<%BANNER%>

Two-Dimensional (Depth-Averaged) Modeling of Flow and Phosphorus Dynamics in Constructed Wetlands

Permanent Link: http://ufdc.ufl.edu/UFE0021277/00001

Material Information

Title: Two-Dimensional (Depth-Averaged) Modeling of Flow and Phosphorus Dynamics in Constructed Wetlands
Physical Description: 1 online resource (241 p.)
Language: english
Creator: Min, Joong Hyuk
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: 2d, constructed, everglades, hydrodynamics, modeling, phosphorus, stormwater, wetland
Environmental Engineering Sciences -- Dissertations, Academic -- UF
Genre: Environmental Engineering Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Constructed wetlands are increasingly being used worldwide to facilitate nutrient removal, in particular, phosphorus in Florida, from agricultural runoff or conventional wastewater treatment plant effluents. Two-dimensional (2-D) flow dynamics, solute transport, and phosphorus dynamics models are developed to simulate spatio-temporal variations of flow and/or phosphorus dynamics in constructed wetlands. The MIKE 21 hydrodynamics (HD), advection-dispersion (AD), and ECO Lab models were adopted as the basic framework with modifications and enhancement of phosphorus kinetic pathways to incorporate ecosystem dynamics among water column, floc and upper soil layers, and vegetative communities including emergent aquatic vegetation (EAV), submerged aquatic vegetation (SAV), and periphyton. The models were calibrated and validated for two constructed wetland systems in Florida: the Orlando Easterly Wetland (OEW) Cell 7 and the Stormwater Treatment Area (STA) 5 northern flow-way. The spatio-temporal water level fluctuations, tracer (bromide/chloride), and phosphorus concentration profiles were reasonably simulated through linkages between the HD and the AD or ECO Lab model, and key model parameters were estimated. The OEW modeling study is focused on impacts of topographic and vegetative heterogeneity on short-circuiting flow through sensitivity analysis deduced during model calibration on a bromide breakthrough curve. The short-term simulation results show that relic ditches or other ditch-shaped landforms and the associated sparse vegetation along the main flow direction intensify the short-circuiting pattern, considerably reducing hydraulic efficiency. In the northern flow-way of STA 5, the HD model calibration for six long-term monitoring sites accurately represented observed annual variations in hydroperiod. On average, the root-mean-square error (RMSE) for predicting daily water level was less than 0.10 m. Manning?s roughness coefficients for dense EAV and SAV areas, which were estimated as a function of vegetation type and density, ranged from 0.67 to 1.0 s/m1/3 and 0.12 to 0.15 s/m1/3, respectively. The AD model calibration for four long-term monitoring sites agreed very well with the measured annual variations in chloride concentration with the average RMSE of 13.48 mg/L. Longitudinal dispersivity was estimated to be 2 m and was over an order of magnitude higher than the transverse one. Results of conservative phosphorus transport simulation confirm the findings of recent STA studies that the EAV systems are less efficient for phosphorus retention compared to the SAV systems and current STA system is not sufficient for reducing dissolved organic phosphorus (DOP) concentration to very low levels. Linked with the HD model, the calibrated ECO Lab phosphorus dynamics model better simulated observed annual variations in soluble reactive phosphorus (SRP) and DOP level than particulate phosphorus (PP) level, which is primarily due to uncertainty of model parameters on spatio-temporal variations of mass transfer mechanisms between water column and floc layer. Several limitations of this modeling study are addressed, followed by recommendations for future study to develop a more robust scientific and management modeling tool for constructed wetlands, overcoming the shortcomings of traditional treatment wetland modeling approaches.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Joong Hyuk Min.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Wise, William R.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021277:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021277/00001

Material Information

Title: Two-Dimensional (Depth-Averaged) Modeling of Flow and Phosphorus Dynamics in Constructed Wetlands
Physical Description: 1 online resource (241 p.)
Language: english
Creator: Min, Joong Hyuk
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: 2d, constructed, everglades, hydrodynamics, modeling, phosphorus, stormwater, wetland
Environmental Engineering Sciences -- Dissertations, Academic -- UF
Genre: Environmental Engineering Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Constructed wetlands are increasingly being used worldwide to facilitate nutrient removal, in particular, phosphorus in Florida, from agricultural runoff or conventional wastewater treatment plant effluents. Two-dimensional (2-D) flow dynamics, solute transport, and phosphorus dynamics models are developed to simulate spatio-temporal variations of flow and/or phosphorus dynamics in constructed wetlands. The MIKE 21 hydrodynamics (HD), advection-dispersion (AD), and ECO Lab models were adopted as the basic framework with modifications and enhancement of phosphorus kinetic pathways to incorporate ecosystem dynamics among water column, floc and upper soil layers, and vegetative communities including emergent aquatic vegetation (EAV), submerged aquatic vegetation (SAV), and periphyton. The models were calibrated and validated for two constructed wetland systems in Florida: the Orlando Easterly Wetland (OEW) Cell 7 and the Stormwater Treatment Area (STA) 5 northern flow-way. The spatio-temporal water level fluctuations, tracer (bromide/chloride), and phosphorus concentration profiles were reasonably simulated through linkages between the HD and the AD or ECO Lab model, and key model parameters were estimated. The OEW modeling study is focused on impacts of topographic and vegetative heterogeneity on short-circuiting flow through sensitivity analysis deduced during model calibration on a bromide breakthrough curve. The short-term simulation results show that relic ditches or other ditch-shaped landforms and the associated sparse vegetation along the main flow direction intensify the short-circuiting pattern, considerably reducing hydraulic efficiency. In the northern flow-way of STA 5, the HD model calibration for six long-term monitoring sites accurately represented observed annual variations in hydroperiod. On average, the root-mean-square error (RMSE) for predicting daily water level was less than 0.10 m. Manning?s roughness coefficients for dense EAV and SAV areas, which were estimated as a function of vegetation type and density, ranged from 0.67 to 1.0 s/m1/3 and 0.12 to 0.15 s/m1/3, respectively. The AD model calibration for four long-term monitoring sites agreed very well with the measured annual variations in chloride concentration with the average RMSE of 13.48 mg/L. Longitudinal dispersivity was estimated to be 2 m and was over an order of magnitude higher than the transverse one. Results of conservative phosphorus transport simulation confirm the findings of recent STA studies that the EAV systems are less efficient for phosphorus retention compared to the SAV systems and current STA system is not sufficient for reducing dissolved organic phosphorus (DOP) concentration to very low levels. Linked with the HD model, the calibrated ECO Lab phosphorus dynamics model better simulated observed annual variations in soluble reactive phosphorus (SRP) and DOP level than particulate phosphorus (PP) level, which is primarily due to uncertainty of model parameters on spatio-temporal variations of mass transfer mechanisms between water column and floc layer. Several limitations of this modeling study are addressed, followed by recommendations for future study to develop a more robust scientific and management modeling tool for constructed wetlands, overcoming the shortcomings of traditional treatment wetland modeling approaches.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Joong Hyuk Min.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Wise, William R.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021277:00001


This item has the following downloads:


Full Text

PAGE 1

1 TWO-DIMENSIONAL (DEPTH-AVERAGED) M ODELING OF FLOW AND PHOSPHORUS DYNAMICS IN CONSTR UCTED WETLANDS By JOONG HYUK MIN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Joong Hyuk Min

PAGE 3

3 To my parents and my beloved wife, Moon-Jung

PAGE 4

4 ACKNOWLEDGMENTS My first acknowledgment must go to my advi sor and supervisory committee chair, Dr. William R. Wise, for his support, advice, a nd help throughout my doctoral study. I deeply appreciate his thoughtful guidance and constant enc ouragement in this academic endeavor. I also thank the other members of my committee, Dr. Thomas L. Crisman, Dr. K.R. Reddy, and Dr. Andrew I. James for their valuab le advice and help on my researc h. In particular, their academic excellence on wetland sciences and modeli ng has provided a strong foundation of my interdisciplinary research. This work was made possible by Alumni Fellows hip provided by the University of Florida. I am also fortunate to have the opportunity to study with many outstanding faculty and academic colleagues who I have met in the Department of Environmental Engineering Sciences. In particular, I thank Dr. Jaehyun Cho, Dr. Christ opher Martinez, Armin Feldman, Sarah Anderson, Gordon Brown, and Alisa Marchionno in Hydr ologic Science group for their friendship, encouragement and practical help. In addition, I a ppreciate many Korean graduate students in the department. I am sorry I can not list all of them here. I thank the Danish Hydraulic Institute Cor poration for providing access to and support for the software used in my dissertation researc h. The flow and phosphorus measurement data for this study were provided by South Florida Water Management District. Sp ecial thanks go to Dr. Jana Newman, who helped me find useful info rmation on the Stormwater Treatment Areas and provided several key input data for model. Her kind reply and technical comments were always helpful. Finally, this dissertation would not have been possible without my beloved familys love and support. I express my sincere gratitude to my parents for their spiritual and financial support with affectionate encouragement. I also owe a huge debt of gratitude to my brother and his

PAGE 5

5 family. To my wife, Moon-Jung, I can never th ank her enough. With her never-failing love, understanding, patience, and encouragement, she ha s been always with me. I thank my beloved, one-year-old son, Toby, for his birt h and growth, which always br ing me indescribable joy and happiness. Most of all, I truly thank God for the eternal li fe as well as everything above that He allows in my life.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4LIST OF TABLES................................................................................................................. ........10LIST OF FIGURES................................................................................................................ .......12ABSTRACT....................................................................................................................... ............15CHAPTER 1 INTRODUCTION..................................................................................................................17Study Background............................................................................................................... ...17Study Objectives............................................................................................................... ......222 LITERATURE REVIEW.......................................................................................................24Wetland Hydrology/Hydraulics..............................................................................................24Wetland Flow Models.....................................................................................................24Wetland Hydraulic Performance.....................................................................................27Phosphorus Dynamics in Wetlands........................................................................................29Phosphorus Retention/Release........................................................................................29Phosphorus Retention Models.........................................................................................313 HYDRODYNAMICS-WATER QUALITY INTEGRATED MODEL SETUP....................37Wetland Model Dimension.....................................................................................................37Overview of Hydrodynamics-Water Qu ality Integrated Model: MIKE 21............................38Basic Parameters.............................................................................................................38Hydrodynamics (HD) Module.........................................................................................40Governing equations................................................................................................40Model conditions and parameters.............................................................................42Advection-Dispersion (AD) Module...............................................................................43Governing equation..................................................................................................43Model conditions and parameters.............................................................................44Water Quality/Ecological Engi neering (ECO Lab) Module...........................................45Phosphorus Dynamics Model.................................................................................................47Model Scope....................................................................................................................47State Variables................................................................................................................ .49Phosphorus Transformation Processes............................................................................50Atmospheric phosphorus deposition........................................................................51Mineralization..........................................................................................................52Formation/decay (sor ption/desorption)....................................................................53Sedimentation/resuspension.....................................................................................55

PAGE 7

7 Uptake by biota........................................................................................................57Diffusion...................................................................................................................58Sequestration............................................................................................................58Temperature effect...................................................................................................59Parameters................................................................................................................604 MODEL APPLICATION TO THE OR LANDO EASTERLY WETLAND (OEW) CELL 7......................................................................................................................... ..........72Introduction................................................................................................................... ..........72Study Area..................................................................................................................... ..72Previous Studies..............................................................................................................73Bromide tracer test and hydraulic analysis..............................................................73One-dimensional (1-D) transient storage model......................................................74Model Scope....................................................................................................................74Specific Aims..................................................................................................................75Methods........................................................................................................................ ..........76Moment Analysis and Hydraulic Performance...............................................................76Hydrodynamic Model Setup............................................................................................78Advection-Dispersion Model Setup................................................................................79Results and Discussion......................................................................................................... ..80Bromide Tracer Experiment Simulation.........................................................................80Sensitivity Analysis I: The Effect of Bathymetry on Short-circuiting Flow...................81Sensitivity Analysis II: The Effect of Vegetation on Short-circuiting Flow...................83Implications of This Modeling Study..............................................................................87Summary........................................................................................................................ .........905 MODEL APPLICATION TO THE STORMW ATER TREATMENT AREA (STA) 5 NORTHERN FLOW-WAY: I. FLOW DYNAMICS AND SOLUTE TRANSPORT..........98Introduction................................................................................................................... ..........98Study Area..................................................................................................................... ..98Specific Aims................................................................................................................100Materials and Methods.........................................................................................................100Data Used in Models.....................................................................................................101Basic Parameters Setup.................................................................................................103Hydrodynamic Model Setup..........................................................................................104Advection-Dispersion Model Setup..............................................................................106Model Calibration and Validation.................................................................................107Sensitivity Test..............................................................................................................108Results and Discussion.........................................................................................................108Hydrodynamics Model..................................................................................................108Hydroperiod simulation..........................................................................................109Parameter estimation: hydraulic roughness coefficients........................................110Sensitivity analysis.................................................................................................111Advection-Dispersion Model........................................................................................112Chloride transport simulation.................................................................................112

PAGE 8

8 Parameter estimation: longitudina l and transverse dispersivity.............................113Sensitivity analysis.................................................................................................114Summary........................................................................................................................ .......1146 MODEL APPLICATION TO THE STORMW ATER TREATMENT AREA (STA) 5 NORTHERN FLOW-WAY: II. PHOSPHORUS DYNAMICS..........................................147Introduction................................................................................................................... ........147Materials and Methods.........................................................................................................149ECO Lab Model Setup..................................................................................................150Phosphorus data used in model..............................................................................150Initial conditions.....................................................................................................152Constants used in model.........................................................................................154Phosphorus dynamics: EAV vs. SAV....................................................................156Results and Discussion.........................................................................................................157Conservative Phosphorus Transport Model..................................................................157Estimation of phosphorus retention........................................................................157Implication of this modeling study.........................................................................160Phosphorus Dynamics Model........................................................................................161Phosphorus dynamics simulation...........................................................................161Sensitivity analysis.................................................................................................165Summary........................................................................................................................ .......1667 CONCLUSIONS AND RECOMMENDATIONS...............................................................208Conclusions.................................................................................................................... .......208Recommendations................................................................................................................ .209APPENDIX A RECALIBRATION RESULTS OF HYD RODYNAMICS SIMULATION FROM MAY 1, 2003 TO DECEMBER 31, 2004............................................................................214B CALIBRATION OF THE CHLORIDE TRANSPORT AD MODEL ON LONGITUDINAL AND TRANSVERSE DISPERSIVITY................................................217C RESULTS OF CONSERVATIVE SOLUTE TRANSPORT SIMULATION ( 1 RMSE) FROM MAY 1, 2003 TO APRIL 30, 2004 ON SRP CONCENTRATION PROFILES....................................................................................................................... .....221D RESULTS OF CONSERVATIVE SOLUTE TRANSPORT SIMULATION ( 1 RMSE) FROM MAY 1, 2003 TO APRIL 30, 2004 ON DOP CONCENTRATION PROFILES....................................................................................................................... .....223E RESULTS OF CONSERVATIVE SOLUTE TRANSPORT SIMULATION ( 1 RMSE) FROM MAY 1, 2003 TO AP RIL 30, 2004 ON PP CONCENTRATION PROFILES....................................................................................................................... .....225

PAGE 9

9 LIST OF REFERENCES.............................................................................................................227BIOGRAPHICAL SKETCH.......................................................................................................241

PAGE 10

10 LIST OF TABLES Table page 3-1 State variables used in the ECO Lab phosphorus dynamics model...................................613-2 Processes used in the ECO Lab phosphorus dynamics model...........................................623-3 Constants used in the ECO Lab phosphorus dynamics model..........................................633-4 Atmospheric total phosphorus (TP) bulk deposition rates in Fl orida reported in previous studies............................................................................................................... ...653-5 Estimation of atmospheric phosphorus depos ition used in STA 5 northern flow-way phosphorus models.............................................................................................................653-6 Literature values for mineralization/de gradation rates of or ganic matter in water column, floc, and soil layer................................................................................................663-7 Literature values for various phosphorus decay/desorption rates......................................673-8 Literature values for critical veloc ity and phosphorus sedimentation/resuspension rate in wetlands............................................................................................................... ...683-9 Literature values (maximum growth rate and half saturation constant) for phosphorus uptake by emergent ( Typha spp. ), submerged aquatic vegetation, and periphyton...........693-10 Literature values for phosphorus se questration rate in the Everglades..............................704-1 Physical parameters used in models and metrics showing the hydraulic performance of the OEW Cell 7..............................................................................................................915-1 Monthly water budget for the HD mo del of STA 5 northern flow-way..........................1175-2 Vegetation type of STA 5................................................................................................1185-3 Monthly water budget recalibrated for th e AD model of STA 5 northern flow-way......1195-4 Manning roughness coefficients used in the HD model, which are assigned with respect to type and density of vegetati on habitat in STA 5 northern flow-way..............1205-5 Coefficients for effectiv e roughness used in the SFWMM.............................................1205-6 Water level sensitivity on hydraulic resistance represented by Mannings roughness coefficient in STA 5 northern flow-way..........................................................................1215-7 Dispersion coefficient in constructed wetlands reported in previous studies..................122

PAGE 11

11 5-8 Sensitivity analysis on longitudinal dispersivity ( y is fixed to 0.1m) in STA 5 northern flow-way............................................................................................................1225-9 Sensitivity analysis on transverse dispersivity ( x is fixed to 2m) in STA 5 northern flow-way....................................................................................................................... ...1226-1 Calibrated or estimated values of constants used in the ECO Lab phosphorus dynamics model...............................................................................................................1686-2 Differences of phosphorus dynamics model between EAV and SAV wetland ecosystem...................................................................................................................... ...1706-3 Annual phosphorus budget for STA 5 north ern flow-way in Water Year (WY) 2004 (May 03-April 04) estimated through conservative phosphor us transport simulation....1716-4 Comparison of the annual average phosphor us concentrations between measurement and simulation during the model calibrati on period (May 1, 2003 to April 30, 2004)....1726-5 RMSE values of the phosphorus dynamics model during the model calibration period (May 1, 2003 to April 30, 2004)......................................................................................1726-6 Comparison of the average phosphorus c oncentrations betwee n measurement and simulation during the model validation period (May 1, 2004 to December 31, 2004)....1736-7 RMSE values of the phosphorus dynamics model during the m odel validation period (May 1, 2004 to December 31, 2004)..............................................................................1736-8 Results of sensitivity analysis of SRPw on major parameters used in the phosphorus dynamics model...............................................................................................................1746-9 Results of sensitivity analysis of DOP on major parameters used in the phosphorus dynamics model...............................................................................................................1766-10 Results of sensitivity analysis of PP on major parameters used in the phosphorus dynamics model...............................................................................................................178B-1 Calibration results of the chloride tr ansport AD model according to the ratio of longitudinal ( x) vs. transverse dispersivity ( y) in STA 5 northern flow-way...............217

PAGE 12

12 LIST OF FIGURES Figure page 2-1 Phosphorus dynamics in wetlands.....................................................................................363-1 A conceptual diagram of constructed wetland phosphorus dynamics model applied in this study..................................................................................................................... .......714-1 Location and plan view of study area, the OEW Cell 7.....................................................924-2 Fluctuation of wetland water vol ume during the simulation period..................................934-3 Bromide tracer experiment simula tion: transient 2-D bromide plume..............................944-4 Bromide tracer experiment simulation: model fit on the measured BTC..........................954-5 The effect of bathymetry on short-circuiting flow.............................................................964-6 The effect of vegetation dist ribution on short-circuiting flow...........................................975-1 Location map of study area, STA 5.................................................................................1235-2 A schematic of STA 5 indicating the hydrau lic structures as well as the flow and vegetation pattern.............................................................................................................1245-3 Daily-based precipitation and evapotranspi ration rates in STA 5. A) Precipitation. B) Evapotranspiration...........................................................................................................1255-4 Daily-based surface water flows in ST A 5 northern flow-way. A) Inflows. B) Outflows....................................................................................................................... ....1265-5 Bathymetry prediction map of STA 5 northern flow-way generated using IDW interpolation scheme........................................................................................................1275-6 Raster-typed bathymetry predicti on map of STA 5 northern flow-way..........................1285-7 Vegetation distribution shapefile of STA 5 northern flow-way......................................1295-8 Mannings roughness coefficients assigne d at each model grid cell of STA 5 northern flow-way. A) Cell 1A. B) Cell 1B.....................................................................1305-9 Inlet chloride concentration pr ofiles in STA 5 northern flow-way..................................1325-10 Simulation results of the HD model on hydr operiod fluctuation at the six water level measurement points in STA 5 northern flow -way during the mode l calibration period.1335-11 Simulation results of the HD model on hydr operiod fluctuation at the six water level measurement points in STA 5 northern flow -way during the mode l validation period..136

PAGE 13

13 5-12 Sensitivity analysis results of Mannings roughness coefficient ( n ) on hydroperiod fluctuation at the six water level measur ement points in STA 5 northern flow-way......1395-13 Simulation results of the AD model on ch loride concentration profiles at the four water quality measurement points in STA 5 northern flow-way during the model calibration period.............................................................................................................1425-14 Simulation results of the AD model on ch loride concentration profiles at the four water quality measurement points in STA 5 northern flow-way during the model validation period..............................................................................................................1445-15 Calibration curves of ch loride transport model on longitudinal and transverse dispersivity................................................................................................................... ....1466-1 Temporal variations of inlet phosphoru s species at G342A. A) Time-series SRPw, DOP, and PP concentration profiles. B) Re lative portion of inle t phosphorus species...1806-2 Temporal variations of inlet phosphoru s species at G342B. A) Time-series SRPw, DOP, and PP concentration profiles. B) Re lative portion of inle t phosphorus species...1816-3 Temporal variations of inlet phosphoru s species at G349A. A) Time-series SRPw, DOP, and PP concentration profiles. B) Re lative portion of inle t phosphorus species...1826-4 Raster-typed phosphorus cont ent prediction maps in the floc and the upper soil layer of STA 5 northern flow-way used in the ECO Lab phosphorus dynamics model..........1836-5 Temperature profile used in the phosphor us dynamics model, measured at G343C.......1856-6 A schematic of phosphorus dynamics in the EAV system of STA 5..............................1866-7 A schematic of phosphorus dynamics in the SAV system of STA 5...............................1876-8 Results of conservative solute tran sport simulation from May 1, 2003 to December 31, 2004 on SRPw concentration profiles........................................................................1886-9 Results of conservative solute tran sport simulation from May 1, 2003 to December 31, 2004 on DOP concentration profiles..........................................................................1906-10 Results of conservative solute tran sport simulation from May 1, 2003 to December 31, 2004 on PP concentration profiles.............................................................................1926-11 Calibration results of flow and phosphor us dynamics coupled simulation from May 1, 2003 to April 30, 2004 on SRPw concentration profiles..............................................1946-12 Calibration results of flow and phosphor us dynamics coupled simulation from May 1, 2003 to April 30, 2004 on DOP concentration profiles...............................................1966-13 Calibration results of flow and phosphor us dynamics coupled simulation from May 1, 2003 to April 30, 2004 on PP concentration profiles...................................................198

PAGE 14

14 6-14 Validation results of flow and phosphorus dynamics coupled simulation from May 1, 2004 to December 31, 2004 on SRPw concentration profiles..........................................2006-15 Validation results of flow and phosphorus dynamics coupled simulation from May 1, 2004 to December 31, 2004 on DOP concentration profiles...........................................2026-16 Validation results of flow and phosphorus dynamics coupled simulation from May 1, 2004 to December 31, 2004 on PP concentration profiles...............................................2046-17 Results of the ECO Lab phosphorus dynamics model on six state variables in floc and soil layers at six field monitoring sites......................................................................2066-18 Results of the ECO Lab phosphorus dynamics model on Pmacro and Pperi at randomly chosen twelve model grid cells. A) Pmacro. B) Pperi..........................................................207A-1 Recalibration results of hydrodynamics simulation from May 1, 2003 to December 31, 2004 on hydroperiod fluctuation................................................................................214C-1 Results of conservative solute transp ort simulation ( 1 RMSE) from May 1, 2003 to April 30, 2004 on SRPw concentration profiles...............................................................221D-1 Results of conservative solute transp ort simulation ( 1 RMSE) from May 1, 2003 to April 30, 2004 on DOP con centration profiles................................................................223E-1 Results of conservative solute transp ort simulation ( 1 RMSE) from May 1, 2003 to April 30, 2004 on PP concentration profiles....................................................................225

PAGE 15

15 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy TWO-DIMENSIONAL (DEPTH-AVERAGED) M ODELING OF FLOW AND PHOSPHORUS DYNAMICS IN CONSTR UCTED WETLANDS By Joong Hyuk Min August 2007 Chair: William R. Wise Major: Environmental Engineering Sciences Constructed wetlands are increasi ngly being used worldwide to facilitate nutrient removal, in particular, phosphorus in Fl orida, from agricultural runo ff or conventional wastewater treatment plant effluents. Two-dimensional (2-D) flow dynamics, solute transport, and phosphorus dynamics models are de veloped to simulate spatio-t emporal variations of flow and/or phosphorus dynamics in constructed wetlands. The MIKE 21 hydrodynamics (HD), advection-dispersion (AD), and ECO Lab models were adopted as the basic framework with modifications and enhancement of phosphorus kinetic pathways to incorporate ecosystem dynamics among water column, floc and upper soil layers, and vegetative communities including emergent aquatic vegetation (EAV), submerged aquatic vegetation (SAV), and periphyton. The models were calibrated and valid ated for two constructed wetl and systems in Florida: the Orlando Easterly Wetland (OEW) Cell 7 and the Stormwater Treatment Area (STA) 5 northern flow-way. The spatio-temporal water level fluctuations, trac er (bromide/chloride), and phosphorus concentration profiles were reasonabl y simulated through linkages between the HD and the AD or ECO Lab model, and key model parameters were estimated. The OEW modeling study is focused on impacts of topographic and vegetative heterogeneity on short-circuiting flow through sensitivity analysis deduced during model

PAGE 16

16 calibration on a bromide breakthrough curve. The short-term simulation results show that relic ditches or other ditch-shaped la ndforms and the associated sparse vegetation along the main flow direction intensify the short-ci rcuiting pattern, considerably reducing hydraulic efficiency. In the northern flow-way of STA 5, the HD m odel calibration for six long-term monitoring sites accurately represented observed annual vari ations in hydroperiod. On average, the rootmean-square error (RMSE) for predicting daily water level was less than 0.10 m. Mannings roughness coefficients for dense EAV and SAV ar eas, which were estimated as a function of vegetation type and density, ranged from 0.67 to 1.0 s/m1/3 and 0.12 to 0.15 s/m1/3, respectively. The AD model calibration for four long-term m onitoring sites agreed very well with the measured annual variations in chloride con centration with the aver age RMSE of 13.48 mg/L. Longitudinal dispersivity was estimated to be 2 m and was over an order of magnitude higher than the transverse one. Results of conservative phosphorus tr ansport simulation confirm the findings of recent STA studies th at the EAV systems are less e fficient for phosphorus retention compared to the SAV systems and current STA sy stem is not sufficient for reducing dissolved organic phosphorus (DOP) concentration to very low levels. Linked with the HD model, the calibrated ECO Lab phosphorus dynamics model be tter simulated observed annual variations in soluble reactive phosphorus (SRP) and DOP level than particulate phosphorus (PP) level, which is primarily due to uncertainty of model para meters on spatio-temporal variations of mass transfer mechanisms between water column and floc layer. Several limitations of this modeling study ar e addressed, followed by recommendations for future study to develop a more robust scientific and management modeling tool for constructed wetlands, overcoming the shortcomings of trad itional treatment wetland modeling approaches.

