastaland ocean models, coastal and estuarinescience, Vol. 4. AmericanGeographical Union, pp. 1-19.2. Cerco, C.F. and Meyers, M. (2000).“Tributary refinements to ChesapeakeBay model”, Journal of EnvironmentalEngineering, ASCE, 124 (2), 164-174.3. Chapra, S.C. (1997). Surface WaterQuality Modeling. The McGRAW-in the lake, as well as implementing othercomprehensive watershed management andengineering measures.Summary and ConclusionsA hydrodynamic and water quality model wasdeveloped to simulate the fate and transport ofnutrient, along with eutrophication dynamics inLake Dianchi, China. The modeling study hasshown that the EFDC modeling framework iscapable of representing the hydrodynamics andwater quality process in the lake. The modelexplicitly simulates the physical, chemical, andbiological processes which are very importantfor understanding the eutrophication dynamics inthe lake. Overall, the model was shown to havereproduced the observed water qualityconcentrations at locations throughout the lake.The preliminary scenario analysis shows that itwould be very hard to achieve observabledepression in algal bloom intensity in the lake.Even by reducing all nutrients from thetributaries except for Panlongjiang River andDaqinghe River, the resulted long term responsein algal bloom intensity to the load reduction isnegligible. This result explains why thetremendous pollution control investment duringthe past decades didn’t result in appreciablewater quality improvement in the lake. Futureresearch would be focused on refining the modelwith more intensive and accurate data, and thenapplying the model to evaluate various potentialpollution control schemes and help designfeasible and efficient ways to restore the waterquality in Lake Dianchi.HILL Company, INC., New York,USA.4. DiToro, D.M. and Fitzpatrick, J.J.1993. Chesapeake Bay Sediment FluxModel. Contract Report EL-93-2, USArmy Engineer Waterways ExperimentStation, Vicksburg, MS.5. Hamrick, J.M. (1992). A threedimensionalenvironmental fluiddynamics computer code: Theorecticaland computational aspects. Specialpaper 317, the College of William andMary, Virginia Institute of MarineScience, Williamsburg, Va.6. Hamrick, John M. (1996). User’sManual for the Environmental Fluid

Dynamics Computer Code. SpecialReport No. 331 in Applied MarineScience and Ocean Engineering.Department of Physical Sciences,School of Marine Science, VirginiaInstitute of Marine Science, TheCollege of William and Mary,Gloucester Point, VA.7. Park, K., Kuo, A., Shen, J., andHamrick, J. (1995) (rev. by Tetra Tech,Inc. 2000). A Three-dimensionalHydrodynamic-Eutrophication Model(HEM-3D): Description of WaterQuality and Sediment ProcessSubmodels (EFDC Water QualityModel). Special Report No. 327 inApplied Marine Science and OceanEngineering.8. Lung, W.S. (2001). “Water QualityModeling for Wasteload Allocationsand TMDLs.” John Wiley& Sons, Inc.,NY.9. Tetra Tech, Inc (2008). Hydrodynamicand water quality modeling to SupportRestoration of the Salton Sea,California. Prepared for Salton SeaAuthority, La Quinta, CA.10. Zou, R., Bai, S., and Parker, A. (2008a)Hydrodynamic and eutrophicationmodeling for a tidal marsh impactedestuarine system using EFDC. Coastaland Estuary Modeling, 561-589.11. Zou, R., Lung, W.S., and Wu, J.(2008b), Multiple-pattern parameteridentification and uncertainty analysisapproach for water quality modeling.Ecological Modeling, doi:10.1016/J.ecolmodel.2008.11.02112. Zou, R., Carter, S., Shoemaker, L.,Parker, A., and Henry, T., (2006), AnIntegrated hydrodynamic and waterquality modeling system to supportnutrient TMDL development forWissahickon Creek. Journal ofE le v a tio n (m )1888188818871887188618861885Environmental Engineering, ASCE,132(4), 555-566.Observed Elev Simulated 2003J F A M J S O DDateFigure 2 Simulated Elevation vs. DataFigure 1 Computational Grid for Dianchi ModelFigure 3 Simulated Flow Field in Lake Dianchi

Figure 1 EFDC Grid for Lake DianchiTP(mg/L)TN(mg/L)DO(mg/L)NH4(mg/L)1.61.20.80.40.0Jan-0320.015.010.05.00.05.04.03.02.01.00.01.21.00.80.60.40.20.0Jan-03Jan-03Jan-03Huiw anzong NH4Mar-03May-03Huiw anzong DOMar-03May-03Huiw anzong TNMar-03May-03Huiw anzong TPMar-03May-03Jul-03Jul-03Jul-03Jul-03Sep-03Sep-03Sep-03Sep-03Nov-03Nov-03Nov-03Nov-03Jan-04Jan-04Jan-04Jan-04TP(mg/L)TN(mg/L)DO(mg/L)NH4(mg/L)1.61.20.80.40.0Jan-0320.015.010.05.00.05.04.03.02.01.00.01.21.00.80.60.40.20.0Jan-03Jan-03Jan-03Guanyszong NH4Mar-03May-03Guanyszong DOMar-03May-03Guanyszong TNMar-03May-03Guanyszong TPMar-03May-03Jul-03Jul-03Jul-03Jul-03Sep-03Sep-03Sep-03Sep-03Nov-03Nov-03Nov-03Nov-03Jan-04Jan-04Jan-04Jan-04Figure 4 Simulated Water Quality Vs Dataat Huiwan ZhongFigure 5 Simulated Water Quality Vs Dataat Guanyinxi

NH4(mg/L)1.61.20.80.40.0Jan-03Haikouxi NH4Mar-03May-03Jul-03Sep-03Nov-03Jan-04CHC(mg/L)250.0Haikouxi CHC200.0150.0100.050.00.0Jan-03Mar-03May-03DO(mg/L)20.0Haikouxi DO15.010.05.00.0Jan-03Mar-03May-03Jul-03Sep-03Nov-03Jan-045.04.03.02.01.0Haikouxi TN0.0Jan-03Mar-03May-03Jul-03Sep-03Nov-03Jul-03Jan-04TN(mg/L)1.21.00.80.60.40.20.0Haikouxi TPJan-03Mar-03May-03Jul-03Sep-03Nov-03Jan-04Sep-03Nov-03Jan-04Figure 7 Blue-green algae for baselineCHC(mg/L)250.0Haikouxi CHC200.0150.0100.050.00.0Jan-03Mar-03May-03Jul-03Sep-03Nov-03Jan-04Figure 8 Blue-green algae with reductionTP(mg/L)Figure 6 Simulated Water Quality Vs Dataat Haikouxi