PAGE 17

17 CHAPTER 1 INTRODUCTION Study Background Constructed wetlands are increa singly being used to facil itate nutrient removal from conventional wastewater treatment plant effluent and agricultural runo ff (Kadlec and Hammer, 1988; Hammer, 1989; Brix, 1994; Ka dlec and Knight, 1996; Gopal, 1999; Haberl, 1999; Mitsch et al., 2000; Schaafsma et al ., 2000; Chimney and Goforth, 2001; Braskerud, 2002; Black and Wise, 2003; Chimney and Goforth, 2006). This is a result of the acknowledgement, development, and application of the natural ability of wetlands to transform and store organic matter and nutrients among the newly-recognized multiple functions and values of wetlands, which once were regarded as wastelands (M itsch and Gosselink, 1993; Brix, 1994). In addition, these extensive applications of the wetland function of water quality improvement are attributed to the fact that using constructe d wetlands is generally one of the most ecologically friendly and cost-effective green technologi es for removing nutrients, pr imarily nitrogen and phosphorus (Goforth, 2001). The Everglades is an internationally-rec ognized, subtropical wetland ecosystem in South Florida. It was historically a vast oligotrophic fr eshwater wetland that almost covered the landscape of South Florida before the 1900s (Chimney and Goforth, 2001). However, the ecologically unique wetland system has been ad versely impacted by the altered hydrology that was originally manipulated for flood control and th e influx of nutrient-rich runoff generated from urban development and agricultural activities (Newman and Lynch, 2001). Eutrophication caused by excess loadings of nutri ents into a variety of aquatic bodies is one of the biggest environmental concerns aroun d the world. In the Everglades, deteriorated water quality and substantial shifts in plan t and microbial communities, which are highly

PAGE 18

18 sensitive to phosphorus availability, have been reported in downstream areas of agricultural runoff discharge (Davis, 1994; Doren et al., 1997; Nungesser and Chimney, 2001). Phosphorus, which is a vital element of all natural systems and a major component in most fertilizers, has been designated as the trigger that has put the fragile ecosystem at risk. Excess phosphorus found in stormwater discharge from agricultural and urban areas ultimately reaches the Everglades. It accelerates algal growth and an overabundance of e xotic species, replacing native biota, which is harmful to the marsh ecosystem. Since Floridas 1994 Everglades Forever Act was set into action, se veral construction, research, and regulation activities for restoring the remaining Everglades ecosystem have been conducted, including application of constructed wetlands referred to as Stormwater Treatment Areas (STAs) and implementation of Best Mana gement Practices (BMPs) (Goforth, 2001). The South Florida Water Management District (SFWMD) is the government agency responsible for the design, construction, and maintenance of the STAs. With respect to the phosphorus concern, the Act mandated to set both inte rim (50 ppb as TP) and long-term water quality goals (likely or nearly 10 ppb as TP) for ultimate restoration and preservation of the Everglades. Of the many efforts to reduce phosphorus concentration enteri ng the Everglades to fulfill the interim goal, it was reported that significant phosphorus load re ductions have been successfully achieved through STAs (Goforth, 2001). To date, the Everglades STAs, which is th e largest constructed wetland system in the world (Chimney and Goforth, 2001), is a typi cal example showing that application of constructed wetlands is an effective and pow erful technology for reduction of phosphorus concentrations. According to th e operational performance results (May 1995 to April 1999) of Everglades Nutrient Removal Project (ENRP), which is now incorporated into STA-1W,

PAGE 19

19 phosphorus concentrations at the outflow pump station ranged from 10 to 39 ppb as TP compared to inflow concentrations ranging fr om 57 to 201 ppb as TP and total 77.1 % TP load reduction was achieved (Nungesser and Chimney, 2001). Early results on average phosphorus removal performance of all the operationa l STAs revealed that outflow phosphorus concentrations of 25 to 35 ppb as TP may be su stainable in the STAs (Goforth, 2001). When the STAs are used in combination with the Advanced Treatment Technologies (ATTs), which are alternative phosphorus reduction technologies to meet the long-term outflow phosphorus concentration goal, such as SAV (Knight et al ., 2003)/periphyton-based STAs (Bays et al., 2001; DeBusk, et al., 2004; McCormick et al., 2006) a nd hybrid systems of chemical and biological treatment (Gu et al., 2001), phosphorus discharge is expected to be reduce d to very low levels (10 to 20 ppb as TP) (Chimney and Goforth, 2001; Goforth, 2001). Recently, several projects to optimize nutrient removal performance by the STAs are being extensivel y conducted in addition to the development of the ATTs (Goforth, 2001; Newman and Lynch, 2001; Dierberg et al., 2002; Dierberg et al., 2005; Pietro et al., 2006). Phosphorus is not conservative in wetland syst ems; that is, it inte racts strongly with wetland soils and biota, which play a key role in short or long-term phos phorus storage (Kadlec, 1997). In terms of nutrient removal in wetlands as the phosphorus cy cle has a significant difference from nitrogen in that there is no co mplete phosphorus removal mechanism similar to denitrification process in nitrogen cycle. Some phosphorus input into a wetland system is temporarily retained as various forms. Phosphorus retention can be defined as the capacity of a wetland to remove phosphorus from the water column through physical and biogeochemical processes and retain it in a form not easily released under normal conditions (Reddy et al., 1999). As the retention capacity is spatially hetero geneous and temporally not unlimited within a

PAGE 20

20 wetland system and is highly dependent on the change of biogeochemical condition such as redox and pH, mainly resulted from alte red hydrology, phosphorus retention has been traditionally considered as a black box due to its complexity even in many constructed wetland systems. As a result, treatment efficiency ha s been mainly estimated and compared through empirical input-output analysis or simple firs t-order kinetic model (K adlec, 1994b; Kadlec and Knight, 1996; Reddy et al., 1999). The most important factor sustaining the stru cture and function of wetland systems is the flow (Hammer, 1989; Arnold et al., 2001). Unders tanding flow characteris tics in constructed wetlands is also essential becau se it basically determines the availability of pollutants for assimilation by biota and sorption by soils. In addition, it significantly changes temporally because storm events in the wet season can ge nerate huge runoff; on the other hand, drought in the dry season makes the wetland surface almost dr y. In spite of its importance, wetland flow dynamics has been traditionally considered under unrealistic hydraulic conditions such as steadystate and plug-flow condition, which are base d on unrealistic physical settings such as rectangular wetland shape, constant flat bathymetry (slope), and single value of flow resistance. This is the inherent limitation of traditional wetland phosphorus retention models, which mainly depend on one-dimensional (1-D), conceptual, a nd parameter-lumped modeling approaches. In addition, it is neither easy to fi nd the physical meanings of their key parameters associated with phosphorus retention nor possible to estimate the impacts of va rious physical and ecological factors affecting the flow and phosphorus dynamics independently. Kadlec (2000) suggested that new paradigms are required that incorporate the ab ility to describe short-circuiting and spatial distribution of vegetation, indicating the inad equacy of traditiona l phosphorus retention modeling approaches. Even though th ese traditional modeling approach es have been very useful

PAGE 21

21 as a management modeling tool during the early stage of system development, and a predictive tool of long-term treatment efficiency, more advanced modeling approaches for flow and phosphorus dynamics in constructed wetlands are n ecessary to more systematically understand and predict the internal processes of a bloc k box, overcoming the critical shortcomings of previous modeling efforts (Clement et al., 1998). Multi-dimensional, physically-based, fu lly distributed hydrodynamics/phosphorus dynamics modeling approaches have been suggested as one of the alternatives (Feng and Molz, 1997; Persson et al., 1999; Howell et al., 2005). However, several shortcomings of these modeling approaches have been also reported: (1) requirement of enormous input data and physical parameters, (2) complexity of model calibration/validation, (3) model uncertainty, and (4) requirement of huge time and effort (Fernandez et al., 2006). In spite of the shortcomings, these approaches have been more widely applied in a variety of water bodies, such as stream (DeAngelis et al., 1995), ditches (Janse, 1998), lakes (Chen, 1994; Chen and Sheng, 2005; Chao et al., 2006), reservoirs (Clement et al., 1998), estuaries (Chen, 1994; Wang et al., 2003a and b), and ripari an wetlands (van der Peijl and Verhoeven, 1999; Wang and Mitsch, 2000). This is due to the recent tremendous progress in computer technology, which allows use of both temporal and spatial data resulting in more sophisticated numerical models. However, these approaches ha ve rarely been applied for flow and nutrient dynamics in large-scaled, subtr opical constructed wetlands. This study presents the development and application of a coupled model for flow and phosphorus dynamics with limited field data collected at two constructed wetland systems in South Florida. This study is expected to be a stepping stone for all the future modeling efforts on the development of more advanced flow and

PAGE 22

22 nutrient dynamics models applicable to a variety of constructed wetland systems as well as to the Everglades stormwater treatment ar eas in operation or in preparation. Study Objectives In order to manage the eutrophication problem effectively, it is imperative to have a systematic understanding of th e behavior and fate of target nutrients in an aquatic body. Phosphorus dynamics in constructed wetlands are not only determined by biogeochemical processes, but by flow dynamics and associated so lute transport processes such as sedimentation and resuspension. For example, fast moving fl ow generated by a storm event can lead to resuspension of bottom sediments in a treatmen t cell, instantaneously increasing particulate phosphorus level in water column. The increased particulate phosphor us can be transformed into soluble reactive phosphorus by degr adation or desorption in the water column, which can form favorable conditions for eutrophication. In addition, hydrodynamics a nd solute transport processes often play a role as the dominant control factor on phosphorus dynamics in wetland ecosystems, because a physical process usually occurs more quickly than a biogeochemical one (Chen, 1994). Due to the importance of hydrodynamics and solute transport processes to phosphorus dynamics in constructed wetlands, this study intended to develop a hydrodynamics-solute transport-water quality integrated model fo r flow and phosphorus dynamics in constructed wetlands, directly applicable pa rticularly to large-scaled, tr eatment wetland systems in South Florida, using a pre-developed numerical mode ling program, MIKE 21, developed by the Danish Hydraulic Institute (DHI) for two-dimensional (2-D) free surface flows o ccurring in the areas such as coastal hydraulics and oceanography. The integrated modeling approaches have been mainly developed and applied in various water bodies; however, multidimensional, coupled modeling efforts for flow and phosphorus dynamics in constructed wetlands have been rarely

PAGE 23

23 reported. For this, the numerical model was modified to fit the physical settings of a constructed wetland, and some field data were collected in the OEW Cell 7 and the STA 5 northern flow-way to explore the applicability of a 2-D, physica lly based, fully distributed, dynamics modeling approach in wetland ecosystems. Through model application, the next goals are: (1) to figure out ke y physical factors or processes to regulate the flow and phosphorus dynamics in constructed wetlands, (2) to investigate the effects of hydrodynamics and solu te transport processes on phosphorus cycle in constructed wetlands quantitativel y, (3) to suggest the value or range of key model parameters commonly used in other flow and phosphorus dy namics models, and (4) to obtain useful implications (from data collection to model modi fication level) for future modeling efforts on the development and application of more adva nced coupled model for flow and phosphorus dynamics in constructed wetlands as a tool of predictive model. To accomplish these objectives, first of all, spatial and temporal field data available on physical setting, hydrology/hydraulic, soil, and vege tation of two study areas were collected and analyzed primarily through database website for ti me series hydrology and water quality data (i.e. DBHYDRO, which is a SFWMDs online environm ental database) and geographic information system (GIS) as well as literatu re review or personal contact. Secondly, modified numerical flow and phosphorus models were applied to the fi eld data. Then, simulated model results on hydroperiod, chloride, and phosphorus species were analyzed and compared to measured data. Based on the results of model calibration, valida tion, and sensitivity analysis, some conclusions were elicited.

PAGE 24

24 CHAPTER 2 LITERATURE REVIEW Wetland Hydrology/Hydraulics Hydrology is one of the most fundamental components sustaining the structure and function of wetlands. It controls the hydroperi od of a wetland ecosystem and affects the formation of wetland-specific soil and vege tation types and their temporal and spatial distribution. However, it is not a simple to simulate wetland flow because it is usually very difficult to measure each component of the we tland water budget such as surface watergroundwater interaction and ev apotranspiration precisely. In spite of the difficulty, many researchers ha ve been trying to develop more advanced flow models to mimic the complex, transient flow pattern. In this section, the historical wetland flow modeling efforts are reviewed, which is fo llowed by a brief discus sion on several physical factors affecting wetland hydraulic performance. Wetland Flow Models Kadlec (1994a) suggested three types of mode ls that can explain experimental tracer response curves in free-water wetlands: (1) plug flow with dispersion, (2 ) Tank-In-Series (TIS), and (3) a series-parallel networ k of Continuous Stirred Tank R eactors (CSTRs). In the first model, mixing is assumed to follow an advection-dispersion equation. The second model is based on the conceptual compartmentalization of wetland into a series of well-mixed regions. The third model, called a network m odel, consists of a series of well-mixed zones in a main flow path, which interchanges water and dissolved substa nces with side zones or storage zones that are not in the main flow path. Such a network idea was originally propos ed by Levenspiel (1972) for chemical reactors; however, the same concep t has been applied to various aquatic systems including wetlands (Kadlec, 1994a; Werner and Kadlec, 2000) Werner and Kadlec (2000)

PAGE 25

25 extended Levenspiels Model G to the case of an infinite side tanks called zones of diminished mixing (ZDMs), represented by CSTRs connected to main channel represented by a plug flow stage. Although those models are not site-specific and can simulate more realistic residence time distribution of non-ideal flows of constructed wetlands because various natural and constructed wetlands actually show intermediate flow char acteristic between plug flow and complete mixing, it is difficult to find any physical meaning for th e key parameters of those models, and they all have an inherent limitation as a 1-D, conceptual model, which is not enough to describe 2-D, transient wetland flow characteristics. The math ematics and detail description of each model were well discussed by Kadlec (1994a), Kadl ec and Knight (1996), a nd Werner and Kadlec (2000). In addition to these conceptual models, th ere are several wetland hydrologic models. Konyha et al. (1995) described a continuous hydrologic model called SWAMPMOD that can be used in the design of constructed wetlands. In th e paper of Su et al. (2000), a semi-distributed hydrological model (modified SLURP) was used to simulate water level fluctuations over a 28year period for a 3-ha prairie wetland in Sa skatchewan. A spatial hydrology model (HYDROMODEL), which was developed with Arc/Info Macro language routines in the GRID environment, was used to assess the spatial ex tent of the Suwannee Rivers sill effects on the swamp hydrology, to predict hydrologi cal changes expected with si ll removal, and to examine the sills role in affecting vegetation distribut ion in the Okefenokee Sw amp, Georgia (Loftin et al., 2001). In this model, the 22 vegetation type s were regrouped and co ded to assign Mannings coefficients to each model grid to reflect th e change of surface water flow rates through vegetation. Potential evapotrans piration was calculated by usi ng Thornthwaites equation and was applied being considered as actual evap otranspiration. Groundwater exchange was not

PAGE 26

26 modeled because the exchange was assumed to be negligible in the swamp. System sensitivities to changes in topography manipulation, outflow, a nd evapotraspiration rate were analyzed, and the results were discusse d (Loftin et al., 2001). Water budget models form another type of wetland hydrology model (Walton et al., 1996, Arnold et al., 2001; Dadaser-Celik et al., 2006). A Wetlands Dynamic Water Budget Model (WDWBM), which provides magnitude for the wa ter budget components as well as hydraulic components such as water depths, discharges, an d flow velocities throughout the modeled area, was developed and applied to an Arkansas sw amp (Walton et al., 1996). This model includes three dynamically-linked modules: surface water, vertical processe s, and horizontal groundwater flow module. On the other hand, the modified Soil and Water Assessment (SWAT) model, one of the water budget models, was used to evalua te suitability of mitigation bank design along Trinity River, Texas (Arnold et al., 2001). This model was used to assess whether storm flow and baseflow could maintain the proposed bottomland wetland ecosystem. In terms of hydrology-ecology dynamics, Poiani and Johnson (1993) developed a spatial, deterministic model to simulate the hydrologi c and vegetation dynamics based on data from semi-permanent prairie wetlands lo cated in North Dakota that were slightly brackish and located at groundwater discharge or flow-through areas In the hydrologic submodel programmed in BASIC, the surface water levels were simulated ba sed on input data of precipitation, runoff, and potential evapotranspiration. Gr oundwater processes were not c onsidered, and evpotranspiration was calculated by applying Thornthwaites method (potential evaporati on) and transpiration scalar. The vegetation submodel programmed within a GIS estimated the amount and distribution of emergent plant cover and open water. Water elevation data calculated from the hydrologic submodel were used as input data to th e vegetation submodel. An actual distributions

PAGE 27

27 of vegetation types observed from aerial photogr aphs were compared to the results of each simulation using actual and calculated water leve ls to assess the combined performance of both submodels, and implications and applicabil ity for ecological wetland managements were discussed (Poiani and Johnson, 1993). Several examples of other wetland flow models, including multidimensional wetland hydrodynamics model, are summarized. The loca l hydrology of a cypre ss pond/flatwood forest system was modeled using WETLANDS, a multid imensional water flow and solute transport numerical model that provides dynamic linkage s among pond water, unsaturated soil zones, and groundwater (Mansell et al., 2000). The SIDRA-M AGE model, which coupled the field-scale hydrology model SIDRA with the channel routi ng model MAGE, was used to simulate the change in subsurface and open channel flows caused by the increase of drained areas in a freshwater marsh for eight dr ainage schemes (Giraud et al ., 1997). Development of a 2-D, diffusion-based, wetland hydraulics model (W ETFLOW) was presented by Feng and Molz (1997). This model was applied to low gradient, 1-D (for laboratory e xperiment) and 2-D flow fields (for a wetland pond), respec tively, and the results showed th e significance of the effect of topography changes on flow depth and veloci ty variations. Finally, MODFLOW wetland simulation module was developed by Restrepo et al. (1998) and applied with the groundwater flow model in the north-central Miami-Dade County, Florida (Wils nack et al., 2001). The results of this application showed that the module was suitable fo r simulating regional wetland hydrology within the Everglades. Wetland Hydraulic Performance In order to simulate the removal or retenti on of a variety of pollutants in constructed wetlands effectively, understand ing those hydraulic characterist ics controlled by the physical setting, such as shape, bathymetry, slope, exis tence of hydraulic struct ure and its type, and

PAGE 28

28 vegetation, is essential because flow basically plays a role of medi um to carry pollutants to biota and soil. The residence time distribution (RTD) determined by tracer experiment in a wetland has been used as an important tool to unde rstand wetland hydraulics (K adlec and Knight, 1996; Persson, 2000; Werner and Kadlec, 2000; Ho lland et al., 2004). Specifically, through the moment analysis of an RTD, several paramete rs for measuring and comparing the hydraulic performances of various constructed wetlands are elicited (Persson et al., 1999; Martinez and Wise, 2003b). Short-circuiting flow is one of the hydrauli c characteristics most commonly observed in various constructed wetland systems. In genera l, naturally-formed or man-made low elevation areas exist in wetlands, even within constr ucted wetlands. Water flow makes these zones interconnected and canalized, which show deeper water depth and less vegetation density (less hydraulic resistance) compared to the adjacent shallow-depth, flat-bottomed, vegetative areas, causing hydraulic short-circuiting (Dierberg et al., 2005). They re duce hydraulic retention time in a treatment cell and leave large isolated, st agnant areas only tangentially affected by the incoming water, decreasing the treatment efficienc y. Hence, it is one of the biggest obstacles to successful treatment wetland design and management (Persson, 2000). Several researchers have reported the factors a ffecting short-circuiting flow in constructed wetlands (Thackston et al., 1987; Kadlec and Kn ight, 1996; Persson et al., 1999; Persson, 2000; Koskiaho, 2003; Persson and Wittgren, 2003; Holland et al., 2004; Dierberg et al., 2005; Jenkins and Greenway, 2005; Wrman and Kronns, 2005) as follows: Wetland shape or the le ngth-to-width ratio Wetland depth or water level Wetland flow rate Wetland bathymetry or basin morphology Wetland vegetation: type, dens ity, and spatial distribution Wetland hydraulic structur es: type and location

PAGE 29

29 Wetland internal structures such as subsurface berms and islands Wind Phosphorus Dynamics in Wetlands Many aquatic bodies, including surface flow c onstructed wetland systems, are phosphoruslimited. The Everglades in South Florida, a huge oligotrophic marsh with historical phosphorus levels around 10 ppb as TP, is one of the most representative phosphorus limited ecosystems in the world. Under this condition, excess phosphorus loading breaks up the ecological balance of an ecosystem set over a long time, increasing prim ary production of various plants, which are not originally dominant. In order to resolve these problems, cons tructed wetlands are increasingly being used to reduce the external loading from uplands prior to entering downstream aquatic water bodies effectively (Gopal, 1999; Carleton et al., 2001). Figure 2-1 shows a conceptual diagram of phosphorus dynamics in wetlands. The general characteristics of phosphorus dynamics in construc ted wetlands are not di fferent than those of natural wetlands. However, the main concern in constructed wetlands is to enhance phosphorus retention and inhibit re lease into water column to meet the goal mentioned above. Phosphorus retention/release processe s are a natural part of phosphorus cycle in wetlands. Therefore, it is extremely important to understand the phosphorus dynamics in wetlands systematically to maximize the treatment efficiency and to develop an optimized treatment model. After a brief summary on phosphorus retention/release in wetl ands, historical phosphorus retention modeling efforts are reviewed, followed by a brief disc ussion on two traditional modeling approaches. Phosphorus Retention/Release In general, phosphorus retention mechanisms can be classified into two categories: (1) biotic processes such as the uptake by m acrophytes, periphyton, and microorganisms and (2) abiotic processes including sorp tion and exchange by wetland so ils, chemical precipitation, and

PAGE 30

30 mechanical sedimentation (Reddy et al., 1999). As retention processes are differently regulated by the analytical forms of phosphorus, which are commonly categorized into four groups: (1) dissolved inorganic phosphorus (DIP) or solu ble reactive phosphorus (SRP), (2) dissolved organic phosphorus (DOP), (3) particulate inor ganic phosphorus (PIP), and (4) particulate organic phosphorus (POP), it is essential to cons ider all chemical fractio ns of total phosphorus (TP) when we need to understand the specific retention mechanisms more systematically. For instance, only SRP is generally considered bioavaila ble. Hence, when it comes to biotic retention processes, SRP concentration can be more critic al than TP. On the othe r hand, physical retention mechanisms such as sedimentation and resusp ension of phosphorus are highly associated with only particulate phosphorus (PP) concentration (Chen and Sh eng, 2005). Therefore, the key mechanism and efficiency of phosphorus retention in constructed wetlands are not always the same but highly variable according to the charact eristic of phosphorus chemical fractions in input water and the design and perf ormance of constructed wetlands. Once temporally retained, phosphorus is ofte n released as a form of dissolved and particulate phosphorus into water column through biotic and abiotic proc esses. This internal loading often constitutes a substantial part of the total loading, preventing the fulfillment of established water quality criteria even though external loading is significantly reduced (Fisher and Reddy, 2001; Sndergaard et al., 2001). First, the change of physicochemical parameters, such as high pH, reduced redox, and high temper ature, accelerates disso lved phosphorus release into the water column of constructed wetla nds. Anaerobic release of Fe-bound phosphorus is well known when redox potential is decreased (Gchter and Meyer, 1993; Reddy et al., 1999) and high temperature increases th e bacterial activit y, giving lower redox pot ential and a higher production of enzymes (Kamp-Nielson, 1975). In addi tion, it is well known that a sharp diffusion

PAGE 31

31 gradient, desorption and dissoluti on of precipitates and complexes, and enzymatic hydrolysis of organic phosphorus (decomposition of vegetation a nd litter) cause biotic or abiotic release of dissolved phosphorus. On the other hand, releas e mechanisms of particulate phosphorus into water column in wetlands are: (1) mechanical resuspension by wind and current, (2) biological resuspension by bioturbation from benthic or ganisms or through the release of wetland gas bubbles like CH4 and H2S, and (3) sloughing of periphyton into the water column (Reddy et al., 1999; Sndergaard et al., 2001). Phosphorus Retention Models For optimizing the design and management of constructed wetlands and predicting the performance of the treatment wetlands under vari ed conditions such as altered hydroperiod and vegetation type/density, a modeling approach is essential. In th e review paper of Reddy et al. (1999), classification and examples of various phosphorus retention models in stream and wetland systems, ranging from simple correlations to complex dynamics models, were systematically reviewed. An empirical mass balance approach based on in put-output analysis is the simplest model used to describe phosphorus retention in various wetland sy stems (Kadlec and Newman, 1992; Kadlec and Knight, 1996; Reddy et al., 1999). Next, first-order kinetic models referred to as k C* model or the Vollenweider mo del have been most frequently used to explain exponential decrease of phosphorus concentration along the flow direction that has be en observed at longterm field scale (Mitsch et al ., 1995; Walker, 1995; Kadlec, 1999a and b; Kadlec, 2000; Black and Wise, 2003) and to compare the efficiency of treatment wetlands (Kadlec, 1994b; Kadlec and Knight, 1996; Carleton et al., 2001). This model is based on the assumptions that phosphorus removal rate is directly proportional to the c oncentration of phosphorus at a given location and the first-order kinetic constant, k called volumetric or areal phos phorus uptake rate coefficient,

PAGE 32

32 and lumps all phosphorus retention processes oc curring in constructed wetlands (Kadlec, 1997). Hence, the performance of a constructed we tland for phosphorus removal is expressed by a single constant. In addition to the relatively simple two models mentioned above, there are several phosphorus models developed either to describe a specific pro cess in phosphorus cycle or to apply phosphorus model for specific purposes. Sedime nt flux models are specifically designed to describe phosphorus flux from bottom sediment s (Di Toro and Fitzpatrick, 1993; Lijklema, 1993). Various phosphorus models have also been developed primarily to describe changes in ecosystem by environmental conditions (Kad lec and Hammer, 1988; Mitsch and Jrgensen, 1990) or incorporated into watershed model to s imulate phosphorus transport in a rural or urban basin; for example, SWMM (Huber and Dick inson, 1988), HSPF (Donigian and Huber, 1990), and SWAP (Arnold et al., 1994). These models are useful to simulate a specific process associated with the phosphorus transport and dyn amics; however, they are insufficient as a comprehensive modeling tools for complex phos phorus dynamics among water column, soil, and vegetation in constructed wetlands. Water flow is a basic component in most so lute transport or retention models because water is the medium where most solutes includi ng phosphorus are transporte d or transformed. In addition, in terms of treatment wetland efficien cy, as manipulating the hydrologic regime to increase phosphorus removal efficiency is a desi rable strategy for cons tructed wetlands (Wang and Mitsch, 2000), understanding phosphorus dyna mics based on accurately-simulated wetland flow is essential. It provides a useful tool to understand phos phorus behavior under non-steady state, real flow conditions, to test various hypotheses according to changed hydrologic regime and ultimately making more realistic predictions.

PAGE 33

33 Hydrodynamic transport models, historically developed mainly for streams or rivers, however, were seldom applied to wetland system s (Reddy et al., 1999). They can be applied for phosphorus retention simulation in constructed wetlands using the sa me principle once they are coupled to appropriate water qu ality dynamics models, which have been developed to simulate the transformations of solutes or pollutants through the interactions with air, sediment, or algae, in river and lake systems. Under this flow-wat er quality integrated dynamics modeling approach, the cycle or retention mechanisms of each phosphorus species described as appropriate mathematical expressions are coupled into a r eaction term of an advection-dispersion equation (Chen, 1994; Wang and Mitsch, 2000; Chen and Sh eng, 2005). Most of the water quality models, including WASP4 (Ambrose et al., 1988), CE-QUAL-RIV1 (Environmental Lab., 1990), and RCA (HydroQual, 1995), have been used to simula te phosphorus transport/ retention coupled to primarily 1-D hydrodynamic models (Reddy et al ., 1999). These integrated approaches are considered more advanced and comprehensive th an other models reviewed earlier because they allow one to describe and simulate details of most internal phosphorus retention processes in wetlands, which have been traditionally regard ed as a black box, on the basis of hydrodynamic simulation. Management vs. scientific modeling approach In summary, there have been two broad type s of approach as to traditional phosphorus retention modeling: (1) management modeling a pproach and (2) scientific modeling approach. The first approach is a way to describe every re tention process occurring in constructed wetlands through a certain lumped parameter as shown in Vollenweider and first-order k C* model. This approach generally requires line arly collected samples along the flow direction from inlet to outlet point to calibrate the parameter. Often, in case that the hydraulics within constructed wetlands is already determined through solute tr ansport modeling based on conservative tracer

PAGE 34

34 experiment, only input and output phosphorus concentration prof iles are sufficient to estimate the value of the parameter. Although the former approach is simple to repr esent and compare the efficiency of nutrient removal in constructed wetlands for management purposes and has been empirically proven to some extent, it does not provide any detailed information about each physical, biological, and geochemical retention process. That is, the physic al meanings of lumped parameters are not so clear that they can not provide any scientific information on identification of a key retention mechanism or quantification of th e retention processes. Furtherm ore, despite the popularity of use due to its simplicity, the first-order treatment wetland model includes several unrealistic hydraulic assumptions, such as uniform flow, st eady state, and plug-flow reactor conditions, inherently limiting applicabil ity of the model under real hydrodynamic conditions showing transient, highly dispersive, and sp atially heterogeneous flow charac teristics. As a result, Kadlec (2000) and Black and Wise (2003) a ddressed the inadequacy of the first-order model in that the key parameters such as uptake rate coefficient, k and apparent background concentration, C* were not independent on hydraulic loading rate and inlet phosphorus con centration. Therefore, the first-order model, which has been most comm only applied as the desi gn basis of constructed wetlands for phosphorus removal like the Everglades STAs (Walker, 1995), is not sufficient as a forecasting tool for detailed phosphorus dynamics in constructed wetlands. On the other hand, the scientific modeling ap proach is often too complex and contains unnecessary details because it phys ically describes retention pr ocesses by identifying numerous wetland compartments such as litter, sediment, live or dead biomass a nd regulating appropriate transfer among the compartments (Kadlec, 1997). Hence, it usually requires a lot of data for model calibration/validation and numerous coefficients or constants that have to be determined

PAGE 35

35 to regulate the relationships between state variables. Regarding the shortcomings of these dynamics modeling approaches, Fernandez et al (2006) indicated three shortcomings: (1) it usually requires extensiv e model input data and physical para meters. In many cases, some of the model input data or parameters are not available, which makes it difficult to develop the model. This also makes it hard to calibrate and validat e the model. (2) the underlying uncertainties in the parameterization lead to uncertainties in mode l prediction. (3) time and effort required to run those models are usually considerably more th an needed by application of conceptual, highly simplified lumped-parameter models. In most situ ations with limited field data, these dynamics modeling approaches may not be required to estimate the long-term behavior of phosphorus retention in a wetla nd (Reddy et al., 1999). Although there have been always trade-offs between these two types of approaches, the latter approach is adapted in this study because it is impossible with the former approach to reach the new paradigm mentioned by Kadlec (2000).

PAGE 36

36 Figure 2-1. Phosphorus dynamics in wetlands.

PAGE 37

37 CHAPTER 3 HYDRODYNAMICS-WATER QUALIT Y INTEGRATED MODEL SETUP Wetland Model Dimension For simplicity, a one-dimensional representation of constructed wetland surface has been applied, sometimes enumerating several reasons not to use higher dimension models: (1) relatively simple geometry of man-made treatment wetland systems and (2) insufficient data to validate a higher dimension model (Kadlec and Ha mmer, 1988). As a result, most 1-D modeling approaches to wetland flow primarily are ba sed on conceptually-lumped models, mainly expressed as mixtures of plug flow reactor (PFR) and/or CSTR models (Kadlec and Knight, 1996). However, sheet flow characteristics in cons tructed wetlands can not be described fully using a 1-D flow model, even though flow simula tion axis is chosen to correspond to the main direction of bathymetry slope in a constructed wetland with simple rectangular geometry. This is proved by the fact that the effective volumes of currently operating constructed wetland systems are highly diverse as shown by Martinez and Wise (2003b). Many treatment wetlands really experience spatially different shor t-circuiting and long time retenti on, which is mainly attributed to the topographical and vegetative heterogeneit y in a constructed wetl and. Basically, a 1-D model does not provide any specifi c information on the spatial heterogeneity of flow within a treatment cell. Of course, it also has been successfully described th rough the BTC fitting of tracer test with 1-D, conceptual wetland m odels; however, without a tracer experiment, the model calibration is not possible an d the approach is primarily a pplied to only single input-outlet systems. For example, in the case of treatment cells with multiple inlet and outlet points, commonly observed in South Florida STAs, appli cation of 1-D model is significantly limited.

PAGE 38

38 On the other hand, flow and nutrient dynamics in wetlands are different from those in deep water bodies such as lakes and estuaries, which te mporally show dramatic vertical heterogeneity of nutrient level in water column. Even though ther e is a vertical differe nce of flow velocity, physicochemical parameters, and nutrient concentr ations in the wetland water column (Chimney et al., 2006), water depth in most constructed wetla nds is relatively so shallow, compared to the one of lakes or reservoirs, that stratification is usually neglected. Most of all, it is unlikely that enough data would be available at any well-studied site to calibrat e and implement a 3-D model. Therefore, in this study, a 2-D, physically based, distributed flow and water quality dynamics modeling approach is selected. Overview of Hydrodynamics-Water Quality Integrated Model: MIKE 21 In this section, a hydrodynamics -water quality integrated num erical model, MIKE 21, is reviewed based on its th ree interconnected fund amental submodules for environmental hydraulic simulation: HD, AD, and ECO Lab. Although the MIKE 21 HD and AD models were developed for 2-D free surface flows occurring in the ar eas such as coastal hydraulics and oceanography, they seem appropriate for simulation of hydrau lic and environmental phenomena in wetlands because basic hydraulic principles are almost identical and the model can be easily adjusted to the specific wetland conditions th rough relatively simple conversion of a few model options. Basic Parameters Key basic parameters to be determined in MI KE 21 are as follows: bathymetry, simulation time step, boundary, source and sink, and flooding and drying depth. To develop a bathymetry map, the map project ion, geographical position of grid origin, and model domain size (length of model grid in x and y direction) of study area should be first determined. Based on the generated blank bathymet ry file, topographic elevation data measured in field are assigned. In most cases as the elevations are reported as a point data type and the

PAGE 39

39 number of measurement is not enough to cover the entire model domain, several interpolation schemes are usually required. MIKE 21 enables th e model users to use on e of three types of inverse distance weighting (IDW) interpolati on scheme. As commonly used interpolation schemes like Kriging are not supported by MIKE 21, users have to inte rpolate their point measurement data set to obtain a raster-typed conti nuous bathymetry map us ing other computer programs. MIKE 21 requires the user to specify either the water surface elevati on or the flux at all open boundary points. Unlike other aquatic bodie s, surface inflows and outflows in most constructed wetlands are regulated by point source/sink-type hydraulic structures such as pump stations, weirs, and culverts. Therefore, the model boundary is closed to represent the berm around treatment wetland cells in this study (no surface flow at boundary), and the inflows and outflows are instead specified using source and sink option at each grid cell corresponding to the location of inflow and outfl ow hydraulic structures. Also, th e source and sink option is used to represent net groundwater inflow or outflow. One of main characteristics of MIKE 21 flow model is that flooding and drying conditions of wetland can be represented with respect to the change of hydroperiod using flood and dry option. According to SFWMD (2005b), detention de pth is defined as a ponding depth at a grid cell below which no water transfer into the adj acent grid cell is allo wed even if a hydraulic gradient exists. In the SFWMM model, around 0.1 ft was suggested for the depth in various wetland types. Likewise in MIKE 21 HD modul e, flooding and drying de pths have to be determined. Above the flooding depth, hydraulic roughne ss-controlled water flow is horizontally transferred along the hydraulic gradient; however, below th e flooding depth, water does not

PAGE 40

40 move even if a hydraulic gradient exists. If water depth is less than the drying depth, the grid cell is considered as a completely dried area. Hydrodynamics (HD) Module The MIKE 21 HD module is a numerical model fo r simulation of water levels and flows in free surface water bodies. It simula tes transient 2-D flows in a vertically-homogeneous layer. This module is very important for developmen t of flow-water quality integrated dynamics models because the MIKE 21 HD time series output results are used as input for other MIKE 21 modules such as AD and ECO Lab. Governing equations The HD module is based on the depth-averaged Saint-Venant equations (Equation 3-1 to 33) describing the evolution of the water level and two Cartesian velocity components u and v of which solutions are numerically obtained from a finite difference form of the equations (Rung and Olesen, 2003). The following equations, which are depth-aver aged mass and momentum conservation, are the general governing equations for MIKE 21 HD model. t d y q x p t (3-1) 0 ) ( ) ( ) ( 12 2 2 2 2 a w x q xy xx wp x h fVV h y h x h C q p gp x gh h pq y h p x t p (3-2)

PAGE 41

41 0 ) ( ) ( ) ( 12 2 2 2 2 a w y p xy yy wp y h fVV h x h y h C q p gq y gh h pq x h q y t q (3-3) where h(x, y, t) d(x, y, t) and (x, y, t) are water depth (= d m), time varying water depth (m), and surface elevation (m), respectively. The othe r symbols used in Equation 3-1 to 3-3 are as follows: p and q(x, y, t) = fluxes in x and y directions ( p = uh and q = vh m2/s; u and v are depth averaged velocities in x and y directions), C(x, y) = Chezy resistance (m1/2/s), f(V) = wind friction factor, V Vx, and Vy(x, y, t) = wind speed and the components in x and y directions (m/s), (x, y) = Coriolis parameter (1/s), pa(x, y, t) = atmospheric pressure (kg/m/s2), w = water density (kg/m3), and xx, xy, and yy = effective shear stress components (kg/m/s2). Although all these components can affect flow dynamics in the real world and be implemented in MIKE 21, several assumptions were made in this study to simplify the HD model by considering dominant impacting factor s in constructed wetland setting. In this study, simplified forms of the above th ree partial differential equati ons were based on the following assumptions: The fluid is incompressible and has uniform density. Bottom slope is small, and the bathymetry is fixed. Chezy or Mannings equation, which originally applies to steady uniform flow, can be used to describe hydraulic resistan ces due to bottom and/or vegetation. Coriolis effect, surface resistance by wind, and shear stresses by tur bulence are ignored. Wind or wave action in shallow aquatic bodies pl ays an important role in sediment dynamics (Chen, 1994); however, this effect can be redu ced in densely-vegeta ted constructed wetland systems like EAV treatment cells because emer gent vegetation dampens wave energy and

PAGE 42

42 shelters the water surface from wind stress (Nepf, 1999; Braskerud, 2001). In addition, it was reported that SAV was effective to limit wi nd-driven sediment and associated nutrient resuspension (Dennison et al., 1993; Barko a nd James, 1997; Horppila and Nurminen, 2003). To integrate the numerical formulation of the above equations for mass and momentum conservation in the space-time domain, MIKE 21 HD uses a so-called Alternating Direction Implicit (ADI) finite difference scheme (DHI, 2005b). Refer to DHI (2005b) for in-depth description of numerical formul ation and solution algorithm app lied in the MIKE 21 HD model, including mathematical and numerical background for each of the terms in Equation 3-1 to 3-3. Model conditions and parameters Key hydrodynamic parameters to be determined in MIKE 21 are as follows: initial surface elevation, boundary condition, source and sink in cluding precipitation and evaporation, and hydraulic resistance. The 2-D map for initial surface elevation is generated using the IDW linear interpolation scheme of MIKE 21 from field data collected at several measurement points in study area on the simulation start date. If enough data are not avai lable, constant water level can be simply specified. In this study, there is a no flow bounda ry condition because any open boundary around constructed wetland cell is not assi gned at the stage of Basic pa rameters setup. Instead of open boundary, the time series magnitude (m3/s), velocity (m/s), and in let direction of surface water inflow are specified at each grid point assign ed as an isolate source, and the time series magnitude of surface water outflow is specified at each grid point assigned as an isolate sink. All these data are based on field measurement and estimation. For groundwater flow, if net groundwater seepage is dominant, sink cells are assigned at the groundwater recharge areas with the time series discharge (m3/s); on the other hand, if net groundw ater inflow is dominant, source

PAGE 43

43 cells are assigned at the groundwater discharge areas with the time series discharge and velocity. Precipitation and evaporation data are also specified here. Usually there are three types for the assignment: (1) tempo-spatially homogeneous, (2) temporally vari ous but spatially homogeneous, and (3) tempo-spatially heter ogeneous. In many cases, the second option is commonly used if the study ar ea is not spatially huge. Free surface water movement in natural or constructed wetlands, sometimes called sheetflow or overland flow, is usually gove rned by the hydraulic resistance (roughness). Mannings or Chezys coefficien t is commonly used to quantif y hydraulic resistance caused by bottom sediment and vegetation stem/litter friction. Either a constant value applied over the whole model area or 2-D map-typed values should be determined. Selection of roughness coefficient values appropriate in constructed wetland systems will be discussed in Chapter 4 and 5 in more detailed. Advection-Dispersion (AD) Module The MIKE 21 AD module is used to simulate th e transport of solutes subject to advectiondispersion process. In terms of the decay ch aracteristics of a com ponent, either no decay (conservative solute) or linear decay (first or der decay solute) component is usually assigned. Governing equation The MIKE 21 AD module numerically solves the advection-dispersion equation for substances in water column in two dimensions. Linear decay and source/sink term are included with advection and dispersion term of a com pound. The governing partia l differential equation for the AD module is as follow: S C h F y C D h y x C D h x vhC y uhC x hC ty x ) ( ) ( ) ( (3-4)

PAGE 44

44 where C Dx/ Dy, F and S are a solute concentration, x and y directional dispersion coefficient (m2/s), first-order decay rate co efficient (1/s), and sources/s inks, respectively. The above equation is numerically solved using a third order finite diffe rence explicit scheme, QUICKEST, for 2-D AD equation. The details of numerical algorithm and solu tion technique applied in the AD model are described in Ek ebjrg and Justesen (1991). Model conditions and parameters Key advection-dispersion parameters to be de termined in MIKE 21 are: component and the type (conservative or reactive), initial concen trations, boundary conditions, source and sink, and dispersion coefficients. In the same way as the specification of initia l surface elevation reviewed earlier, the 2-D map for initial concentration of components or a constant valu e is required. In the STA 5 northern flow-way modeling study, chloride and phosphorus concentrations measured at seven sampling points on the simulation start date were used to make the 2-D map using the IDW linear interpolation sche me. For each component, a constant or time series inlet component concentrations must be specified at each source grid cell defined in the model. The dispersion coefficients in x and y direc tion are among the most important parameters in the AD module. Dispersion is a general term to refer to the scattering of a solute in fluid due to random processes (molecular or turbulent diffusi on) as well as the effect of spatial velocity gradients. In MIKE 21, there are two options to assign the coefficients: (1 ) independent of flow velocity and (2) proportional to the flow velocity. If the former is selected, dispersion coefficients in both x and y direction must be sp ecified as a form of either 2-D map type or a constant value applied to all points in the m odel domain. On the other hand, if the latter is chosen, x and y directional dispersivities must be assigned. In this case, the dispersion

PAGE 45

45 coefficients can be continuously changed during th e entire simulation period in accordance with the flow velocities calculated by the HD module at each simulation time step. Model instability problems occur when too high values of dispersion co efficients are used (DHI, 2005a). Therefore, a reasonable range of di spersion coefficient must be assigned to avoid instability when the first option (proportional to the flow velocity) is selected. Given model grid size and time step, the upper limit of dispersi on coefficient is determined by the following equation (DHI, 2005a), which is related to the Courant number ( Cx = u t/ x and Cy = v t/ y). 5 02 2 t y D x Dy x (3-5) In the model application to the OEW Cell 7 (Chapter 4), a 2-D map of dispersion coefficient is generated based on a 2-D map of flow velocity at a sp ecific time step using the first option; on the other hand, the s econd option is applied in the m odel application to the STA 5 northern flow-way (Chapter 5 and 6). Each disper sivity in the x and y direction was determined through calibration on chloride transport simulati on, and the upper limit of dispersion coefficient was calculated by Equation 3-5. Water Quality/Ecological Engi neering (ECO Lab) Module ECO Lab is a MIKE 21 module developed to simulate water qualit y, eutrophication, heavy metals, and ecology (DHI, 2004a). There are two options to use this module: (1) using the predefined ECO Lab templates and (2) deve loping users own ecosystem model through modifying the original templates or creating a new template. The demand on customized water quality/ecosystem models is generally increasi ng because the pre-developed models sometimes have too many unnecessary components or lack sp ecific processes. ECO Lab is easy to modify (or create) and implement the math ematical descriptions of envi ronmental/ecological processes, allowing the customized model description in accordance with the variety of ecosystem.

PAGE 46

46 It can simulate dissolved substances, particulat e matters of dead or living materials, living biological organisms, and other components and de scribe physical and biogeochemical processes and interactions among ecosystem state variable s. Coupled to the HD and AD module described above, transport processes based on AD module are integrated in the ECO Lab simulation. Hence, state variables in ECO La b can either be transported by advection-dispersion processes based on hydrodynamics, or have a fixed feat ure like rooted vegetation (DHI, 2004a). The dynamics of mobile ECO Lab state variab les can be expressed by a set of solute transport equations written as: P S y c D x c D y c v x c u t cy x 2 2 2 2 (3-6) where P denotes ECO Lab processes. The linear or non-linear relationships among the state variable are coupled thro ugh the ECO Lab source term P The above equation is numerically solved using an explicit time integration sche me (Euler, 4th order Runge Kutta, or 5th order Runge Kutta with quality check) when the concentration of the next time step is calculated. Refer to DHI (2004a) for the details of the numerical solution technique. Key ECO Lab parameters to be determined in MIKE 21 are: integration method, initial conditions, state variables, consta nts, forcing functions, and proce sses. The way to set the initial conditions of state variables follows the same pr ocedure as specifying the initial concentrations of components in the AD module. Each state vari able is expressed by a series of processes, which link the state variables via mathematically -described physical, chemical, or biological reactions. Each process is usually represen ted as a kinetic form and defined through a combination of state variable constants, and/or forcing functions. Time series input concentration profiles of each state variable and the initial condition on model domain are usually dependent on field measurement; however, determining the consta nts is not a simple

PAGE 47

47 task. These are usually specified (1) through fi eld or laboratory experiments, (2) from the literature sources, (3) by model calibration, or (4) using appropriate estimation or valid assumption. The next section presents a phosphorus dynami cs model for large-scaled, subtropical constructed wetlands, which is developed using ECO Lab. The model will serve as a practical example to show how each ECO Lab parameter is defined and interconnected with one another. Phosphorus Dynamics Model Phosphorus concentrations in constructed we tlands change in time and space due to seasonal climate changes, input/output mass disc harges, surface water-groundwater interactions, biogeochemical reactions, as well as flow dynami cs. To simulate the dynamics accurately, every controlling factor should be a ppropriately incorporated in th e dynamics model. However, many assumptions are generally made in numerical m odeling approaches for phosphorus because some processes in the cycles are very complicated and not fully understood. Therefore, it is neither possible nor effective to consid er all the components in wetla nd ecosystems and the associated processes and parameters in a model. This show s that it is required to optimize the dynamics model as simple as possible according to the application purpose and scope of a model. Model Scope A phosphorus dynamics model developed usi ng ECO Lab, presented in the following sections, has a specific aim to apply the model to large-scaled, subtr opical constructed wetland systems such as the STAs in South Florida. The application results of phosphorus dynamics model to one of the STA systems are discussed in Chapter 6. Figure 3-1 illustrates a conceptual diagra m of constructed wetland phosphorus dynamics model applied in this study. Compared to the traditional diagram shown in Figure 2-1, a couple of differences are observed:

PAGE 48

48 Phosphorus dynamics in the floc layer are consid ered. The floc layer, mainly consists of decaying plant materials, exists between water column and the upper soil layer, playing a critical role in regulating th e level of phosphorus species in water column. In this study, particulate phosphorus in the water column is directly deposited into the floc layer, not the soil layer and mechanical and/or non-mechanical resu spension are also generated from the floc layer toward the water colum n. The phosphorus cycle in the floc layer is assumed to be similar to that in the upper soil layer. Phytoplankton phosphorus is removed from the list of state variables. According to McCormick et al. (1998), phytoplankton, consis ted primarily of re suspended periphyton cells, was sparse in water samples collected from both eutrophic and oligotrophic sites in the northern Everglades. They also repor ted that phytoplankton productivity was generally undetectable at each site. Therefor e, phytoplankton is not considered a state variable in this study. Of the three types of periphyton, benthic pe riphyton (epipelon) is not considered. The ecosystem of STAs is more similar to one of eutrophic area than oligotrophic area in Water Conservation Area (WCA) 2A due to the high phosphorus loading rate. The field study results of McCormick et al. (1998) revealed that pe riphyton biomass was low and limited to epiphyton and metaphyton (epipelon was undetectable) in openwater areas at eutrophic sites and was undetect able in cattail stands in th e northern Everglades. Based on their finding, no processes related to benthic periphyton are consider ed in this study. Although it is generally known that EAV roots mainly uptake porewater SRP in soil layer of 20 to 30 cm depth, in this model, EAV is assumed to absorb the SRP only in 10 cm upper soil layer. In other words, deep soil layer is not consider ed in this study. The phosphorus kinetic pathways shown in Figur e 3-1 are incorporated into the reaction term, P in Equation 3-6. Like many present wate r quality models, th e phosphorus dynamics model used first order kinetics fo rmulations to describe most of the transformation processes, except for those associated with biota growth and sedimentation/resuspension. For the growth rate of macrophytes and periphy ton, Michaelis-Menten kinetics fo rmulations were used. The phosphorus dynamics model is calibrated and fi nally validated by comparing the simulated concentrations of each critical state variable w ith the measurement data, which are described in Chapter 6. The following sections provide detailed de scription of ECO Lab phosphorus dynamics model.

PAGE 49

49 State Variables Table 3-1 shows 12 state variables used in the ECO Lab phosphorus dynamics model. They are classified into three spatial domains in constructed wetland ecosystems (water column, floc layer, and upper 10-cm soil layer). Only four state variables in water column (SRPw, DOP, PIP, and POP) are mobile by horizontal solute transport processes, while the other two state variables in water column (Pperi and Pmacro) are fixed at each grid cell during the simulation. Six state variables in floc and soil layer are not transported by the AD module. Due to the difference in physico-chemical conditions, the phosphorus cycles in water column and in floc/soil layers are different. The floc and soil layer phosphorus components are IP, OP, and porewater SRP, respectively. Although the more detailed soil phosphorus fractionation data are often available in South Florida wetlands, they are still not enough to verify the tempo-spatial variation of floc/soil phosphorus spec ies in the dynamics model. Since horizontal transport is generally very minimal in the floc and so il layers, compared to that in water column, mass flows of phosphorus co mponents are primarily vertical. The kinetic pathways of 12 state variables ar e as follows, and each specific transformation process is described in the later section: SRPw: Atmospheric deposition + DOP minerali zation + Desorption of PIP Adsorption into PIP Uptake by macrophytes Uptake by periphyton Diffusion from floc layer into water column. DOP: Atmospheric deposition + Desorption of POP Adsorption into POP DOP mineralization. PIP: Atmospheric deposition + Adsorption from SRPw Desorption of PIP Sedimentation into IPf + Resuspension from IPf. POP: Atmospheric deposition + Adsorpti on from DOP Desorption into DOP + Mortality of macrophytes + Sloughing of Pperi Cohesion into Pperi Sedimentation into OPf Resuspension from OPf. Pperi: Uptake by periphyton + Cohe sion of POP Sloughing of Pperi.

PAGE 50

50 Pmacro: Uptake by macrophytes Mortality of ma crophytes + Uptake by plant root Plant root decay. SRPf: Degradation of OPf + Desorption of IPf Adsorption into IPf Diffusion from floc layer into water column Diffusion from soil layer into floc layer. OPf: Sedimentation of POP Resuspensi on into POP Degradation into SRPf Sequestration into OPs. IPf: Adsorption of SRPf Desorption into SRPf + Sedimentation of PIP Resuspension into PIP Sequestration into IPs. SRPs: Degradation of OPs + Desorption of IPs Adsorption into IPs Uptake by plant root Diffusion from soil layer in to floc layer + Inflow of SRPs from deep soil layer Leaching of SRPs into deep soil layer. OPs: Plant root decay Degradation into SRPs + Sequestration from OPf Sequestration into deep soil layer. IPs: Adsorption from SRPs Desorption into SRPs + Sequestration from IPf Sequestration into deep soil layer. Phosphorus Transformation Processes As shown in Figure 3-1, major pathways in phosphorus cycle are: (1) atmospheric phosphorus deposition, (2) mineralization/degrad ation of organic phos phorus, (3) sorptiondesorption reaction of inorgani c and organic phosphorus, (4) sedimentation/resuspension of particulate phosphorus, (5) SRP uptake by biot a, (6) diffusion, and (7) soil phosphorus sequestration. Processes and constants used in the ECO Lab phosphorus dynamics model are summarized in Table 3-2 a nd Table 3-3, respectively. Some phosphorus transformation processes, such as atmospheric deposition, soil phosphorus sequestration, and sedimentation/re suspension, are expressed in terms of flux (g/m2/day). These are usually divided by unit water or soil depth to fit th e units of the equation (the change of concentration pe r unit time). In the model grid cells with extremely shallow water depths during dry period, model instability can occur because the very small water depth can abruptly generate a huge amount of flux. In this study, whenever fl ux terms are applied,

PAGE 51

51 minimum water depth ( min_wd ) is defined and MAX function, as shown below, is used to avoid the instability of phosphorus dynamics model. Atmospheric phosphorus deposition The importance of atmospheric deposition as a source of phosphorus varies from place to place. Atmospheric phosphorus deposition signif icantly influences water column phosphorus dynamics in pristine areas of the Everglades; on the other hand, the impact may be not dominant in nutrient-enriched areas close to agricultural runoff inputs, because the atmospheric portion is relatively small compared to surface water in flow containing highl y elevated phosphorus concentration (Raghunathan et al., 2001). Atmospheric phosphorus deposition is e xpressed by the depositional flux of each phosphorus species as follows: f_atm1 /MAX( wd min_wd ) (3-7) f_atm2 /MAX( wd min_wd ) (3-8) f_atm3 /MAX( wd min_wd ) (3-9) f_atm4 /MAX( wd min_wd ) (3-10) where f_atm1 f_atm2 f_atm3 and f_atm4 are atmospheric deposition flux of SRPw, DOP, PIP, and POP, respectively (Table 3-3). Table 3-4 shows atmospheric total phosphor us (TP) bulk depositi on rates in Florida reported in the previous studi es. The bulk deposition means wet plus dry deposition. Unlike nitrogen, phosphorus dry deposition is predominant (approximate statewide ratio; wet:dry = 1:5), and the deposition rate varies acr oss the state from 17 to 111 mg/m2/year (Brezonik et al., 1983). It is generally higher in areas adjacent to agricu ltural land use than pristine and coastal areas. Rural areas high phosphorus deposition rates (e. g. Belle Glade) and higher portions of dry deposition (wet:dry = 1:7) are attributed to sm oke and ash generated from the burning of sugar

PAGE 52

52 cane foliage prior to harvesting (Hendry et al., 1981), and may also be caused by tilling, which exposes large areas of dry, unvegetated peat to wind erosion (Grimshaw and Dolske, 2002). Considering the geographic location of th e current operating STAs adjacent to the Everglade Agricultural Area (EAA), atmospheric phosphorus deposition in this modeling study is estimated based on measurement data on SRP a nd TP wet deposition (Table 3-5). With the 1:5 ratio of wet vs. dry deposition (Brezonik et al., 1983), 85.5 mg/m2/year of bulk deposition rate is estimated. With the simple assumption that ha lf of wet deposition is dissolved phosphorus, DOP wet deposition rate is estimated as 2.33 mg/m2/year. The ratio of PIP vs. POP is again assumed to be 1:1, so PIP and POP wet depos ition rates are estimated as 3.56 mg/m2/year, respectively. Likewise, PIP and POP dry depositio n rates are estimated as 35.63 mg/m2/year, respectively. Mineralization Mineralization/degradation of organic phosphor us is a biological decomposition process mediated by extracellular enzymatic hydrolysis. In general, the mineralizati on rate of DOP is one of relatively fastest processes in phosphorus cy cle, which is accelerated by increased pH and high temperature. Mineralization rate constant is important in phosphorus model because it regulates phosphorus transformation from DOP to SRPw in water column, from OPf to SRPf in floc layer, and from OPs to SRPs in soil layer, respectively. Dissolved organic phosphorus forms in the floc and soil layers, which are intermediate species in the degradat ion processes of various organic matters, were not considered. The mi neralization/degradation processes are usually modeled by a first order kine tic equation as follows: k_min DOP (3-11) k_deg_f OPf bd_f / poro_f f_excop_f (g P/m3/day) or (3-12) k_deg_f OPf f_excop_f (mg P/kg/day) k_deg_s OPs bd_s / poro_s f_excop_s (g P/m3/day) or (3-13)

PAGE 53

53 k_deg_s OPs f_excop_s (mg P/kg/day) where k_min k_deg_f and k_deg_s are mineralization or degradation rate constant (unit: day-1) in the water column (DOP to SRPw), floc (OPf to SRPf), and soil layer (OPs to SRPs), respectively. Typical values of k_min vary from 0.1 to 1 day-1 (Chen, 1994), and it is a function of pH and temperature. Because the total amount of floc/soil organic phosphorus is not involved in degradation processes, biodegrad able phosphorus fractions of floc ( f_excop_f ) and soil organic phosphorus ( f_excop_s ) were considered in this study. Some literature values for rate constants are presented in Table 3-6. Due to more reduced redox condition in soil layer, degradation rate constants in floc and soil layers are likely much smaller than mineralization constants in water column. Formation/decay (sorption/desorption) Most previous water quality models used an equilibrium model for sorption/desorption reactions; however, a kinetic a pproach is required in cases of dynamics model using short simulation time step because it is not enough to reach equilibrium between the dissolved and particulate forms of phosphorus species (Chen, 1994) A first-order kinetic equation is frequently used as a kinetic model for sorption-desorption reactions (Appan and Wang, 2000). Reddy et al. (1998a) reported that phosphorus sorption charac teristics were generally well described by a linear model if phosphorus concentrations added fo r isotherm experiments are less than 10 mg/L. In this study, considering gene rally low water column phosphorus concentrations found in Florida constructed wetland ecosystems, linear and Langmuir model were used to represent adsorption/desorption process in water column a nd floc/soil layer, respectively. Each sorption and desorption process was described as a form of kinetic formulation defined in the derivation procedures of linear and Langmuir isotherm model (Chapra, 1997).

PAGE 54

54 Various decay processes define d by first order kinetics in this study are expressed as follows: k_decay1 POP (3-14) k_decay2 PIP (3-15) k_decay3 Pperi (3-16) k_decay4 Pmacro(1 frac_root ) (3-17) k_decay5 bd_f / poro_f IPf f_excip_f (g P/m3/day) or (3-18) k_decay5 IPf f_excip_f (mg P/kg/day) k_decay6 Pmacro frac_root (g P/m3/day) or (3-19) k_decay6 / bd_s Pmacro frac_root (mg P/kg/day) k_decay7 bd_s / poro_s IPs f_excip_s (g P/m3/day) or (3-20) k_decay7 IPs f_excip_s (mg P/kg/day) where k_decay1 k_decay2 k_decay3 k_decay4 k_decay5 k_decay6 and k_decay7 are desorption/biodegradation of POP into DOP, desorption of PIP into SRPw, sloughing of Pperi into POP, mortality of macrophytes, desorption of IPf into SRPf, root decay of macrophytes into OPs, and desorption of IPs into SRPs, respectively (Table 3-3). The unit of all these decay constants is day-1. Like the degradation processes of OPf and OPs described in the previous section, exchangeable phosphorus fractions of floc ( f_excip_f ) and soil inorganic phosphorus ( f_excip_s ) were considered for the desorption processes. These decay reactions are assumed to include dissolution, biodegradation, and desorption proce sses. Some literature values for the rate constants are summarized in Table 3-7. Various formation processes defined in this study are expressed as follows: k_form1 DOP (3-21)

PAGE 55

55 k_form2 SRPw (3-22) k_form3 POP (3-23) k_form4 SRPf( Smax_f IPf f_excip_f ) bd_f / poro_f (g P/m3/day) or (3-24) k_form4 SRPf( Smax_f IPf f_excip_f ) (mg P/kg/day) k_form5 SRPs( Smax_s IPs f_excip_s ) bd_s / poro_s (g P/m3/day) or (3-25) k_form5 SRPs( Smax_s IPs f_excip_s ) (mg P/kg/day) where k_form1 k_form2 k_form3 k_form4 and k_form5 are adsorption of DOP into POP, adsorption/precipitation of SRPw into PIP, cohesion of POP into Pperi, adsorption/precipitation of SRPf into IPf, and adsorption/prec ipitation of SRPs into IPs, respectively (Table 3-3). The unit of k_form4 and k_form5 is L/mg/day and the others are day-1. Smax_f and Smax_s denote maximum adsorption capacity (mg/kg) in floc and soil la yer, respectively. These formation reactions are assumed to include precipitation and complexi ng reactions as well as sorption processes including adsorption and absorption. Sedimentation/resuspension Sedimentation is one of the main removal processes of phosphorus in water column, in particular, particulate phosphorus. Two par ticulate phosphorus species, PIP and POP, are influenced vertically by sedimentation as we ll as horizontally by fluid motion in the water column. In this study, sedimentation does not indicate only traditional settling process of suspended materials, but also all mass transfer from water column to floc layer such as living or dead macrophytes and periphyton falling. In contrast, resuspension is one of the main release processes of phosphorus into water column. As phos phorus concentrations in floc or soil layer are usually several orders of ma gnitude higher than those in th e water column, resuspension of bottom sediments and sequential decay/desorption processes of the resuspended materials can significantly increase water column phosphorous level.

PAGE 56

56 To simulate sedimentation and resuspension processes in this phosphorus modeling study, the concept of critical flow velo city is applied. It is defined as flow velo city at the specific condition that phosphorus resuspension rate is equal to sedimentation rate in water column. It is also similar to the concept of critical orbital ve locity required for resuspension used by Blom and Toet (1993) and Blom and Aald erink (1998). Hence, if flow velocity calculated by the HD module at a grid cell is below th e critical velocity, only sediment ation occurs. On the contrary, once flow velocity exceeds the cr itical velocity, sedimentation is intermitted and mechanical resuspension occurs. In this study, non-mech anical (mainly biological) resuspension is considered regardless of flow velocity regime to represent the vari ous biotic and abiotic resuspension processes described in Chapter 2. In this study, sedimentation of PIP and POP and resuspension of IPf and OPf are expressed as follows: v_dep1 /MAX( wd min_wd )PIP (only if U < v_crit ) (g P/m3/day) or (3-26) v_dep1 /MAX( wd min_wd )/ bd_f PIP (only if U < v_crit ) (mg P/kg/day) v_dep2 /MAX( wd min_wd )POP (only if U < v_crit ) (g P/m3/day) or (3-27) v_dep2 /MAX( wd min_wd )/ bd_f POP (only if U < v_crit ) (mg P/kg/day) (if U > v_crit ) ( v_resus1+v_resus2) /MAX( wd min_wd )IPf bd_f (if U < v_crit ) v_resus2 /MAX( wd min_wd )IPf bd_f (g P/ m3/day) or (3-28) (if U > v_crit ) ( v_resus1+v_resus2) /MAX( wd min_wd )IPf (if U < v_crit ) v_resus2 /MAX( wd min_wd )IPf (mg P/kg/day) (if U > v_crit ) ( v_resus3+v_resus4) /MAX( wd min_wd )OPf bd_f (if U < v_crit ) v_resus4 /MAX( wd min_wd )OPf bd_f (g P/ m3/day) or (3-29) (if U > v_crit ) ( v_resus3+v_resus4) /MAX( wd min_wd )OPf

PAGE 57

57 (if U < v_crit ) v_resus4 /MAX( wd min_wd )OPf (mg P/kg/day) where v_dep1 v_dep2 v_resus1 v_resus2 v_resus3 and v_resus4 represent PIP depositional rate, POP depositional rate, mechanical resuspension rate of IPf, non-mechanical resuspension rate of IPf, mechanical resuspension rate of OPf, and non-mechanical resuspension rate of OPf, respectively (Table 3-3). Some literature values for critical velocity, phosphorus sedimentation, and resuspension rate are summarized in Table 3-8. Uptake by biota When water flows through a constructed wetland, plants absorb nutrients from the water column and soil layer, using them in metabolism or storing them in their tissues. Certain types of vegetation planted in a certain sequence in constr ucted wetlands are very efficient to remove nutrients. Of phosphorus species, only SRP, a me asure of dissolved bioavailable phosphorus, is taken up by biota as a nutrient. In a phosphorus limiting ecosystem like the Everglades, biota growth is dependent on the amount of SRP in water column and soil layer. Although the type of wetland biot a varies, it is classified in to three groups: (1) EAV, (2) SAV, and (3) periphyton in this study, considering the lack of fi eld data and the complexity of model. EAV species primarily obtain SRP from soil layer via root uptake, while SAV species directly uptake the considerable amount of SRP from water column. Hence, phosphorus uptake by root and by shoot/foliage is separate and three periphyton types, epiphyton, metaphyton, and epipelon, are aggregated to one peri phyton species in this model. The uptake rate of SRP by biota is proportional to the growth rate of biota. In general, biota growth in phosphorus limiting ecosystems is expressed by Michaelis-Menten kinetics as follows: kg_macro [SRPw/(SRPw+ ks_macro )]Pmacro(1 frac_root ) (3-30) kg_root [SRPs/(SRPs+ ks_root )]Pmacro frac_root (3-31)

PAGE 58

58 kg_peri [SRPw/(SRPw+ ks_peri )]Pperi (3-32) where kg_macro kg_root and kg_peri are macrophyte, macrophyte root, and periphyton maximum growth rate constants (1/day) and ks_macro ks_root and ks_peri are macrophyte, macrophyte root, and periphyton uptake half satu ration constants (mg/L), respectively. In general, biotic growth rate constants are a func tion of temperature, light intensity, and nutrient concentration. In this study, it is assumed that macrophyte an d periphyton growth are only a function of SRP concentration in water column/soil layer and temperature; that is, the aquatic system is phosphorus-limiting. Hence, other limiting fa ctors such as nitrogen and light intensity are not considered. In these kine tics, the half saturation constant s, defined to parameterize the dependence of biota growth on phosphorus, are the SRP concentration at which the growth rate is one half of the maximum. Table 3-9 presents some literature values of maximum growth rate and half saturation constant fo r phosphorus uptake by EAV (mainly Typha spp. ), SAV, and periphyton. Diffusion Diffusion of SRP between water column and floc layer or between floc and soil layer is expressed by Ficks law as follows: poro_f D (SRPfSRPw)/ d_diff_f /MAX( wd min_wd ) (3-33) poro_sD( SRPsSRPf)/d_diff_s/d_ave_f (3-34) where poro_f / poro_s D d_diff_f / d_diff_s and d_ave_f are floc/soil layer porosity, effective diffusion coefficient (m2/day), depth of diffusive exchange in floc/soil layer, and average depth of floc layer, respectively (Table 3-3). Sequestration Phosphorus sequestration was defined by Turner et al. (2006) as th at phosphorus removed from the water column through various retention pr ocesses and retained in stable forms in soil.

PAGE 59

59 The accretion of inorganic and organic phosphorus in to soil layer via floc layer is the dominant process. In this study, floc and soil phosphorus se questrations are described as follows: f_seq1 / d_ave_f / bd_f (3-35) f_seq2 / d_ave_f / bd_f (3-36) f_seq3 / d_ave_s / bd_s (3-37) f_seq4 / d_ave_s / bd_s (3-38) where f_seq1 f_seq2 f_seq3 and f_seq4 are sequestration flux of IPf into IPs, OPf into OPs, IPs into deep soil layer, and OPs into deep soil layer, respectiv ely (Table 3-3). Each flux (unit: g/m2/day) is divided by average depth of floc ( d_ave_f ) and soil layer ( d_ave_s ). Table 3-10 shows some literature values for phosphorus sequestration rate in the Everglades. Temperature effect Most phosphorus transformation processes illustrated in Fi gure 3-1 are directly or indirectly affected by temperatur e. Biologically mediated reactions are directly affected by the increased biological activity at higher temperature. Chemical reac tions are also accelerated with increased temperature partially due to higher mol ecular activity. As a general rule of thumb, the rates of most reactions in natural waters will approximately increase twice for a temperature rise of 10C (Chapra, 1997). Temperature dependence of biologically mediat ed reactions is desc ribed by Arrhenius equation as: k ( T ) = k (20) T (3-39) where the temperature ( T ) is expressed in C, k (20) is a rate cons tant at 20C, and is a temperature coefficient genera lly ranging from 1.0 to 1.1.

PAGE 60

60 Parameters Phosphorus transformation processes described in the previous secti ons include a lot of model parameters that should be determined be fore running the models. Determining the model parameters is generally difficult because they ar e temporally and spatially heterogeneous within a constructed wetland system, depending on num erous physiochemical factors such as temperature, pH, redox, and DO. Some coefficien ts and rate constants can be found in the literature, but many have not been documented. Table 3-3 shows the lis t of 56 constants used in the ECO Lab phosphorus dynamics model. As described in the previous sections, the range of literature values of these constants is very wide. The specification of constants for the model application (Chapter 6) is determined based on typical literature values (or from the range), and some parameters are adjusted in order to obtain the best model fit. In a few cases which appropriate literature values are not available, values are initially guessed and fi nally estimated by model calibration.

PAGE 61

61 Table 3-1. State variables used in th e ECO Lab phosphorus dynamics model. Symbol Description Units Horizontal transport SRPw DOP PIP POP Pperi Pmacro Water column Soluble reactive phosphorus Dissolved organic phosphorus Particulate inorganic phosphorus Particulate organic phosphorus Phosphorus in periphyton Phosphorus in macrophytes mg/L mg/L mg/L mg/L mg/L mg/L Mobile Mobile Mobile Mobile Stationary Stationary SRPf OPf IPf Floc layer Soluble reactive phosphorus in floc layer porewater Soil organic phosphorus in floc layer Soil inorganic phosphorus in floc layer mg/L mg/kg mg/kg Stationary Stationary Stationary SRPs OPs IPs Upper soil layer (0-10 cm) Soluble reactive phosphorus in soil layer porewater Soil organic phosphorus in soil layer Soil inorganic phosphorus in soil layer mg/L mg/kg mg/kg Stationary Stationary Stationary

PAGE 62

62 Table 3-2. Processes used in the ECO Lab phosphorus dynamics model. Symbol Description (Unit: g P/m3/day or mg P/kg/day) Formulation 1. SRPw_atmdep 2. DOP_atmdep 3. PIP_atmdep 4. POP_atmdep 5. POP_decay* 6. POP_form* 7. DOP_mineral* 8. PIP_decay 9. PIP_form 10. PIP_sed 11. PIP_resus* 12. POP_sed 13. POP_resus* 14. SRPw_periup* 15. SRPw_macup* 16. SRPw_diff 17. Pperi_decay* 18. Pperi_form* 19. Pmacro_decay* Water column Atmospheric deposition of SRPw Atmospheric deposition of DOP Atmospheric deposition of PIP Atmospheric deposition of POP POP decay into DOP (desorption/biodegradation) POP formation from DOP (adsorption) Mineralization of DOP into SRPw PIP decay into SRPw (desorption) PIP formation from SRPw (adsorption/precipitation) Sedimentation of PIP into floc layer Resuspension of PIP from floc layer Sedimentation of POP into floc layer Resuspension of POP from floc layer Uptake of SRPw by periphyton Uptake of SRPw by macrophytes SRPw diffusion between floc layer and water column Pperi decay into POP (sloughing) Pperi formation from POP (cohesion) Pmacro decay into POP (mortality of macrophytes) Flux Flux Flux Flux 1st order 1st order 1st order 1st order 1st order Critical veloc. Cri.+Flux Critical veloc. Cri.+Flux M-M M-M Diffusion 1st order 1st order 1st order 20. OPf_degrad* 21. IPf_decay 22. IPf_form 23. SRPf_diff 24. IPf_seq 25. OPf_seq Floc layer Degradation of OPf into SRPf IPf decay into SRPf (desorption) IPf formation from SRPf (adsorption/precipitation) SRPf diffusion between upper soil and floc layer Sequestration of IPf into upper soil layer Sequestration of OPf into upper soil layer 1st order Langmuir iso. Langmuir iso. Diffusion Flux Flux 26. OPs_degrad* 27. IPs_decay 28. IPs_form 29. SRPs_macup* 30. Pmacro_rdecay* 31. SRPs_leakdn 32. SRPs_leakup 33. IPs_seq 34. OPs_seq Upper soil layer (0-10 cm) Degradation of OPs into SRPs IPs decay into SRPs (desorption) IPs formation from SRPs (adsorption/precipitation) Root uptake of SRPs by macrophytes Root decay of macrophytes into OPs Leaching of SRPs into deep soil layer Inflow of SRPs from deep soil layer Sequestration of IPs into deep soil layer Sequestration of OPs into deep soil layer 1st order Langmuir iso. Langmuir iso. M-M 1st order Flux Flux Flux Flux The temperature dependence of biologically medi ated reactions is cons idered as a form of Arrhenius equation as follow: k ( T ) = k (20) T where the temperature is expressed in C and is temperature coefficient of Arrhenius equation.

PAGE 63

63 Table 3-3. Constants used in the ECO Lab phosphorus dynamics model. Symbol Description Unit 1. f_atm1 2. f_atm2 3. f_atm3 4. f_atm4 5. k_decay1 6. k_decay2 7. k_decay3 8. k_decay4 9. k_decay5 10. k_decay6 11. k_decay7 12. k_form1 13. k_form2 14. k_form3 15. k_form4 16. k_form5 17. k_min 18. k_deg_f 19. k_deg_s 20. v_crit 21. v_dep1 22. v_dep2 23. v_resus1 24. v_resus2 25. v_resus3 26. v_resus4 27. kg_macro 28. kg_root 29. kg_peri 30. ks_macro 31. ks_root 32. ks_peri 33. D 34. d_diff_f 35. d_diff_s 36. d_ave_f 37. d_ave_s 38. frac_root 39. bd_f 40. bd_s 41. poro_f 42. poro_s Atmospheric deposition flux of SRPw Atmospheric deposition flux of DOP Atmospheric deposition flux of PIP Atmospheric deposition flux of POP Decay rate constant (POP DOP) Decay rate constant (PIP SRPw) Decay rate constant (Pperi POP) Decay rate constant (Pmacro POP) Decay rate constant (IPf SRPf) Root decay rate constant (Pmacro OPs) Decay rate constant (IPs SRPs) Formation rate constant (DOP POP) Formation rate constant (SRPw PIP) Formation rate constant (POP Pperi) Formation rate constant (SRPf IPf) Formation rate constant (SRPs IPs) Mineralization rate constant (DOP SRPw) Degradation rate constant (OPf SRPf) Degradation rate constant (OPs SRPs) Critical velocity of flow Deposition of PIP Deposition of POP Mechanical resuspension of IPf Non-mechanical (biologi cal) resuspension of IPf Mechanical resuspension of OPf Non-mechanical (biological) resuspension of OPf Macrophyte max. growth rate constant Macrophyte root max. growth rate constant Periphyton max. growth rate constant Macrophyte uptake half saturation constant Macrophyte root uptake half saturation constant Periphyton uptake half saturation constant Effective diffusion coefficient Depth of diffusive exchange in floc layer Depth of diffusive exchange in upper soil layer Average depth of floc layer Average depth of upper soil layer Root fraction of macrophyte Bulk density of floc layer Bulk density of upper soil layer Floc layer porosity Upper soil layer porosity g/m2/day g/m2/day g/m2/day g/m2/day 1/day 1/day 1/day 1/day 1/day 1/day 1/day 1/day 1/day 1/day L/mg/day L/mg/day 1/day 1/day 1/day m/s m/day m/day m/day m/day m/day m/day 1/day 1/day 1/day mg/L mg/L mg/L m2/day m m m m kg/L kg/L

PAGE 64

64 Table 3-3. Continued. Symbol Description Unit 43. Smax_f 44. Smax_s 45. f_excop_f 46. f_excop_s 47. f_excip_f 48. f_excip_s 49. f_seq1 50. f_seq2 51. f_seq3 52. f_seq4 53. f_leak1 54. f_leak2 55. 56. min_wd Max. adsorption capacity in floc layer Max. adsorption capacity in upper soil layer Biodegradable phosphorus fraction of OPf Biodegradable phosphorus fraction of OPs Exchangeable phosphorus fraction of IPf Exchangeable phosphorus fraction of IPs Sequestration flux of IPf into IPs Sequestration flux of OPf into OPs Sequestration flux of IPs into deep soil layer Sequestration flux of OPs into deep soil layer Leaching flux of SRPs into deep soil layer Leaching flux into SRPs from deep soil layer Temperature coefficient of Arrhenius equation Minimum water depth mg/kg mg/kg g/m2/day g/m2/day g/m2/day g/m2/day g/m2/day g/m2/day m

PAGE 65

65 Table 3-4. Atmospheric total phosphorus (TP) bu lk deposition rates in Florida reported in previous studies. Source Value (ave.) Unit Description Hendry et al., 1981 Brezonik et al., 1983 Walker, 1993 Walker, 1995 Redfield, 1998 Raghunathan et al., 2001 Grimshaw and Dolske, 2002 17-96 17-111 (51) (66) 96 0-116 43 20-80 >50 47 1.3 0.3* mg/m2/yr Seven South Florida locations Statewide (Wet : Dry = 1:5) Rural (agricultural) areas Belle Glade (Wet : Dry = 1:7) Assumed range in South FL WCA-2A Statewide Urban and agricultural areas Everglades Statewide (Sep. 92-Oct. 93) The mean rate of wet atmospheric phosphorus deposition across the state. Table 3-5. Estimation of atmospheric phosphorus deposition used in STA 5 northern flow-way phosphorus models. Wet : Dry = 1 : 5 Wet deposition (mg/m2/year) Dry deposition (mg/m2/year) Bulk deposition (mg/m2/year) Bulk deposition (ton/year) SRPw 4.79*4.79 0.039 DOP 2.332.33 0.019 PIP 3.5635.6339.19 0.319 PP POP 3.5635.6339.19 0.319 TP 14.25**71.2585.50 0.696 Measured data provided by Chimney (2007). ** Measured data provided by Pietro et al. (2006).

PAGE 66

66Table 3-6. Literature values for minerali zation/degradation rates of organic matter in water column, floc, and soil layer. Source Value Unit Comment Water column Bowie et al. (1985) Chen (1994) Chao et al. (2006) 0.2-0.22 0.5 (0.1-1.0) 0.15 1/day Lake Okeechobee, FL Shallow oxbow lake phosphorus model Floc layer Findlay et al. (1990) Mitsch and Gosselink (1993) 0.00086 0.002-0.007 1/day Bottom detritus decay rate Litter decomposition rates of Typha spp. in freshwater marshes Upper soil layer Bowie et al. (1985) Kadlec and Hammer (1988) Jrgensen and Bendoricchio (2001) Wang et al. (2003b) 0.001 (0.0004-0.01) 7.86-6 0.0004-0.001 0.002-0.046 1/day Decay rate in active sediment layer used in modeling study of Wang and Mitsch (2000) Organic sediment PO4 3Organic decay coefficients used in modeling

PAGE 67

67Table 3-7. Literature values for various phosphorus decay/desorption rates. Source Value Unit Comment Water column k_decay1 (POP DOP) Bowie et al. (1985) Jrgensen and Bendoricchio (2001) Lung (2001) Chen and Sheng (2005) k_decay2 (PIP SRPw) Chen and Sheng (2005) k_decay3 (Pperi POP) Bowie et al. (1985) Asaeda and Van Bon (1997) Wang and Mitsch (2000) k_deacy4 (Pmacro POP) Davis and van der Valk (1978) Davis (1984) Kadlec and Knight (1996) van der Peijl and Verhoeven (1999) Grace (2003) Wang and Mitsch (2000) DeBusk and Reddy (2005) Alvarez and Becares (2006) Chimney and Pietro (2006) 0.2-0.22 0.22 0.03 6 8 0-0.8 0.003-0.17 0.1 0.1143 0.0063 0.003-0.0075 0.0013-0.0031 0.0057 0.011 f(time) 0.0028 0.0043-0.0052 0.0014-0.0026 0.0059 0.0568 0.0255 1/day 1/day 1/day 1/day g/m2/day 1/day POP SRPw: 0.02-0.1 day-1 Suggested as a first approximation value Muddy zone in Lake Okeechobee Muddy zone in Lake Okeechobee Benthic algae non-predatory mortality rate Total phytoplankton non-pred atory mortality rate Periphyton loss rate determined by calibration Falling rate of standing dead Annual release rate of P by cattail leaves at the enriched zone of WCA-2A Decomposition rates of Typha spp. in wetlands Max. relative death rate of above-ground plant biomass 4 year-1 Macrophyte death rate determined by calibration Mean decomposition rate fo r cattail litter at the P-enriched area of WCA-2A (Range: 0.001-0.0092) Typha latifolia in summer (20C) Typha latifolia in winter (5C) Mean decomposition rate of submerged Typha Mean decomposition rate of SAV Literature median of decomposition rates for SAV Upper soil layer k_decay6 (Pmacro OPs) van der Peijl and Verhoeven (1999) 0.00143 1/day Max. relative death rate of below-ground plant biomass

PAGE 68

68Table 3-8. Literature values for cri tical velocity and phosphorus sediment ation/resuspension rate in wetlands. Source Value Unit Comment Critical velocity Blom and Toet (1993) Blom and Aalderink (1998) 0-0.04 0.00032 0.305 m/s Critical orbital velocity requi red for resuspension (field) Critical orbital velocity requi red for resuspension (field) Critical orbital velocity required for resuspension based on flume experiment (Range: 0.290-0.320) Sedimentation rate of PIP Blom and Toet (1993) Chen (1994) Blom and Aalderink (1998) Wang and Mitsch (2000) Chao et al. (2006) 0.864-36.288 1.296 0.36 6.84 0.43 0.003-0.007 0.01 m/day g/m2/day m/day Apparent settling velocity based on field data Settling velocity of sediment at Lake Okeechobee (Range: 0.0894-2.16) Apparent settling velocity based on flume experiment (Range: 0.0576-0.792) Apparent settling velocity based on field data (Range: 5.064-8.64) Sedimentation rate of TP determined by model calibration Simulated TP sedimentation flux Shallow oxbow lake phosphorus model Sedimentation rate of POP Larsen et al. (1974) Bowie et al. (1985) Wang and Mitsch (2000) DHI (2004c) 0.02 0.04 0.01-4.0 0.5714 0.2 (0.07-0.7) m/day Algal phosphorus settling velocity Non-algal particulate phosphorus settling velocity Total phytoplankton settling velocity Plankton settling rate velocity determined by calibration Depositional rates for particulate organic matter (1-10 m) A typical value of 0.2 m/day corresponding to a particle Size of 5 m is suggested. Resuspension rate Mechanical Wang and Mitsch (2000) Chen and Sheng (2005) Non-mechanical Fisher and Reddy (2001) 0.00143 0.067 0.0065 g/m2/day Determined by model calib ration (Range: 0.00024-0.0126) Measured at the enriched zone in WCA-2A using intact soil core/benthic chamber

PAGE 69

69Table 3-9. Literature values (maximum growth rate and half saturation cons tant) for phosphorus uptake by emergent ( Typha spp. ), submerged aquatic vegetation, and periphyton. Source Value Unit Comment Max. growth rate of macrophyte Emergent aquatic vegetation Davis (1984) Davis (1991) Wang and Mitsch (2000) Noe and Childers (2007) Submerged aquatic vegetation Rodgers et al. (1983) Janse (1998) 0.0041-0.01 0.0018-0.0115 0.00085-0.0045 0.00008 0.00055 0.003 0.0044 0.004-0.17 0.2-0.5 0.2-0.4 g/m2/day 1/day Annual uptake rate of P by cattail leaves at the enriched zone of WCA-2A Annual average P assimilation rate of Typha domingensis leaves in the Everglades P pumped out of deep sediment by macrophytes Oligotrophic Slough in Everglades Cladium marsh in Everglades Partially enriched Cladium/Typha marsh in Everglades Enriched Typha marsh in Everglades Foliage growth rate Submerged rooted plants Submerged non-rooted plants Max. growth rate of periphyton Bowie et al. (1985) Sand-Jensen and Borum (1991) Scinto and Reddy (2003) 0.042-0.5 0.02-2.0 1376 930 446 1/day 1/day g P/g dw/h g P/g dw/h g P/g dw/h Max. P uptake rates by green algae, blue-green algae, flagellates, and benthic algae Phytoplankton Epiphyton at the interior site of WCA-2A Metaphyton at the interi or site of WCA-2A Epipelon at the interior site of WCA-2A Half satur. const. of periphyton uptake Bowie et al. (1985) Chen (1994) Scinto and Reddy (2003) Chao et al. (2006) 0.004-0.08 0.01 0.264 0.499 0.508 1.0 mg/L Total phytoplankton Calibrated value on phytoplankton uptake for Lake Okeechobee P dynamics model (Range: 0.005-0.03) Epiphyton at the interior site of WCA-2A Metaphyton at the interi or site of WCA-2A Epipelon at the interior site of WCA-2A Shallow oxbow lake phosphorus model

PAGE 70

70Table 3-10. Literature values for phosphorus sequestration rate in the Everglades. Source Value Unit Comment Craft and Richardson (1993) Reddy et al. (1993), Richardson and Craft (1993) Turner et al. (2006) Inorganic phosphorus EAV SAV Organic phosphorus EAV SAV 0.0013 0.0004 0.0011-0.0033 0.0044 0.0025 0.0015 0.0003 g/m2/day P accumulation rate at the enriched area in WCA-2A P accumulation rate at the une nriched area in WCA-2A Average accumulation rates in the Everglades peats STA-1W Cell 1 STA-1W Cell 4 STA-1W Cell 1 STA-1W Cell 4

PAGE 71

71 Figure 3-1. A conceptual diagram of constructed wetland phosphorus dynam ics model applied in this study.

PAGE 72

72 CHAPTER 4 MODEL APPLICATION TO THE ORLAN DO EASTERLY WETLAND (OEW) CELL 7 Introduction Solutes, including tracers, nutrients, and polluta nts, are transported through primary flow paths in the main channel of a wetland and have limited contact with stagnant water areas along the banks, in dense emergent vegetations, and in the subsurface hyporheic zone (Keefe et al., 2004; Harvey et al., 2005). To enhance efficien cy of a treatment wetland, a systematic understanding of various factors a ffecting the solute transport or retention is required. Wetland hydraulics is one of the most fundamental and significant factors, because the hydraulic efficiency is directly related to the treatment efficiency of constructed wetlands (Martinez and Wise, 2003b). This chapter is focused on the impact of topographic and vegetative heterogeneity on the formation of short-circuiting flow zone through modification of those factors. One reason these two factors were chosen among many ones reviewed in Chapter 2 is that it is usually difficult to obtain these data in the field dir ectly. As a result, few studies on these, compared to the other factors, have been reported ba sed on actual wetland systems, rath er than hypothetical ones. In addition, they are the most critical component s for application of mu lti-dimensional, fully distributed, dynamic models in a constructe d wetland with a given shape and hydraulic structures. As a result, several implications draw n in this chapter provid e a useful starting point for the modeling efforts in Chapter 5 and 6. Study Area The study area, Orlando Easterly Wetland (OEW), located east of the City of Orlando and about three kilometers west of th e St. Johns River in central Flor ida (Figure 4-1), is one of the oldest and largest constructed tr eatment wetlands in the United States (Black and Wise, 2003;

PAGE 73

73 Martinez and Wise, 2003b). Since 1 987, it has been used to reduce nutrient loading (nitrogen and phosphorus) of the treated wastewater discharged to the St. Johns River (Kadlec and Newman, 1992). Figure 4-1 shows the plan view of northern part of the entire wetland system including 17 compartmentalized cells (and a downstream lake ) divided by earthen berms. The inflow is typically distributed into three fl ow paths of sequential cells at in fluent structure (closed circles in Figure 4-1), and flow control in each cell is achieved through adjustable, sharp-crested rectangular weirs (clo sed rectangles). For a detailed 2-D, hydrodynamic and solute tr ansport modeling, Cell 7, located in the middle of the northern flow train, was sele cted. The cell has a topographic slope of approximately 0.2% along the main flow direction fr om influent weir (3-X) to the effluent weir (7-X), and the average drop in bed elevation acr oss the cell is about 0.6 m. Martinez and Wise (2003b) reported that site gradi ng was not conducted in the inte rior of the wetland cells during conversion of the site from pasture land for cat tle operation to treatment wetland. Artificial landforms, such as ditches and roads, still exist within the cells. Cell 7 is a wet prairie with dominant vegetation of cattail ( Typha spp. ) and bulrush ( Scirpus spp. ) (Martinez and Wise, 2003b). Previous Studies Bromide tracer test and hydraulic analysis Martinez and Wise (2003b) conducted bromide tracer experiments for mostly the OEW treatment cells. From RTD-moment analyses of th e treatment cells, the hydraulic efficiencies of the individual treatment cells were estimated; th e entire wetland system was operating at near 50% efficiency during the period of test. The existence of pref erential flow paths resulting in short-circuiting flow with the associated forma tion of dead zones was proposed as the main factor causing these low hydraulic efficiencies wi thin the OEW treatment cells. For a detailed

PAGE 74

74 description of the bromide tracer tests and the results, see White et al. (2002) and Martinez and Wise (2003b). One-dimensional (1-D) transient storage model In Martinez and Wise (2003a), one-dimensiona l transport model with inflow and storage (OTIS) was applied to simulate the OEW bromide tracer experiment results and quantit atively to estimate short-circuiting and transient storage within the constructed wetland. This transient storage model consists of a main channel de scribed by advection-di spersion equation with completely mixed storage zones and has been main ly applied to river or stream systems where a 1-D approach is reasonable (Bencala and Walters 1983; Harvey et al., 1996). Despite usefulness and conceptual simplicity, the model has several limitations. As the authors discussed, it cannot account for the spatial distribution of multiple pa rallel flow paths and can only provide lumped parameters, such as main channel and storage zone cross-sectional ar eas and an exchange coefficient between main channel and storage zone of which the physical meaning is not clear. This shows that a 2-D, physically based, distri buted modeling approach is desirable to overcome the limitations of 1-D, wetland flow models. Model Scope Two-dimensional, physically based, distributed hydraulic and solute transport models are often used to obtain valuable insights on the hydrodynamic behaviors of water systems (Persson et al., 1999). This is due to the fact that they allow researchers to estimate the effect of each factor separately, based on realis tic flow conditions. In this chapter, in order to examine quantitatively the impact of bathymetry and vegetation density on the formation of shortcircuiting flow, MIKE 21 HD and AD modules were used for a cons ervative transport simulation on bromide tracer experiment carried out at the OE W Cell 7. Due to lack of field data on recent bathymetry and vegetation distribution, a full app lication of the distribu ted modeling feature was

PAGE 75

75 limited. As a result, the assignment of bottom elevation and Mannings roughness coefficient, which represents bathymetry and hydraulic resistance, respectively, on each grid was conceptually conducted along the flow zonation delin eated by the 2-D, transient distribution of flow velocity vectors during the calibration (sensitivity analysis) procedure. Each impact was estimated and compared in terms of hydraulic e fficiency, defined by Perss on et al. (1999), which is based on breakthrough curves (BTCs) generated by each simulation. Specific Aims Bathymetry and hydraulic roughness due to ve getation are the most critical model components for distributed hydrodynamic models in a constructed wetland with a given shape. Getting accurate bathymetry is im portant since it plays a key role in controlling water depth and flow velocity in a constructed wetland. In gene ral, uneven bathymetry causes the difference of water depth, resulting in the difference of fl ow velocity sometimes resulting in hydraulically dead zones. On the other hand, the impact on hydrau lic resistance of a variety of vegetations is one of most unique characteristic s of wetland flow system. In ge neral, hydraulic resistance, denoted by Mannings roughness coefficient ( n ), is a function of the sp atial distribution, density, and type of vegetation as well as water depth. Ho wever, it is usually very difficult directly to obtain these data in the field; as a result, few studies on these ha ve been reported based on actual wetland systems in spite of their significance. In particular, there are few available references on the selection of Mannings n values for various vegetated wetland areas in 2-D hydrodynamics models. The main purpose of a modeling work descri bed in this chapter is to explore the applicability of a modeling appro ach presented in Chapter 3 for the development and application of more advanced flow dynamics and solute tr ansport/retention models in constructed wetland systems in spite of the lack of direct field measurement. That is, the primary interest of this

PAGE 76

76 chapter does not exist on the development of a 2-D, physically based, fully distributed model, which is precisely calibrated with field measur ement, but exists on the derivation of useful implications obtained through model sensitivity anal ysis regarding the effects of bathymetry and vegetation density on short-circuiting flow in a constructed wetland system. Methods Moment Analysis and Hydraulic Performance Key parameters used to characterize hydrauli c performance of a we tland are derived from moment analysis of an RTD (Kadlec and Kn ight, 1996; Martinez and Wise, 2003b; Holland et al., 2004). The zeroth absolute moment ( Mo) of an RTD is equivalent to the total mass of tracer recovered, and the first normalized absolute moment ( 1) yields the centroid of the RTD, which is the mean residence time, t (days). The second normalized central moment ( 2) is the variance, 2 (days2), of the RTD, which accounts for the spreading. These moments can be used to describe the outlet response of a wetland system to a Dirac input function for a tracer (Sardin et al., 1991). However, if the input boundary conditi on is a continually ch anging time series, the moments can be formulated under the assumption that the system behaves linearly (Das and Kluitenberg, 1996). MRF = 0 inf inf 0) ( ) ( ) ( ) ( dt t C t Q dt t C t Qeff eff (4-1) inf n eff n n ( n = 1 and 2) (4-2) where Qeff(t) and Qinf(t) are the volumetric flow rates (m3/s) of effluent and influent, respectively; Ceff(t) and Cinf(t) the tracer concentration (mg/L) of ef fluent and influent, respectively; and eff n and inf n are the first (n=1) or second (n=2) norma lized moment of effluent and influent, respectively. Mass recovery fraction (MRF) is defi ned in Equation 4-1 as the ratio of tracer mass

PAGE 77

77 recovered at the outlet point to the total mass of tracer injected. Equation 42 serves as a general formula for calculating 1 and 2 for wetland systems with arbitrary input boundary conditions. In this case, the mean residence time is considered as a time interval between each centroid of inlet and outlet RTD curves. Equation 4-1 and 4-2 were used for the moment analysis in this study. In general, nominal residence time, (days), is defined as AVG AVGQ V (4-3) where VAVG is the volume of a wetland cell (m3) and QAVG is the average cell flow (m3/day). In Martinez and Wise (2003b), hydraulic efficiency was defined as a ratio of experimentally determined mean residence time ( t) to nominal residence time ( ), which was suggested as the effective volume ratio (e) by Thackston et al. (1987). t e (4-4) Persson et al. (1999) proposed an altern ative metric for hydraulic efficiency () that combines existing metrics of flow uniformity (1-1/N) and effective volume (e). p pt t t t t N e ) ( 1 1 1 (4-5) where N is the number of CSTRs in series requ ired to simulate flow conditions and tp is the peak time for a BTC. The number of CSTRs in series (N) is calculated using the difference in the time-to-peak concentration and the mean residence time as follows: N = t/( t tp) (Kadlec and Knight, 1996).

PAGE 78

78 For a detailed descriptio n of the RTD moment analysis or the parameters for measuring hydraulic performance, the reader is referred to Levenspiel (1972), Kadlec and Knight (1996), Werner and Kadlec (1996), and Persson et al. (1999). Hydrodynamic Model Setup The 2-D HD model was used to simulate th e dynamic flow conditions observed at OEW Cell 7. Of the basic parameters required for the m odel setup, the bathymetry of modeled area was formed on a 30-by-30 meter grid, initially base d on the constructed we tland blueprint for the OEW reclamation project (Post, Buckley, Schuh & Jernigan, Inc., 1985) and finally determined after calibration. The simulation period was set from November 17, 2000 2:00 PM to December 10, 2000 8:00 AM (22.75 days), which corresponds to the period in which flow data were measured and the samples for bromide concentrat ion analysis were collected. Each simulation time step for both flow and conservative tr ansport modeling was 5 seconds. The wetland boundary was surrounded by the cells where land values were assigned, resulting in no flow boundary conditions along the wetland boundary cells to reflect the surrounding berm. Inflow and outflow that occurred thr ough inlet and outlet weirs were si mulated using the source and sink options. Two cells corresponding to the location of each weir were selected, and time series inflow/outflow data calculated at each weir were assigned to th e cells. In MIKE 21, it is possible to simulate both flooding and dr ying condition of wetland with th e transient flow condition. In this study, if the water depth at a cell is less th an 0.02 meter, the cell was considered to become dry; however, once the depth is greater than 0.03 meter, the cell was regarded as a flooded condition subject to flow. The in itial surface elevation was set as a constant value of 6.85 m corresponding to the initial water level over top elevation of outlet weir, and for flow resistance, which is expressed as a form of Manning roughne ss coefficient, a constant value of n = 0.11 was initially set and later adjusted for model calibration. The time series results of HD simulation

PAGE 79

79 such as flow velocity and water depth in every gr id cell were used as input data for the bromide transport simulation in the AD module. Rainfall was minimal to non-existent during the period of field experiment (Martinez and Wise, 2003a). As precipitation and evaporation during the simulation period were negligible, compared to the flow through the Cell 7, they were not included in this simulation. Also, wind effects were not considered. Advection-Dispersion Model Setup The OEW Cell 7 bromide tracer experiment carried out on November-December 2000 was simulated using the conservative solute tran sport option of MIKE 21 AD module. For the AD simulation, the initial concentra tion of bromide was set to zero, and the time series bromide concentration profile measured at the inlet was assigned at the source cell. It is generally known that selecting 2-D dispersion coefficients, one of the most important parameters in advectiondispersion simulation, is very difficult (DHI, 2004b). Of severa l empirical estimates (DHI, 2004b), the following formulation was used to es timate the coefficients (they were finally determined through calibration process). Dx = a xvx / Dy = a yvy (4-6) where Dx and Dy are the x and y direction dispersion coefficient (m2/s), x = y is the constant grid spacing (30 m), and vx and vy are the local current velocity components (m/s). As the local current velocity is not constant with respect to time, water ve locity on a time step or average water velocity of each cell was required in both x and y directions to determine vx and vy. In this simulation, a time step when the peak concentrati on of bromide measured at the inlet weir was observed was selected. The calibrate d dispersion coefficients were specified as a 2-D coefficient map similar to those for the bat hymetry and hydraulic resistance.

PAGE 80

80 Results and Discussion Bromide Tracer Experiment Simulation Table 4-1 shows parameters of models used and the moment analysis results of the bromide BTC in previous and this study, respec tively. Discrepancies between the previous research and this study are larg ely attributed to the differen ce of simulation period and flow conditions simulated. Although previous research used a slightly longer period for flow and bromide measurement, it was carried out assumi ng steady-state flow conditions for the hydraulic analysis and 1-D transient storage model appl ication for transport (Martinez and Wise, 2003a and b); by contrast, 2-D, transient flow conditions were reflected in this study. Figure 4-2 shows fluctuation of wetland water volume duri ng the simulation period. The volume of each simulation time step was estimated from the wa ter depth-volume relati onship of the OEW Cell 7, which is based on the time series water depth s imulation results, and the average cell volume of 61,877 m3 was used to calculate the nominal residen ce time. Figure 4-3 illustrates six snapshots of the transient distribution on 2-D bromide c oncentration plume at the OEW Cell 7 every 12 hours after 1.5 days from the start date of simula tion. It shows that approximately two-thirds of the entire wetland cell area did not contribute to flow, and the fl ow and solute transport were mainly controlled by the short-circ uiting zone developed narrowly along the central flow path. Figure 4-4 shows the calibrated MIKE 21 model fit (solid line) for bromide measured at the outlet weir during the entire simulation peri od. The bromide concentr ation profile (closed circles) measured at the inlet weir (weir num ber 3-x in Figure 4-1) was used as the input boundary condition at the inlet ce ll for advection-dispersion simulation on bromide concentration (open circles) measured at the outlet weir (weir number 7-x in Figure 4-1). Minimum travel time, tm, defined as the shortest travel time from inlet to outlet, can be considered as a characteristic of the RTD, which identifies short-circuiting of flow in wetlands because it indicates the fastest

PAGE 81

81 flowpath through a wetland (Holla nd et al., 2004). In this stud y, minimum travel time was defined as the time of period between two points of the last measurement before the measured concentration (about 0.3 mg/L) reached 5% of the peak concentration in both inlet and outlet RTD curve. The time of each point calculated us ing this method matched the first rise of each RTD curve. The minimum travel time was de termined to be approximately 14 hours. Several other parameters to measure the hydr aulic performance of a wetland cell are also summarized in Table 4-1. Many are indicators of short-circuiting flow including a relatively short mean residence time (2.38 days), a small effective volume (24%), and very low hydraulic efficiency (15%). In the following sections, the impacts of ch ange in bathymetry and vegetation on shortcircuiting flow within a treatment wetland cel l are presented by revi ewing the results of sensitivity analysis on the two f actors, based on their spatial di stribution delineated by the 2-D bromide plume distribution. Several implications of the analysis re sults and the modeling approach used in this study ar e subsequently discussed. These se nsitivities were deduced during model calibration. Sensitivity Analysis I: The Effect of Bathymetry on Short-circuiting Flow The original bathymetry of Cell 7 for 2-D modeling was made using the inverse distance weighted interpolation scheme of MIKE ZER O Grid Series Editor, based on measured topographic elevation of 35 points, illustrated on the construction blueprint for OEW reclamation project (Post, Buckley, Schuh & Jernigan, Inc., 1 985). Unexpectedly, a considerable discrepancy was found in comparing bromide concentrations meas ured at outlet weir ( open circles) with the BTC (solid line) generated at outlet cell through MIKE 21 conservative tr ansport simulation with original bathymetry as shown in the Figure 4-5. To fit the simulation BTC to the measured bromide concentration profile adequately, it was required to find the bathymetry condition which

PAGE 82

82 contributes to decrease the time to peak or mean residence time and increase the peak concentration. Therefore, to figure out the mo st favorable 2-D bathym etry condition creating short-circuiting flows in Cell 7, five simulation cases with different bathymetry manipulation were considered under the same model conditio ns (Figure 4-5). The hydraulic efficiency calculated from the each BTC simulation is denoted for comparison in the figure. Before bathymetry manipulation, it was assumed that striking change of bromide BTC at outlet cell was primarily caused by th e change of bathymetry in th e cells located along main flow direction (short-circuiting flow zone), identified by 2-D flow velocity profile generated from the simulation using original bathymetry. First, bath ymetry levels of 0.3 m were increased in the 19 cells located within this main fl ow channel (Ridge effect in Fi gure 4-5). This corresponds to a 0.3 m high middle ridge dividing the en tire cell into two areas. In this case, the simulation result illustrates that the peak concentration is rather decreased, and the mean residence time and time to peak are adversely increased as shown by the dotted line of Figure 4-5. Secondly, bathymetry levels of 0.3 m were decreased under the same c ondition as before to create a 0.3 m deep middle ditch (Ditch effect in Figure 4-5). This simple ditch eff ect was favorable to reduce the minimum travel time but still was lacking in fitting the measured pr ofile (Figure 4-5). In the third trial, bathymetry levels of 0.3 m for both side outer cells adjacent to the cells w ithin the main flow channel area were increased without cha nging any bathymetry with in the channel, causing effects similar to construct immersed levees on both side of an imaginary channel (Two Side Ridges effect in Figure 4-5). This simulation showed a better result than the three previous trials including original bathymetry simulation but was st ill far from the reasonab le fitting. Finally, in order to mimic the highly short-circuiting characte ristic of the OEW Cell 7 flow more precisely, the previous Two Side Ridges eff ect was combined with the Ditch effect to enhance the impact

PAGE 83

83 of one large ditch, like a 1-D channel with le vees, formed along the main flow direction (Combined effect in Figure 4-5). The combined effect simulation was repeatedly carried out by using trial and error until the a cceptable curve fitting resulted. The simulation BTC, represented by the bold solid line in Figure 4-5, was generated from the decr easing the distance between two ridges, increasing height of each ridge to about 0.3 0.7 m, and decreasing the bathymetry levels of the main flow channel area to about 0.03 0.1 m. The hydraulic efficiency of this simulation (14%) is most close to that of the measured data (15%). The final adjusted bathymetry was considered as the real bathymetry domain inst ead of original one and used for the calibration of other model parameters. Bathymetric description is one of the mo st important tasks in hydrodynamic modeling process since it plays a key role in determini ng water depths in a model area (DHI, 2004b). This result shows that the bathymetry effect resultin g in the increased water depth at the main flow zone elicits a greater extent of short-circuiting, decreasing the hydrau lic efficiency. This effect of water depth on RTD characteristics was also report ed by Holland et al. (2004) from the results of dye tracer experiments conducted on a stormwater treatment wetland. Martinez and Wise (2003b) suggested that the pr esence of relic ditche s running parallel to the direction of flow and other landforms in the cell bottom was one of major factors creating short-circuiting flows experienced at the OEW. The result of this study strongly supports the fact that relic ditches and other ditch-shaped landf orms developed along the main flow direction enhance the wetland preferential flow char acteristic and reduce residence time. Sensitivity Analysis II: The Effect of Vegetation on Short-circuiting Flow The variation of flow resi stance by the spatial or tem poral distribution of wetland vegetation can affect treatment performance of constructed wetlands (Poiani and Johnson, 1993;

PAGE 84

84 Lee et al., 2004; Wrman and Kronns, 2005). In many cases, however, it is not easy to understand the effect accurately due to the lack of available data. Although hydraulic resistance parameters like Ma nnings coefficient, originated from the theory of open channel hydraulics, are known not to be directly applicable to wetland systems (Kadlec and Knight, 1996), Manning s roughness coefficient has been most widely used to describe the lumped flow resistan ce of drag due to the stems or litter of wetland vegetation as well as the channel bottom (Kadlec and Knight 1996; Loftin et al., 2001; Lee and Shih, 2004; SFWMD, 2005a). In general, Mannings coefficien t is a function of water depth as well as the spatial and temporal distributi on of vegetations in a given we tland type. In this study, roughness coefficients were assigned to each grid cell of the modeled area based on the hydraulicallydivided, conceptual flow zones (main flow channe l, storage zone, and dead zone), not simulation data on the water depth in each cell or the spatia l and temporal distribution of the vegetations. Considering the relatively short simulation period (22.75 days), a temporal effect on the change of vegetation distribution can be easily ignored. Unfortunately, th ere were, however, no reliable data, such as aerial photography or ecological su rvey, on the spatial distribution of vegetation. This is the main reason to assign coefficients al ong the flow zone delineated by the transient 2-D bromide plume distributions. In other words, th is means that spatial variation of the roughness coefficient within each flow zone was not considered. In addition to insufficient field data, this approach was used to evaluate treatment wetla nd (or hydraulic) efficiency in previous studies (Martinez and Wise, 2003a and b), so this applic ation of the same approach allowed a direct comparison between the two studies. In the previous study (Martinez and Wise, 2003a ), a 1-D solute transport model was used to estimate the ratio of the hydraulically-divided conceptual flow zones. However, the approach

PAGE 85

85 did not provide any information on the hydrauli c roughness of each zone, although separation of the zones is originally attributed to the di fference of flow resistance by wetland vegetation. Therefore, the effect of vegetation on short-circu iting flow is reviewed by the sensitivity analysis of Mannings roughness coefficients of three different hydraulic zones on bromide BTC in this study. Figure 4-6 illustrates BTCs a nd hydraulic efficiencies determined from six conservative advection-dispersion simulations in which only roughness coefficients were varied. Open circles indicate the bromide concentrations measured at outlet weir, and six different lines represent the profiles simulated at outlet cell (Figure 4-6). Ha lf of those were simulated using constant Mannings coefficients (n = 0.1, 0.05, and 0.025) through the enti re modeled area, assuming no spatial heterogeneity of vegeta tion. A significant difference among th e simulation results of these three cases was not observed, and th ey all were not sufficient in fully simulating short-circuiting flow at the Cell 7 (Figure 4-6), showing a large discrepancy of m ean residence time and variance compared to the measured data. This result shows that assigning onl y one roughness coefficient for a wetland system, which has been generally us ed in the most wetland studies, may not fully explain the hydraulic roughne ss characteristic. For the other three simulations, Manning coefficients were assigned along the hydraulically divided flow zone s: (1) hydraulic dead zone c onsisting of background area, (2) transient storage zone located at the transitional area betw een dead zone and main flow channel, and (3) main flow channel zone resul ting in short-circuiting flow. The heterogeneous distribution of vegetation that caus ed spatially non-uniform resistan ce to flow is a main factor creating short-circuiting flows, resulting in low efficiency of treatment wetlands (Kadlec and Knight, 1996; Martinez and Wise, 2003b). In this study, the main flow channel zone corresponds

PAGE 86

86 to the cells located in the ditch area between two ridges determined in bathymetry effect analysis. Some part of the cell located at the ridges was assigned as the storage zone, and the other area was classified as dead zone. First, to estimate the roughness coefficien t in the dead zone, the maximum roughness coefficient (n = 0.11 s/m1/3) for medium to dense brush floodplains in winter (n = 0.045 0.11) was selected from Chow (1959). This value was verified through comparison with other data on the roughness coefficient of wetland types sim ilar to the OEW Cell 7. According to SFWMD (2005a), the roughness coefficient can be comput ed on the basis of its vegetation and ponding depth as follows: n = ahm b (4-7) where a and b are the constants that are a function of wetland type or vegetation, and hm is average water depth (feet). The wetland t ype of the OEW Cell 7 is wet prairie (a = 0.150 and b = -0.77). As the average water depth was regarded as approximately 0.6 meter at the Cell 7 during the entire simulation period, the roughness coefficient of 0.13 (s/m1/3) was calculated using Equation 4-7. In addition, Mannings coefficient in wet prairie was reported as 0.10 by Loftin et al. (2001). Therefore, the value (n = 0.11) chosen as the back ground area roughness coefficient in this study seems to be reasonable. With the estimated value of 0.11 for dead zone roughness coefficient, th ree different trials for estimation of coefficients in storage and main flow channel zones were carried out. First, the minimum roughness coefficient value (n = 0.045; Chow, 1959) for medium to dense brush floodplains in winter was assigned for the main flow channel zone, and the value for storage zone was assumed to be same as the one for the dead zone (Variant n=0.11/0.11/0.045 in Figure 4-6). Although the simulation was not completely satisfactory, the result, as illustrated by dotted

PAGE 87

87 line (Figure 4-6), was much closer to the real short-circuiting pa ttern of the Cell 7. For the next trial, the minimum (0.022 s/m1/3) and maximum roughness co efficients (0.033 s/m1/3) for straight and uniform earth dredged ar ea with short grass and few weeds (n = 0.022 0.033; Chow, 1959) were assigned for the main flow ch annel and storage zones, respectively (Variant n=0.11/0.033/0.022 in Figure 4-6). In this case, it seems that the simulation slightly overestimated the bromide data (Figure 4-6). Fina lly, the last simulation was tried using both end values (0.035 s/m1/3 for the main flow channel and 0.07 s/m1/3 for the storage zone) of the roughness coefficient range for scattered brush floodplains with heavy weeds (n = 0.035 0.07; Chow, 1959). The simulated BTC, was between the previous two simulation profiles, is likely to show the best simulation fitting on the measured BTC (Figure 4-6). Hence, the approximate values of 0.035 and 0.07 s/m1/3 can be suggested as the roughness coefficients in the faster-moving flow area a nd the transition zone with the value of n = 0.11 s/m1/3 in the dead zone covering the other area of the cell. According to the model used by Jenkins and Greenway (2005), the non-vegetated part of the wetland was assumed to have a Manning coefficient of 0.035. However, it may be more reasonable to estimate the roughness coefficient value as a certain range than a cons tant value, considering possible deviations like analytical measurement error. In addition, the hydrodynamic metric s of the last three simulation trials did not show big differences among the ca ses, as represented by hydraulic efficiency. Therefore, as the range of Manning roughness co efficient of the main flow channel area, corresponding to the short-circuiting flow zone, 0.022 to 0.045 s/m1/3 may be suggested for a surface flow constructed wetla nd setting like the OEW Cell 7. Implications of This Modeling Study In this study, a 2-D, physically -based, distributed hydraulic a nd solute transport model was developed using MIKE 21, based on the hydraulic measurement and bromide tracer experiment

PAGE 88

88 carried out at the OEW Cell 7. Considering the tr ansient flow conditions, it simulated measured bromide BTC successfully, and the hydraulic perfo rmance was compared to results of previous studies. In this section, four implicati ons of this modeling study are reviewed. Firstly, the application of hydraulic efficiency (), proposed by Perss on et al. (1999) will lead to more reduced hydraulic efficiency of the entire OEW than the value reported by Martinez and Wise (2003a and b) because the efficiency in these studies was defined as the metric of effective volume (e). When the alternative parameter, was applied, the hydraulic efficiency of 24% was reduced to 15% at the OEW Cell 7. Th is was also observed during simulations for bathymetry and vegetation impact. Therefore, this shows the necessity of an effort to compare the hydraulic performances of constructed wetlan d systems more consistently, such as shortcircuiting and hydraulic efficiency, avoiding confusion of terminology. Secondly, impacts of changes in bathymetry an d vegetation distributi on on short-circuiting flow can be compared in terms of hydraulic efficiency (). The deviation from the hydraulic efficiency of measured data (15%) is much hi gher during the simulations for bathymetry effect (14-45%) than the vegetation effect (14-18%). Th is indicates that the bathymetry effect on hydraulic efficiency is more sensitive than the ve getation effect. Recent studi es also highlight the impact of water depth among the factors affec ting hydraulic efficienc y. Holland et al. (2004) reported that water depth, rather than flow rate has a direct impact on the RTD of a wetland, and Kadlec (2005) also suggested th at lower pond efficienci es, compared to wetla nds, are primarily due to a depth effect, not the absence of vegeta tion. Therefore, this modeling study suggests that the impact of water depth change induced by bath ymetry change may be greater than that of vegetation distribution on hydrauli c efficiency, especially show ing that increasing the water depth due to the ditch-shaped bathymetry causes a decrease in hydraulic efficiency.

PAGE 89

89 Next, this study reviewed the impacts of the two physical settings of a constructed wetland on the hydraulics, represented as short-circuiting flow. MIKE 21 described those successfully. In general, most traditional wetland flow models depend on 1-D, conceptual, parameters-lumped modeling approach under unrealistic hydraulic cond itions, such as plug-flow and steady-state conditions. They assume unrestricted chances fo r contact between the incoming water and the organisms responsible for treatment (Persson, 2000). In addition, it is not easy to find the physical meanings of their key parameters and not possible to estimate the effects of various physical and ecological factors affecting wetland hydraulics inde pendently. In contrast, this simulation results show that a 2-D, physically based, distributed model can overcome most limitations that traditional wetland flow models have, provided suitable data for the model are available. This study also plays a role as an ex ample showing which data should be collected for the model development. Finally, water flow is a basic component in mo st solute transport/re tention models because it is the medium where solutes, including various pollutants, are transported or transformed. In terms of treatment wetland efficiency, manipulat ing the hydraulic regime to increase the treatment efficiency is a desirable strategy for constructed wetlands (Wang and Mitsch, 2000), and understanding the removal or retention of pollutants on a ba sis of accurately-simulated wetland hydraulics is essential. In this study, th e MIKE 21 simulation results of the OEW Cell 7 bromide tracer test show its capability as a r obust modeling tool for more advanced flow-solute transport-water quality integrated dynamics models in constructed wetlands. Therefore, it may be broadly applied for development of a useful m odeling tool for quantitatively understanding the internal processes associated with pollutant removal/retention/release under transient flow conditions (e.g. stormwater), testing various hyp otheses according to th e changes of hydraulic

PAGE 90

90 regime, and making more realistic predictions. A series of model development and application procedures covered in the following two chapters (Chapter 5 and 6) serve as an example showing how this approach suggested above is actualized. Summary Short-circuiting flow, commonly experienced but not desirable in many constructed wetlands, reduces hydraulic retentio n times in unit wetland cells, ul timately decreasing treatment efficiency. In this study, a 2-D, transient hydrodynamic and advection-dispersion model was developed using MIKE 21 and calibrated with brom ide tracer data collected at the OEW Cell 7. The treatment cell experiences highly short-circ uiting flow, proved by the metrics showing the hydraulic performance including the efficiency of 15%. Simulation results of this study indicate that the RTD of a treatment wetland cell is high ly sensitive to changes in bathymetry and vegetation distribution. Sensitivity analysis on topographical and vegetative heterogeneity deduced during model calibration sh ows that relic ditches or othe r ditch-shaped landforms and the associated sparse vegetation along the main flow direction (channeling effect) intensify the short-circuiting pattern, considerab ly affecting 2-D solute trans port simulation. The range of the Manning roughness coefficient at the short-circ uiting flow zone was estimated (0.022 0.045 s/m1/3). In terms of hydraulic efficiency, this study indicates that the bathymetry effect on shortcircuiting flow is more important than the vege tation effect. These two effects on short-circuiting flow observed in this study should be fully cons idered; in particular, when a 2-D, physically based, fully distributed modeling approach is used as a modeli ng tool for the advanced design and management of constructed wetland.

PAGE 91

91 Table 4-1. Physical parameters used in models and metrics showing the hydraulic performance of the OEW Cell 7. Previous research* This study Cell area (m2) 117,000117,900 Cell volume (VAVG, m3) 67,00056,600 ~ 73,200 (61,877**) Average cell flow (QAVG, m3/s) 0.0600.072 Nominal residence time ( days) 13.010.00 Mass recovery fraction (MRF, %) 9796 Mean residence time ( t, days) 2.632.38 Variance ( 2, days2) 18.38.8 Effective volume (e, %) 20 (31***)24 Time to peak (tp, days) 1.47 The number of CSTRs in series (N) 2.6 Hydraulic efficiency ( %) 15 Martinez and Wise (2003b). ** Average wetland volume during the simulation period. *** Expressed as the sum of the volume of the main channel and storage zone, determined by OTIS (Martinez and Wise, 2003a).

PAGE 92

92 Figure 4-1. Location and plan view of study area, the OEW Cell 7.

PAGE 93

93 Figure 4-2. Fluctuation of wetland water volume during the simulation period. The volume was estimated from the water depth-volume relationship at the OEW Cell 7, which is developed based on the simulation data.

PAGE 94

94 Figure 4-3. Bromide tracer experiment si mulation: transient 2-D bromide plume.

PAGE 95

95 Figure 4-4. Bromide tracer experiment simu lation: model fit on the measured BTC.

PAGE 96

96 Figure 4-5. The effect of bathym etry on short-circuiting flow.

PAGE 97

97 Figure 4-6. The effect of vegetation di stribution on shortcircuiting flow.

PAGE 98

98 CHAPTER 5 MODEL APPLICATION TO THE STORMW ATER TREATMENT AREA (STA) 5 NORTHERN FLOW-WAY: I. FLOW DYNAMICS AND SOLUTE TRANSPORT Introduction Use of constructed wetlands for nutrient remova l, particularly phosphorus in South Florida, from nutrient-enriched, upland agricultural runoff or wastewater is being extensively recognized as one of the most feasible and effective tec hnologies in the world (Mitsch, 1994). To optimize design and management of constructed wetlands and to predict performance of treatment wetlands under various hydrological and ecological conditions, such as inlet flow/mass loading rate and vegetation type/densit y, a flow dynamics and solute tr ansport coupled model calibrated with field data is necessary. This chapter presents results of flow dyna mics and solute transport model on STA 5 northern flow-way, developed using MIKE 21 HD and AD module. Coupled with an ECO Lab phosphorus dynamics model described in Chapter 3, th is model is also applied to simulate the behavior of water column phosphorus species in the north ern flow-way (Chapter 6). Study Area To fulfill the goals of the 1994 Everglades Fore ver Act to improve surface water quality in the Everglades, various restoration efforts have been accomplished through construction, research, and regulatory activities. One of the main efforts is to construct six strategically located treatment wetlands, referred to as STAs. A regional map showing location of the STAs a nd significant features in the South Florida landscape is presented in Figure 5-1. STA 5, one of the six constr ucted wetlands, is located in Hendry County, Florida. It extends from the L-2 Canal on the west to the Rotenberger Wildlife Management Area on the east (Figure 5-2). The antecedent land use of STA 5 was an agricultural cropland for sugar cane (Gofort h, 2005). STA 5 receives untreated agricultural

PAGE 99

99 stormwater runoff from the C-139 Basin via th e L-2 Canal. If runoff exceeds the hydraulic capacity of STA 5, flow is diverted through G 406. Treated water is collected and discharged either to the Rotenberger Wildlife Management Area or the Miami Canal (Figure 5-2). Seepage collection canals are located al ong the northern and southern bounda ries of STA 5 in order to return seepage via two pump station (G349A and G350A) to the upstream treatment cells (Figure 5-2). The role of these hydraulic structures is to avoid STA 5 dry-out under drought condition. STA 5 consists of four treatment cells (1A, 1B 2A, and 2B), which are divided by a perimeter levee and connected by culverts and weirs. Tota l area of all treatment cells is approximately 1663 ha (SFWMD, 2000). Figure 5-2 illustrates a schema tic of STA 5, indicating flow and vegetation patterns as well as hydraulic structures around the STA. The north ern flow-way of STA 5, which is the main study area in this study, consists of two consecuti vely linked treatment cells (Cell 1A and 1B). They contain approximately 338 and 494 ha of ef fective treatment area, respectively (SFWMD, 2000). Water enters the flow-way from the we st (L-2 borrow Canal) and flows by gravity through the treatment area to the east (Discharge Canal). Within the flow-way, there are two dominant vegetation communities: EAV and SAV. Cell 1A and 1B are EAV and SAV dominant treatment cells, respectively. EAV species found in STA 5 Cell 1A are typically Typha spp. (cattail) and Ludwigia spp. (primrose willow); on the other hand, STA 5 Cell 1B is mostly comprised of invasive exotic SAV species, Hydrilla verticillata, and periphyton, which covers approximately 20% of open water with Hydrilla (Table 5-2). As shown in Figure 5-2, inflow to the northern flow-way from the L-2 Canal is controlled by two gated concrete box inflow culverts (G342A and G342B) and a pump station for seepage retu rn (G349A). The flow induced through the culverts is conveye d through spreader canals, which separate high topographic

PAGE 100

100 elevation area (west of the spread er canals; dry most of the ti me during normal operation) with the upstream treatment cell. Water level of Cell 1A is also controlled by four concrete box culverts with upstream weirs (G343A to G343D ) located on the middle levee between Cell 1A and 1B. Likewise, water level of Cell 1B is controlled by two gated concrete box outflow culverts (G344A and G344B). More detailed descriptions on the operati onal and management history as well as the physical characteristics of STA 5 are contained in the chapters de tailing STA performance in the annual South Florida Environmenta l Reports (Pietro et al., 2006; Pietro et al., 2007) and other SFWMD documents (SFWMD, 2000; Goforth, 2005). Specific Aims The main goals of this chap ter are (1) to develop a dept h-averaged flow dynamics and solute transport model using MIKE 21 HD and AD modules, which can be easily applied in various constructed wetland systems ranged from small-scaled ponds to large-scaled constructed wetlands such as South Florida STAs and (2) to suggest values (or range s) of key parameters commonly applied, but not extensively studied in areas covered by vari ous wetland vegetations, in flow dynamics and solute transport mode ls (e.g. hydraulic roughness coefficients and dispersion coefficient) through the model calibration, which may be directly or indirectly available to constructed wetland systems similar to the STA 5. Materials and Methods Recent tremendous progress in GIS has led to significant advances in various types of physically-based, distributed hydr ologic/hydraulic models (War wick and Haness, 1994; Tsanis and Boyle, 2001). In general, GIS offers spatial data management and analysis tools to assist experienced users in organizing, storing, editing, analyzing, and displaying local and attribute information about geographic data (Xu et al., 2001). Many studies on GIS-based, fully

PAGE 101

101 distributed hydrologic/hydrodynami c modeling have been report ed (Smith, 1993; Leon et al., 2001; Tsanis and Boyle, 2001; Xu et al., 2001; Qi and Grunwald, 2005). Thus, GIS technology not only supplements the shortcomings of fully distributed models mentioned in Chapter 2, but also can enhance model performance. In this section, model setups including the estimation of key model parameters through GIS-based estimation and mapping as well as hydr ologic and water quality data required in the HD and AD models are reviewed. Data Used in Models The hydro-meteorological data of STA 5 norther n flow-way, such as flow, water level, precipitation, and evapotranspira tion, used in the HD model we re collected by SFWMD and are available on the online envir onmental database, DBHYDRO (http://glades.sfwmd.gov /pls/dbhydro_pro_plsql/ show_dbkey_info.main_page ). These flow and water level data from DBHYDRO are typically available as a daily average value for each hydraulic structure around the flow-way illustrated in Fi gure 5-2 (G342A, G342B, G349A, G343B, G343C, G344A, and G344B). The flow data measured at three input hydraulic structures (G342A, G342B, and G349A) and two output hydra ulic structures (G344A and G344B) and precipitation/evapotranspira tion data were used as a model i nput data, and the water level data measured at each hydraulic structure were used for model calibration, validation, and sensitivity analysis. Chloride data used in the AD model were also collected by SFWMD and downloaded from DBHYDRO. In this study, chloride was selected as a tracer for the conservative solute transport model because it was measured biweekly at al most every water quality measurement point during a relatively long period. Time series chloride concentration data measured at three input points were used as a model input, and those measured at middle levee (G343B and G343C) and

PAGE 102

102 output structures (G344A and G344B) were used for model calibration, validation, and sensitivity analysis. Water budget The water budget at STA 5 northern flow-way is comprised of the following components: Surface water inflow through pump and gated culverts (ISF) Surface water outflow through gated culverts (OSF) Rainfall (P) Evapotranspiration (ET) Net groundwater seepage (OGW) Change in storage ( S) Water budget error (R) The budget is developed for varying time periods using the following equation: R O ET O P I t SGW SF SF (5-1) In Equation 5-1, all terms have the same unit; for example, cubic hectometer (hm3 = 1,000,000 m3) per month as shown in Table 5-1. Rainfall and evapotranspiration values were multiplied by the model domain area representing the total effective trea tment area (8,140,000 m2) to obtain a volume of rainfall or evapotranspira tion for a certain period of time. In the study area, daily rainfall data were co llected at G343B, and evapotranspiration was estimated for STA 5 based on data from a near by monitoring station (STA 1W), located approximately 53 km to the northeast of STA 5. The ET data are based on a prediction equation suggested by Abtew (1996). Figure 5-3 shows the precipitation and evapot ranspiration in STA 5 during the entire model simulation period (2.67 years) Daily-averaged headwa ter stage, tailwater stage, and gate opening were measured to calculat e a daily average flow rate using a culvert or pump performance equation at each hydraulic stru cture. The time series surface water inflows and outflows in the northern flow-way are presente d in Figure 5-4. No direct measurement of net groundwater flow was made. Alt hough Parrish and Huebner (2004) and Liyanage and Huebner

PAGE 103

103 (2005) used a linear equation with seepage coeffi cient to compute the gr oundwater seepage rate in STA 5; in this study, the single unknown component was initially determined through minimizing water budget error and finally set wi th the minimum deviation of water level simulations. Liyanage and Huebner (2005) reported that a net seepage loss was dominant in the entire flow-way; in general, the seepage flow direction was into the treatment cells (Cell 1A and 1B) from the L-2 Canal and STA 5 southern flow-w ay (Cell 2A and 2B) and out of the treatment cells toward the northern seepage canal and th e Discharge Canal along the eastern boundary. Monthly water budgeting for the HD simulation of STA 5 northern flow-way is presented in Table 5-1. Surface water inflow through G 342A, G342B, and G349A constitutes 93.2 percent of the total inflow to STA 5 northern flow-way. Rainfall is 6.8 percent of total inflow. Surface water outflow through G344A and G344B is 88.4 percent of total outflow. ET and seepage losses are estimated to be 7.4 percent and 4.2 per cent, respectively. For details on estimation of water budget components through hydro-meteorologica l monitoring at STA 5, refer to Liyanage and Huebner (2005). Basic Parameters Setup Although the northern flow-way of STA 5 was flooded in 1999, performance data for the first three years are not analyzed in this study be cause the interval is us ually considered a period for start-up processes like ve getation colonization (Juston and DeBusk, 2006). Hence, the simulation period for HD model calibration is from May 1, 2002 to April 30, 2003 (1 year) and the simulation period for the model validation starts on May 1, 2003 and ends on December 31, 2004 (1.67 years). Recent flow data after January 1, 2005 are not also considered for model simulation because the northern flow-way was off-line from January 2005 to improve the midlevee hydraulic structures according to Long-Term Plan construction (Pietro et al., 2006). The simulation time step is 1 minute.

PAGE 104

104 Considering the location of inflow hydrauli c structures (G342A and G342B) and seepage return pump (G349A), three point source grid cells are assigned on the left side of the model domain to incorporate the rate, velocity, and direction of inflow. During the simulation period, daily-averaged flow rates were divided by flow passing area of hydraulic st ructures to calculate daily average velocities. Likewise two point sink grid cells ar e assigned on the right side of model domain to incorporate the outflow. Unfo rtunately, there is no option for groundwatersurface water interaction in MIKE 21; however, 15 grid cells along the northern boundary of the flow-way were assigned as a poi nt source/sink to reflect net gr oundwater seepage flow in this modeling study. To simulate flooding and drying condition of wetlands, 0.03 m of flooding depth and 0.001 m of drying depth are selected, respectively. Bathymetry Bathymetric data are among the most critical model inputs for the HD model setup. The latest STA 5 topographic survey data (Wan tman Group, 2005) was provided by the SFWMD for this study. The 216 georeferenced bathymetry m easurement points (NGVD29) are interpolated using inverse-distance weighting (IDW) scheme of GIS to generate continuous bathymetry prediction map for the study area (Figure 5-5). Aver age bathymetry elevati on in Cell 1A (only effective treatment area) is 4.08 m NGVD, and th e average ground elevation in Cell 1B is 3.61 m NGVD. A raster-based bathymetry prediction map shown in Figure 5-6 is used in the HD model, which is made through conversi on to raster in ArcGIS 9.0. Hydrodynamic Model Setup Water levels observed at nine measurement points within the northern flow-way on the simulation starting date (May 1, 2002) were linearly interpolated to set the initial condition for water level of model domain. Differences of wate r level within Cell 1A and 1B were about 0.08

PAGE 105

105 m and 0.04 m, respectively. About 0.3 m of abrupt water level decrease was observed between headwater and tailwater elevati on of hydraulic structures locate d in the middle levee separating Cell 1A and 1B, and approximate ly 0.45 m of water level diffe rence was observed from the upstream (east of spreader canal) to downstream area (west of Discha rge Canal) within the entire flow-way. Water levels of L-2 Canal and Discha rge Canal and the relationships with treatment cells were not considered in this study. Time series rainfall and ev apotranspiration data measured during the simulation period (Figure 5-3) were homogeneously assigned to every grid cell of model domain. Time series profiles of surface water inflow measured at G342A, G342B, and G349A (Figure 5-4A) were incorporated at the three point source grid cells, and those of surface water outflow measured at G344A and G344B (Figure 5-4B) were incorporated at the two point sink grid cells, which were defined in basic parameters setup. Time seri es monthly-averaged groundwater seepage flows (Table 5-1), estimated as a residual of water budge t, were divided by the nu mber (15) of the grid cells assigned for the seepage flow and equally a ssigned to each the grid cell. This means that short-term (daily or weekly level) temporal and spatial variations of grou ndwater seepage flow in the study area are not cons idered in this study. Hydraulic resistance Since hydraulic resistance, usually repres ented as Mannings roughness coefficient (n) values, is closely related to ve getation type/density and water depth, it is spatio-temporally varied. In this study, it is assume d that hydraulic resistance is i ndependent of change of flow depth; that is, it is temporally constant. Therefore, it is described only as a function of vegetation type and density. There are few good references on selection of values for vegetated cover in 2-D HD models (Sutron Corp., 2005).

PAGE 106

106 A GIS shapefile for vegetation distribution of STA 5 northern flow-way (Nick Miller, Inc., 2006) is presented in Figure 5-7. Th e entire vegetation is classified into five different types. Table 5-2 shows the dominant vegetation specie s and ratios. Using tw o attribute fields (vegetation type and percent cove rage), which were stored in th e vegetation shapef ile, two raster files for vegetation type and pe rcent coverage were generated through raster conversion. Then, five different percent coverage maps with resp ect to each vegetation type were extracted using Raster Calculator of ArcGIS 9.0. To date, there is little avai lable knowledge on assignment of the coefficients according to wetland vegetation density. Hence, in this study, the percent coverage in a vegetation type was manually reclassified into four or five classes, considering the histogram clustering pattern. Then, appropriate Mannings roughness coeffi cients were assigned to the each reclassified class. Figure 5-8 illustrates the spatial distribution of Mannings roughness coefficients used in the HD model, assigne d at every grid cell of Cell 1A and 1B as a function of the type and density of vegetation. Advection-Dispersion Model Setup AD model application was accomplished based on chloride data monitored during the period of May 1, 2003 to December 31, 2004, which co rresponds to the validation period of the HD model. Since the AD model is integrated with th e HD model, it is first required to recalibrate the HD model to minimize the AD simulation error caused by the HD simulation error. For this, monthly-averaged net groundwater seepage flows we re modified to generate the best HD model fit during the AD simulation peri od. Table 5-3 shows the monthly water budget reset for the AD model of STA 5 northern flow-way, and the hydrop eriod simulation results of the recalibrated HD model are presented in Appendix A. The average annual model prediction error is approximately 0.11 m.

PAGE 107

107 Chloride concentrations are monitored at seven hydraulic structur es in STA 5 northern flow-way (G342A, G342B, G349A, G343B, G343C G344A, and G344B). For specifying the initial chloride concentration of the study area, concentrations observed at the seven water quality measurement points on the simulation starting date (May 1, 2003) were linearly interpolated to generate the 2-D continuous map. Time series profiles of inlet chloride concentration measured at G342A G342B, and G349A (Figure 5-9) were incorporated at each the point source grid cell. In this model, to represent the atmospheric input of chloride by precipitation, tempo-spatially constant 1 mg/L of chloride concentration was set as the precipitation concentration. Dispersion coefficients in x a nd y direction were set as a time-varying parameter according to the change of flow velocity at each time step. In this case, dispersivities in x and y direction should be specified. As an initial guess, disper sivities of 0.5 and 0.05 m were assigned in x and y direction ( x: y = 10:1), respectively and were finally determined by model calibration ( x = 2 m and y = 0.1 m; the ratio is 20:1). The parameter estimation by model calibration is discussed later in this chapter. Model Calibration and Validation The HD model calibration and validation were pe rformed with historic stages in STA 5 northern flow-way. For the AD model, they were performed with the tim e series profiles of chloride concentration observed in study area. In this study, model calibration was accomplished by calculating the value of root-meansquare error (RMSE) between measured and si mulated data as shown in Equation 5-2. RMSE generally shows how much simulate d values deviated from measured values. Hence, as a metric of model fit, the calibration proce ss was repeated to minimize RMSE.

PAGE 108

108 Root-mean-square prediction error: n t M t Sn i i i1 2) ( ) ( (5-2) where S ( ti) and M ( ti) indicate simulated and measured data at a time step ( ti), respectively, and n is the number of total data during a simulation period. Sensitivity Test After obtaining the best model fit between simulated and measured data by tuning model parameters, model sensitivity analysis on key mode l parameters is carried out. In this study, the sensitivities of simulated water levels (in the HD model) and simulated chloride concentration profiles (in the AD model) at the measurement poi nts of study area to the key model parameter (hydraulic roughness coefficient in the HD model and dispersiv ity in the AD model) were examined. The parameter values estimated by the model calibration were used as baseline values. In each sensitivity mode l run, only single m odel parameter was changed within a certain range. In this study, results of these sensitivity test s are expressed by comparing the RMSE values of calibration and sensitivity test model run or using a RMSE value, which is defined as the difference between RMSEs generated by model sensitivity test (RMSEs) and calibration (RMSEc) as follow: RMSE = RMSEs RMSEc (5-3) Results and Discussion Hydrodynamics Model In this study, time series simulated water le vel profiles, extracted from two grid cells corresponding to the headwater measurement poi nts (G343B_H and G343C_H) of two hydraulic structures located at the middle levee, were comp ared to the measured data to verify simulation

PAGE 109

109 results of the HD model in Cell 1A. Likewise, si mulation results in Cell 1B were verified by comparing measured data with simulated water level profiles, extracte d from four grid cells corresponding to the tailwater measurement points (G343B_T and G343C_T) of two hydraulic structures at the middle levee and headwater measurement points (G344A_H and G344B_H) of two outlet hydraulic structures at the downstream levee. Hydroperiod simulation Understanding spatio-temporal changes of hydroperiod in a constructed wetland system plays a key role in optimizing the management and maximizing the treatme nt efficiency. Figure 5-10 and 5-11 illustrate simulation results of th e HD model on hydroperiod fluctuation at the six water level measurement points during model ca libration (May 1, 2002 to April 30, 2003) and validation period (May 1, 2003 to December 31, 2004), respectively. Calibration results demonstrate that tempo-spatial changes of hydrope riod in the flow-way are well simulated using the HD model (Figure 5-10). The average annua l model prediction error on six measured hydroperiod profiles is approximately 0.09 m. Average simulation RMSE of the HD model in Cell 1A (0.085 m) is less than that in Cell 1B (0.096 m). Although few deviations are observed in a short period of time (i.e. December 2002), simu lation results on the general water level show a decreasing pattern from upstream to downstream, and timing and intensity of water level peaks at each monitoring point agreeing very well with those of measured data. The HD model was also validated with an indepe ndent data set of 1.67 years. As illustrated in Figure 5-11, simulated water level profiles in Cell 1B, particularly at G344A and G344B, are slightly overestimated through the entire simulation period. This i ndicates that r eal groundwater seepage loss in Cell 1B is slig htly higher than the amount estim ated as the monthly-averaged water budget residual. In real ity, the overestimated pattern was not observed when the

PAGE 110

110 groundwater seepage loss was adjusted (entirely in creased) in order to obta in the best HD model fit during this validation peri od (Table 5-3; Appendix A). As discussed in Chapter 4, topographical hetero geneity in bathymetry is one of the primary factors causing uneven flow ways in a wetland system. The location of high bathymetry elevation areas was revealed through the 2-D hydrodynamics si mulation. These areas, in particular located ahead of G 343C, are dry or very shallow even in storm season. Although no obvious major short-circuiting flow ways are observ ed, it is obvious that some local high surface elevation areas, which experience frequent dry-outs in the flow-way, contribute to generate the spatial differences of time series data profiles (water quality as well as water level) among the measurement points. In this study, small variatio n of model prediction errors among the six water level measurement points indicates that this 2D HD model is spatially well-balanced enough to catch differences of 2-D data formed by the sp atial heterogeneity of physical settings like bathymetry and vegetation distribution. All results also reveal that hydroperiods of constructed wetlands can be successfully simulated through the simple applications of a source/sink option in MIKE 21 HD model rather than complex boundary specification processes. Se veral factors causing the simulation errors are addressed in Chapter 7, focused on a couple of li mitations on modeling of flow dynamics used in this study. Parameter estimation: hydraulic roughness coefficients Theoretical approaches to the flow dynamics in free surface natural or constructed wetlands do not have a long history compared to other surface water flow systems. Most theories originated from open channel hydraulics. Ho wever, there are still many unknown or controversial issues in describing the flow dynamics by applying the theories to free surface

PAGE 111

111 wetland systems. One of these is to estimate th e hydraulic resistances of vegetation-covered areas. Coefficient values are initially selected c onsidering the range of Mannings coefficient values suggested in Chapter 4 and previous l iterature data (Kadlec and Knight, 1996; SFWMD, 2005b; Sutron Corp., 2005) and finally determined, as shown in Table 5-4, by the HD model calibration with the historic st age data observed in study area For dense EAV and SAV areas, 0.67 to 1.0 and 0.12 to 0.15 s/m1/3 of the roughness coefficient ra nges were estimated in this study, respectively. Table 5-5 shows coefficients for effective r oughness (N), which are similar to Mannings coefficients, used in the South Florida Wa ter Management Model (SFWMM) (SFWMD, 2005b). The N values are calculated as a function of land use (wetland) type and water depth. In the case of a cattail-dominant wetland, 0.70 s/m1/3 is suggested when a ponding depth of a grid cell is 0.61 m (nominally 2 ft). In the 2-D hydrauli c modeling study of STA 5 accomplished by Sutron Corp. (2005), 0.5 to 1.3 and 0.3 to 0.8 s/m1/3 of the roughness coefficient values were assigned with the change of water depth from 0.15 to 0. 91 m (nominally .5 to 3.0 ft) on cattail and SAVdominant treatment cells, respectively. Although it is not easy to compare Mannings coefficient values estimated in this study di rectly with the historically re ported roughness data, parameter values estimated by model calibration are not sign ificantly different with the ranges suggested by SFWMD (2005b) and Sutron Corp. (2005). Sensitivity analysis A sensitivity analysis was completed to investigate the range of water level sensitivity due to % variation in Mannings roughness coe fficient. A 30% increase of Mannings roughness coefficients caused an average difference of 0.003 m in water level at the six measurement points; on the other hand, 30% d ecrease of the coefficients br ought about an average difference

PAGE 112

112 of 0.009 m in water level (Table 5-6; Figure 512). When Mannings coefficient values were varied by %, water level variations in Cell 1B (G343B _T, G343C_T, G344A, and G344B) were higher than in Cell 1A (G343B_H and G3 43C_H). This shows that water level in SAV system is more sensitive to changing hydraulic resistance than that in EAV dominant system. These sensitivity simulation results show th at estimated Mannings roughness coefficient ( n ) values for STA 5 northern flow-way vegetation are reasonable because the RMSE value of model calibration is less than those of % of variation for sensitivity analysis. Advection-Dispersion Model In this study, time series simulated chloride c oncentration profiles, extracted from the grid cells corresponding to G343B and G343C, were comp ared to measured data to verify simulation results of the AD model in Cell 1A. Likewise, si mulation results in Cell 1B were verified by comparing measured data with simulated chloride concentration profiles, extracted from the grid cells corresponding to G344A and G344B. Chloride transport simulation Figure 5-13 and 5-14 illustrate simulati on results of the AD model on chloride concentration profiles at the f our water quality measurement poi nts in the flow-way during the model calibration (May 1, 2003 to April 30, 200 4) and validation period (May 1, 2004 to December 31, 2004), respectively. Simulated results of chloride concentration profiles agree well with measured data during the entire simula tion period. The average value of annual model prediction errors at the four measurement points is 13.48 mg/L, and prediction errors in Cell 1A (average: 9.62 mg/L) are less than those in Cell 1B (average: 17.33 mg/L). These remarkable calibration and validation results of the AD model ensure the robus tness of HD model as well as AD model.

PAGE 113

113 Parameter estimation: longitudinal and transverse dispersivity Calibration results of the chloride tran sport AD model according to the ratio of longitudinal (x direction; x) vs. transverse dispersivity (y direction; y) in STA 5 northern flowway are presented in Appendix B. The curves of Figure 5-15 were delineated based on the calibration results (Appendix B). In case of the curve with the ra tio of 1:1 between x and y, it has the lowest average RMSE value of 16.21 mg/L when the x is 0.9 or 1.0. As the ratio of x and y gradually increases by 20:1, the lowest av erage RMSE value of each calibration curve decreases to 15.82 mg/L. However, once the ratio is greater than 20:1, the lowest value starts to increase again. As demonstrated in Figure 515, the curves have the minimum average RMSE value when the x is around 2 m, although there are very slight differences among the curves. This suggests that the average RMSE of chlori de transport simulation has the lowest value (15.82 mg/L) when the x is 2 m and the ratios of x and y are 10:1 or 20:1 (Appendix B). These results show that the assumption of homogeneous and isotropic disp ersion coefficient, which is commonly used in 2-D surface water solute trans port models, should be carefully considered in constructed wetland systems similar to STA 5. In this study, 2 and 0.1 m of longitudinal and transverse dispersivities were finally selected thro ugh the model calibration. Using calibrated dispersivity in x (2 m) a nd y direction (0.1 m), ranges of dispersion coefficients ( Dx and Dy) in STA 5 northern flow-way were estimated based on ranges of x and y directional flow velocities ( u and v ). According to HD simulation results on study area, x and y directional flow velocities ar e minimal value of zero when wa ter is stagnant and have a maximum value of 0.14 m/s in the hydraulic struct ure. Rather than usi ng these extreme flow velocity conditions, it seems reas onable to use a range (0.001 to 0.05 m/s) to represent the change of flow velocity within the treatment cells. Considering the practical range of flow velocity, the ranges of Dx and Dy were estimated from 0.002 to 0.1 m2/s and 0.0001 to 0.005

PAGE 114

114 m2/s, respectively. Table 5-7 shows literature va lues for dispersion coefficient in constructed wetlands. The ranges estimated in this study l ook reasonable compared to other dispersion coefficients. Sensitivity analysis Sensitivity analysis results of the chloride transport model on longitudinal and transverse dispersivity in study area are shown in Table 58 and 5-9, respectively. Table 5-8 presents the AD model sensitivity on the change of x (0.1 to 10 m) when y is fixed to 0.1 m; on the other hand, x is fixed to 2 m in order to test the sensitivity on the change of y (0.02 to 2 m) as shown in Table 5-9. Compared to the baseline value of 15.82 mg/L as the average RMSE value, the RMSE values are generally hi gher in sensitivity tests on x than y (Table 5-8 and 5-9). This reveals that the longitudinal disp ersivity is a more sensitive pa rameter in the AD model than the transverse one. In this study, since the main flow direction is almost para llel to the x axis of model domain, x and y are identical to longitudinal and transv erse dispersivity. Th erefore, modelers should be careful in directly applying values or ranges suggested in this study to other constructed wetland systems because the axis of m odel domain is not always parallel to the main flow direction. If parallel, it sa ves time and effort for model calib ration to consider the effect of transverse dispersivity after calibrating a model fi rst with respect to the lo ngitudinal dispersivity. Summary Based on hydro-meteorological field data measured from May 2002 to December 2004 in the northern flow-way of STA 5, the HD model was developed. The model incorporates timevarying, daily-based measurements of water budget components such as stage, flow, rainfall, and evapotranspiration, and monthly-averaged net groundwater seepage is determined through an effort to minimize the monthly water budget error. Water levels measured on the simulation

PAGE 115

115 starting date were linearly interpolated to set the initial condition for water level. GIS-based analyses were carried out for estimating and ma pping bathymetry and hydraulic resistance. In this model, groundwater flow as well as surface flows through hydraulic st ructures is described simply using a point source/sink option. The results of HD model calibrati on (1 year) and valid ation (1.67 years) on the water level profiles observed in field demonstrate that the spatio-temporal changes of hydroperiod in the study area are reasonably simulated. The averag e annual model prediction error on six measured hydroperiod profiles is less than 0.10 m. Thr ough the model calibration process, ranges of Mannings roughness coefficient for each vegetation ha bitat were also estimat ed as a function of vegetation type and density. For dense EAV and SAV areas, 0.67 to 1.0 and 0.12 to 0.15 s/m1/3 of the roughness coefficient ranges are suggested in this study, resp ectively. These results reveal that the hydroperiod of construc ted wetlands can be successfully simulated through simple applications of a source/sink option in MI KE 21 HD model without specifying any flow boundary condition, and emphasize the importance of GIS-based estimation on key model input data as well as tempo-spatially intensive hydrolog ic/hydraulic data collection in developing 2-D, physically-based, fully distributed hydrodynamics models. The AD model application was accomplished with chloride data observed from May 2003 to December 2004 in the northern flow-way. For this, monthly-averaged net groundwater seepage terms were partially modified to generate the best HD model fit during the entire period of AD simulation. Integrated with the reca librated HD model, the AD model successfully simulates the spatio-temporal changes of chlori de concentration profile s observed in the field with an average RMSE value of 13.48 mg/L Through model calibration, 2 and 0.1 m of longitudinal and transverse dispersivities were estimated, respectively. These results show that

PAGE 116

116 the assumption of homogeneous and isotropic disper sion coefficient, which is commonly used in 2-D surface water solute transport models, shoul d be carefully made in constructed wetland systems similar to STA 5. Based on parameter estim ation, ranges of dispersion coefficients in x and y direction in this study area are suggested (Dx = 0.002 to 0.1 m2/s and Dy = 0.0001 to 0.005 m2/s) and compared to literature values. Sensit ivity test results show that the longitudinal dispersivity is a more sensitive parameter than the transverse one.

PAGE 117

117 Table 5-1. Monthly water budget for the HD model of STA 5 northern flow-way. WY Month ISF P IN OSF ET OGW OUT S R May-02 1.5 1.0 2.5 0.0 1.3 2.0 3.3 -0.9 0.1 Jun-02 8.2 2.2 10.4 5.8 0.9 1.7 8.4 1.9 0.1 Jul-02 25.2 1.3 26.6 31.6 1.0 -5.9 26.7 -0.3 0.1 Aug-02 22.2 1.0 23.1 24.4 1.0 -2.9 22.4 0.6 0.1 Sep-02 23.0 0.6 23.5 26.6 0.9 -3.1 24.3 -0.9 0.1 Oct-02 10.3 1.1 11.4 8.1 0.8 2.9 11.8 -0.5 0.1 Nov-02 5.0 0.4 5.4 2.4 0.7 1.8 4.9 0.5 0.1 Dec-02 15.7 0.7 16.4 15.0 0.6 -0.2 15.3 0.9 0.1 Jan-03 9.6 0.1 9.7 8.8 0.7 1.5 11.0 -1.4 0.1 Feb-03 3.6 0.2 3.8 1.2 0.7 1.7 3.6 0.1 0.1 Mar-03 2.2 0.5 2.8 0.6 0.9 1.6 3.0 -0.4 0.1 2003 Apr-03 1.2 0.6 1.9 0.0 1.0 1.4 2.4 -0.7 0.1 Subtotal 127.6 9.8 137.4 124.4 10.6 2.3 137.3 -1.1 1.2 May-03 4.7 1.3 6.0 0.9 1.0 1.8 3.8 2.1 0.1 Jun-03 20.2 1.3 21.5 20.5 1.0 -0.4 21.1 0.4 0.1 Jul-03 20.5 0.8 21.3 18.9 1.1 1.3 21.3 -0.1 0.1 Aug-03 27.5 1.4 28.9 29.2 0.9 -0.6 29.5 -0.8 0.1 Sep-03 24.2 2.5 26.8 25.6 0.8 -0.2 26.2 0.5 0.1 Oct-03 15.0 0.0 15.0 15.1 0.9 0.1 16.0 -1.2 0.1 Nov-03 3.9 0.3 4.2 1.5 0.7 1.9 4.1 0.0 0.1 Dec-03 3.0 0.3 3.4 0.7 0.6 1.7 3.1 0.2 0.1 Jan-04 3.2 0.5 3.7 1.1 0.7 1.4 3.2 0.4 0.1 Feb-04 7.8 0.7 8.6 5.9 0.7 1.5 8.1 0.4 0.1 Mar-04 3.6 0.0 3.6 2.3 1.0 1.6 4.9 -1.4 0.1 2004 Apr-04 1.9 0.3 2.2 0.0 1.1 0.5 1.6 0.5 0.1 Subtotal 135.6 9.4 145.0 121.6 10.6 10.6 142.8 1.0 1.2 May-04 1.7 0.2 1.9 0.0 1.3 1.1 2.4 -0.6 0.1 Jun-04 2.3 1.7 4.0 1.2 1.1 1.4 3.7 0.2 0.1 Jul-04 4.6 1.5 6.1 1.4 1.0 2.9 5.4 0.6 0.1 Aug-04 26.7 1.6 28.3 27.0 0.9 -0.6 27.3 0.9 0.1 Sep-04 30.2 1.3 31.4 35.9 0.8 -4.6 32.1 -0.8 0.1 Oct-04 20.8 0.3 21.1 19.6 0.8 0.0 20.4 0.6 0.1 Nov-04 4.8 0.1 4.9 3.5 0.6 1.5 5.6 -0.8 0.1 2005 Dec-04 1.7 0.2 1.9 0.7 0.6 1.2 2.5 -0.7 0.1 Subtotal 92.8 6.8 99.6 89.4 7.1 2.8 99.4 -0.6 0.8 Total 356.0 26.1 382.1 335.5 28.3 15.8 379.5 -0.6 3.2 % IN % OUT 93.2 6.8 100.088.47.44.2100.0 Unit: hm3 (= 1,000,000 m3)

PAGE 118

118 Note: WY = water year; ISF = surface water inflow; P = precipitation; OSF = surface water outflow; ET = evapotranspiration; OGW = net groundwater seepage; S = change in storage volume; R = water budget residual (= INOUTS). Table 5-2. Vegetation type of STA 5. Type of vegetation habitat Ratio (%) Dominant vegetation species Emergent 46.18Cattail (57%), Mixed cattail and mixed graminoids (30%) Floating 0.60Floating/floating attached emergents Shrub 7.47Primrose Willow (60%), mixed cattail and primrose willow (18%) Open water with or without vegetation 7.71 Open water with Hydrilla 38.05 Hydrilla (81%), Hydrilla with periphyton (19%) Source: STA 5 vegetation map as a format of GIS shapefile (Nick Miller, Inc., 2006)

PAGE 119

119 Table 5-3. Monthly water budget recalibrated for the AD model of STA 5 northern flow-way. WY Month ISF P INOSF ET OGW OUT S R May-03 4.7 1.3 6.0 0.91.02.24.2 2.1-0.3 Jun-03 20.2 1.3 21.5 20.51.00.321.8 0.4-0.6 Jul-03 20.5 0.8 21.3 18.91.11.121.1 -0.10.3 Aug-03 27.5 1.4 28.9 29.20.90.330.4 -0.8-0.8 Sep-03 24.2 2.5 26.8 25.60.8-1.624.8 0.51.5 Oct-03 15.0 0.0 15.0 15.10.90.016.0 -1.20.2 Nov-03 3.9 0.3 4.2 1.50.72.34.4 0.0-0.3 Dec-03 3.0 0.3 3.4 0.70.61.83.1 0.20.0 Jan-04 3.2 0.5 3.7 1.10.71.63.4 0.4-0.1 Feb-04 7.8 0.7 8.6 5.90.71.78.3 0.4-0.1 Mar-04 3.6 0.0 3.6 2.31.01.75.0 -1.40.0 2004 Apr-04 1.9 0.3 2.2 0.01.10.41.5 0.50.2 Subtotal 135.6 9.4 145.0 121.610.611.8144.0 1.00.0 May-04 1.7 0.2 1.9 0.01.31.32.6 -0.6-0.1 Jun-04 2.3 1.7 4.0 1.21.11.63.9 0.2-0.1 Jul-04 4.6 1.5 6.1 1.41.03.15.6 0.6-0.1 Aug-04 26.7 1.6 28.3 27.00.9-0.427.5 0.9-0.1 Sep-04 30.2 1.3 31.4 35.90.8-4.432.3 -0.8-0.1 Oct-04 20.8 0.3 21.1 19.60.8-0.120.3 0.60.2 Nov-04 4.8 0.1 4.9 3.50.61.65.7 -0.80.0 2005 Dec-04 1.7 0.2 1.9 0.70.61.42.7 -0.7-0.1 Subtotal 92.8 6.8 99.6 89.47.14.0100.6 -0.6-0.4 Total 228.4 16.2 244.6 211.0 17.7 15.8 244.6 0.5 -0.4 % IN % OUT 93.4 6.6 100.0 86.3 7.2 6.5 100.0 Unit: hm3 (= 1,000,000 m3)

PAGE 120

120 Table 5-4. Manning roughness coefficients used in the HD model, which are assigned with respect to type and density of vegetati on habitat in STA 5 northern flow-way. Type of vegetation habitat Percent coverage (%) Mannings n (s/m1/3) Emergent <0.1 0.1-0.25 0.5-6.6 0.085 0.130 0.670 Floating 0.003-0.006 0.006-0.02 0.02-0.031 0.070 0.100 0.120 Shrub <0.05 0.05-0.15 0.15-0.25 1.75-2.6 0.070 0.100 0.120 1.000 Open water with or without vegetation <0.025 0.025-0.05 0.05-0.1 0.1-0.21 0.21-0.34 0.060 0.076 0.085 0.100 0.120 Open water with Hydrilla <0.5 0.5-1.3 1.3-2.4 4.0-26.5 0.070 0.120 0.130 0.150 Hydraulic structure (Culvert) 0.024 Table 5-5. Coefficients for eff ective roughness used in the SFWMM. Coefficients of effective roughness, N* N value as a function of ponding depth (ft) Land use/description A b 1ft 2ft 3ft Wetland/freshwater marsh 1.10 -0.77 1.10 0.65 0.47 Wetland/sawgrass plains 1.25 -0.77 1.25 0.73 0.54 Wetland/wet prairie 0.75 -0.77 0.75 0.44 0.32 Rangeland/shrubland (scrub and shrub) 1.05 -0.77 1.05 0.62 0.45 Wetland/STA and above-ground reservoir 1.15 -0.77 1.15 0.67 0.49 Wetland/cattail 1.20 -0.77 1.20 0.70 0.51 Wetland/mixed cattail/sawgrass 1.25 -0.77 1.25 0.73 0.54 Source: Documentation for the SFWMM v5.5 (SFWMD, 2005b). N = A (PONDb), where POND is ponding depth of a grid cell.

PAGE 121

121 Table 5-6. Water level sensitivity on hydrau lic resistance represen ted by Mannings roughness coefficient in STA 5 northern flow-way. RMSE (RMSEs**RMSEc*) Water level measurement Points 30% increase of Mannings roughness coefficients 30% decrease of Mannings roughness coefficients G343B_H -0.0010.005 G343B_T 0.0040.005 G343C_H -0.0060.005 G343C_T 0.0180.006 G344A 0.0050.011 G344B -0.0030.019 Average 0.0030.009 Unit: meter RMSEc is a RMSE between the measured and simulated water level, which is based on hydraulic roughness determined by model calibration process. ** RMSEs is a RMSE between the measured and si mulated water level, which is based on hydraulic roughness modified (%) for sensitivity analysis.

PAGE 122

122 Table 5-7. Dispersion coefficient in constructe d wetlands reported in previous studies. Source Value Unit Comment Kadlec (1994a) 0.0169 0.00976m2/s Des Plaines River Martinez and Wise (2003) 0.157 (0. 010-0.512) OEW(Cell 1-15), Florida Keefe et al. (2004) 0.00253 0.000775 0.00997 0.000235 0.0213 0.000401 Tres Rios Wetlands, Phoenix, Arizona Table 5-8. Sensitivity analysis on longitudinal dispersivity ( y is fixed to 0.1m) in STA 5 northern flow-way. RMSE of chloride simulation (mg/L) x : y x (m) G343B G343C G344A G344B Average 1:1 0.1 10.41 16.16 22.21 22.43 17.802 5:1 0.5 9.45 17.22 21.45 21.22 17.334 10:1 1 9.49 13.29 20.80 20.66 16.060 20:1 2 9.83 13.21 20.08 20.15 15.819 100:1 10 11.46 16.20 19.47 19.98 16.775 Table 5-9. Sensitivity analysis on transverse dispersivity ( x is fixed to 2m) in STA 5 northern flow-way. RMSE of chloride simulation (mg/L) x : y y (m) G343B G343C G344A G344B Average 1:1 2 11.39 12.60 21.06 20.49 16.387 5:1 0.4 10.21 12.87 20.55 20.03 15.915 10:1 0.2 9.91 13.04 20.25 20.08 15.823 20:1 0.1 9.83 13.21 20.08 20.15 15.819 100:1 0.02 9.91 13.42 19.99 20.25 15.892

PAGE 123

123 Figure 5-1. Location map of study area, STA 5.

PAGE 124

124 Figure 5-2. A schematic of STA 5 indicating the hydraulic structur es as well as the flow and vegetation pattern.

PAGE 125

125 Precipitation0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0 5/17/3010/281/264/267/2510/231/214/207/1910/171/15Time (May 1, 2002 to December 31, 2004)Precipitation (mm/day ) A) Evapotranspiration0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 5/17/3010/281/264/267/2510/231/214/207/1910/171/15Time (May 1, 2002 to December 31, 2004)Evapotranspiration (mm/da y B) Figure 5-3. Daily-based precipitatio n and evapotranspiration rates in STA 5. A) Precipitation. B) Evapotranspiration.

PAGE 126

126 Surface water inflows in STA 5 northern flow-way0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 5/17/3010/281/264/267/2510/231/214/207/1910/171/15Time (May 1, 2002 to December 31, 2004)Daily flow (m3/s) G342A G342B G349AA) Surface water outflows in STA 5 northern flow-way0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 5/17/3010/281/264/267/2510/231/214/207/1910/171/15Time (May 1, 2002 to December 31, 2004)Daily flow (m3/s) G344A G344BB) Figure 5-4. Daily-based surface water flows in STA 5 northern flow-way. A) Inflows. B) Outflows.

PAGE 127

127 Figure 5-5. Bathymetry prediction map of ST A 5 northern flow-way generated using IDW interpolation scheme.

PAGE 128

128 Figure 5-6. Raster-typed ba thymetry prediction map of STA 5 northern flow-way.

PAGE 129

129 Figure 5-7. Vegetation distributi on shapefile of STA 5 northern fl ow-way. [Source: Nick Miller, Inc. (2006).]

PAGE 130

130 1.00 1.00 0.10 0.10 0.10 0.08 1.00 0.67 0.67 0.67 0.08 0.67 0.67 0.67 0.67 0.07 0.08 0.67 0.67 0.67 0.12 0.12 1.00 0.10 0.07 0.07 0.10 1.00 0.07 0.67 0.67 0.07 0.07 0.67 0.67 0.67 0.67 0.67 0.67 0.67 0.67 0.67 0.06 0.12 1.00 0.12 0.08 1.00 0.08 0.08 0.67 0.67 0.07 0.07 0.07 0.07 0.67 0.67 0.07 0.07 0.07 0.67 0.67 0.06 0.12 0.20 0.10 1.00 0.07 1.00 0.07 1.00 0.07 0.07 0.67 0.67 0.67 0.67 0.67 0.67 0.06 0.67 0.67 0.67 0.67 0.12 0.12 0.12 0.10 1.00 1.00 1.00 0.07 1.00 0.06 0.67 0.67 0.07 0.67 0.08 0.67 0.67 0.67 0.67 0.67 0.12 0.08 0.12 0.12 0.12 1.00 1.00 1.00 0.12 0.12 0.08 0.67 0.67 0.67 0.67 0.06 0.67 0.67 0.08 0.67 0.67 0.08 0.67 0.67 0.12 0.12 0.12 1.00 0.07 1.00 0.12 0.12 0.08 0.67 0.67 0.67 0.67 0.67 0.67 0.67 0.67 0.67 0.67 0.67 0.12 0.12 0.12 0.12 0.20 1.00 1.00 0.07 0.08 0.12 0.67 0.67 0.10 0.67 0.67 0.67 0.07 0.67 0.67 0.07 0.67 0.13 0.13 0.13 0.13 0.13 0.07 1.00 1.00 1.00 0.08 0.08 0.67 0.10 0.10 0.67 0.67 0.67 0.67 0.67 0.07 0.07 0.13 0.13 0.13 0.13 0.13 0.13 0.07 1.00 1.00 1.00 0.10 0.08 0.67 0.67 0.67 0.67 0.67 0.67 0.67 0.08 0.08 0.13 0.13 0.13 0.13 0.13 0.13 0.07 0.20 1.00 1.00 1.00 0.13 0.13 0.13 0.67 0.67 0.67 0.67 0.67 0.08 0.12 0.12 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 1.00 1.00 1.00 0.13 0.67 0.67 0.67 0.08 0.67 0.67 0.12 0.12 0.12 0.08 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.07 1.00 1.00 0.12 0.12 0.12 0.10 0.67 0.67 0.07 0.12 0.12 0.08 0.12 0.12 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.20 1.00 1.00 1.00 0.12 0.12 0.08 0.07 0.12 0.12 0.12 0.12 0.08 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.08 0.07 0.07 1.00 1.00 0.08 1.00 1.00 0.08 0.10 0.12 0.12 0.12 0.12 0.08 0.13 0.12 0.13 0.13 0.07 0.08 0.08 0.07 0.08 0.07 A) Emergent Floating Shrub Open water with or without vegetation Open water with Hydrilla Figure 5-8. Mannings roughness coeffi cients assigned at each model grid cell of STA 5 northern flow-way. A) Cell 1A. B) Cell 1 B.

PAGE 131

131 0.07 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.07 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.06 0.15 0.07 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.12 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.20 0.12 0.12 0.12 0.08 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.07 0.12 0.12 0.15 0.15 0.15 0.15 0.06 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.12 0.12 0.12 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.07 0.07 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.08 0.15 0.15 0.15 0.08 0.15 0.15 0.15 0.15 0.15 0.15 0.07 0.12 0.12 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.06 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.07 0.12 0.12 0.12 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.12 0.12 0.12 0.15 0.15 0.15 0.15 0.15 0.06 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.08 0.15 0.09 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.08 0.12 0.12 0.15 0.15 0.15 0.15 0.06 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.08 0.15 0.15 0.15 0.15 0.15 0.09 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.20 0.15 0.06 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.06 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.06 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.06 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.02 0.15 0.15 0.02 0.15 0.15 B) Figure 5-8. Continued.

PAGE 132

132 Inlet chloride concentration profiles in STA 5 northern flow-way0 20 40 60 80 100 120 140 160 180 200 220 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Chloride concentration (mg/ L G342A G342B G349A Figure 5-9. Inlet chloride concentrati on profiles in STA 5 northern flow-way.

PAGE 133

133 G343B_H Water Level (RMSE = 0.083 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured SimulatedA) G343B_T Water Level (RMSE = 0.068 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured SimulatedB) Figure 5-10. Simulation results of the HD model on hydroperiod fluctuation at the six water level measurement points in STA 5 northern flow -way during the mode l calibration period (May 1, 2002 to April 30, 2003). A) G 343B_H. B) G343B_T. C) G343C_H. D) G343C_T. E) G344A_H. F) G344B_H.

PAGE 134

134 G343C_H Water Level (RMSE = 0.086 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured SimulatedC) G343C_T Water Level (RMSE = 0.078 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured SimulatedD) Figure 5-10. Continued.

PAGE 135

135 G344A Water Level (RMSE = 0.109 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured SimulatedE) G344B Water Level (RMSE = 0.128 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured SimulatedF) Figure 5-10. Continued.

PAGE 136

136 G343B_H Water Level (RMSE = 0.079 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 4/276/268/2510/2412/232/214/216/208/1910/1812/17Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured SimulatedA) G343B_T Water Level (RMSE = 0.112 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 4/276/268/2510/2412/232/214/216/208/1910/1812/17Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured SimulatedB) Figure 5-11. Simulation results of the HD model on hydroperiod fluctuation at the six water level measurement points in STA 5 northern flow -way during the mode l validation period (May 1, 2003 to December 31, 2004). A) G343B_H. B) G343B_T. C) G343C_H. D) G343C_T. E) G344A_H. F) G344B_H.

PAGE 137

137 G343C_H Water Level (RMSE = 0.087 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 4/276/268/2510/2412/232/214/216/208/1910/1812/17Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured SimulatedC) G343C_T Water Level (RMSE = 0.124 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 4/276/268/2510/2412/232/214/216/208/1910/1812/17Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured SimulatedD) Figure 5-11. Continued.

PAGE 138

138 G344A Water Level (RMSE = 0.241 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 4/276/268/2510/2412/232/214/216/208/1910/1812/17Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured SimulatedE) G344B Water Level (RMSE = 0.273 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 4/276/268/2510/2412/232/214/216/208/1910/1812/17Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured SimulatedF) Figure 5-11. Continued.

PAGE 139

139 G343B_H Water Level4.00 4.10 4.20 4.30 4.40 4.50 4.60 4.70 4.80 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured Calibrated n Increased n (+30%) Decreased n (-30%)A) G343B_T Water Level3.60 3.70 3.80 3.90 4.00 4.10 4.20 4.30 4.40 4.50 4.60 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured Calibrated n Increased n (+30%) Decreased n (-30%)B) Figure 5-12. Sensitivity analysis results of Mannings roughness coefficient ( n ) on hydroperiod fluctuation at the six water level measurem ent points in STA 5 northern flow-way. A) G343B_H. B) G343B_T. C) G343C_H. D) G343C_T. E) G344A_H. F) G344B_H.

PAGE 140

140 G343C_H Water Level4.00 4.10 4.20 4.30 4.40 4.50 4.60 4.70 4.80 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured Calibrated n Increased n (+30%) Decreased n (-30%)C) G343C_T Water Level3.60 3.70 3.80 3.90 4.00 4.10 4.20 4.30 4.40 4.50 4.60 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured Calibrated n Increased n (+30%) Decreased n (-30%)D) Figure 5-12. Continued.

PAGE 141

141 G344A Water Level3.40 3.50 3.60 3.70 3.80 3.90 4.00 4.10 4.20 4.30 4.40 4.50 4.60 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured Calibrated n Increased n (+30%) Decreased n (-30%)E) G344B Water Level3.40 3.50 3.60 3.70 3.80 3.90 4.00 4.10 4.20 4.30 4.40 4.50 4.60 5/16/17/28/29/210/311/312/41/42/43/74/75/8Time (May 1, 2002 to April 30, 2003)Water Level (m) Measured Calibrated n Increased n (+30%) Decreased n (-30%)F) Figure 5-12. Continued.

PAGE 142

142 G343B Chloride Concentration Profile (RMSE = 7.80 mg/L)0 20 40 60 80 100 120 140 160 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured SimulatedA) G343C Chloride Concentration Profile (RMSE = 11.44 mg/L)0 20 40 60 80 100 120 140 160 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured SimulatedB) Figure 5-13. Simulation results of the AD model on chloride concentration profiles at the four water quality measurement poi nts in STA 5 northern fl ow-way during the model calibration period (May 1, 2003 to April 30, 2004). A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 143

143 G344A Chloride Concentration Profile (RMSE = 16.83 mg/L)0 20 40 60 80 100 120 140 160 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured SimulatedC) G344B Chloride Concentration Profile (RMSE = 17.83 mg/L)0 20 40 60 80 100 120 140 160 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured SimulatedD) Figure 5-13. Continued.

PAGE 144

144 G343B Chloride Concentration Profile (RMSE = 12.24 mg/L)0 20 40 60 80 100 120 140 160 180 200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured SimulatedA) G343C Chloride Concentration Profile (RMSE = 15.46 mg/L)0 20 40 60 80 100 120 140 160 180 200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured SimulatedB) Figure 5-14. Simulation results of the AD model on chloride concentration profiles at the four water quality measurement poi nts in STA 5 northern fl ow-way during the model validation period (May 1, 2004 to December 31, 2004). A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 145

145 G344A Chloride Concentration Profile (RMSE = 24.09 mg/L)0 20 40 60 80 100 120 140 160 180 200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured SimulatedC) G344B Chloride Concentration Profile (RMSE = 23.16 mg/L)0 20 40 60 80 100 120 140 160 180 200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured SimulatedD) Figure 5-14. Continued.

PAGE 146

146 15.50 16.00 16.50 17.00 17.50 18.00 18.50 19.00 19.50 20.00 20.50 0.1110 x (m)Average RMSE (mg/L) 1:1 2:1 5:1 10:1 20:1 50:1 100:1 Figure 5-15. Calibration curves of chloride transport model on longitudinal and transverse dispersivity.

PAGE 147

147 CHAPTER 6 MODEL APPLICATION TO THE STORMW ATER TREATMENT AREA (STA) 5 NORTHERN FLOW-WAY: II. PHOSPHORUS DYNAMICS Introduction The primary purpose of STAs in South Florid a is to remove phosphorus flowing into the Everglades. The original design and construc tion of STAs were based on a first-order phosphorus model, which was calibra ted to water column and peat phosphorus data collected in WCA-2A (Walker, 1995). In spite of usefulness as a long-term management model, application is basically limited as a scientific modeling t ool to understand the function and structure of constructed wetlands systematically. In additi on, practical application as a management modeling tool is not complete b ecause the shortor long-term sp atio-temporal variations caused by testing various hypotheses for best management are not provided by a first-order phosphorus uptake model. In this chapter, the ECO Lab phosphorus dynamics model developed in Chapter 3 is calibrated and validated with wa ter column phosphorus data colle cted in STA 5 northern flowway. Linked with the HD and AD models calib rated in the same area (Chapter 5), the phosphorus dynamics model was used to simulate the observed spatio-temporal variations of three phosphorus species (SRPw, DOP, and PP). This modeling approach is expected as an alternative to historical treatment wetland mode ling efforts, overcoming their shortcomings. For this model application, several reasons to choose the northern flow-way of STA 5 are: (1) not all STAs have abundant tempo-spatial flow and water quality measurement data, including phosphorus species in water column, to enable calibration and validation of a long term 2-D flow and phosphorus dynamics model, (2) South Florida STAs, including the ENRP, and the adjacent wetland ecosystems such as WCA-2A, comprise one of the most systematicallystudied wetland ecosystems in the world. So a lo t of background field an d laboratory data on soil

PAGE 148

148 and vegetation, required to develop an ecosystem dynamics model, are av ailable, (3) the basin shape and management practice (EAV-SAV system ) of treatment cells (Cell 1A and 1B) are typical enough to be app licable broadly to other similar cons tructed wetlands, and (4) until now, STA 5 is the most challenging system with resp ect to phosphorus removal efficiency among the STAs (Juston and DeBusk, 2006), mainly due to overloading of flow and phosphorous exceeding hydraulic (2.62 cm/day) and phosphorus loading rates (1.69 g/m2/year) of long-term average design (Pietro et al., 2006). As a result, inpu t and output phosphorus levels are relatively high compared to the other STA systems. On the ot her hand, these high phosphorus concentrations enable one to avoid great simula tion uncertainty due to the extr emely low measured level. In reality, effluent levels of some phosphorus spec ies in well-performing STA systems are not far from the detection limit (4 ppb in this study). Specific Aims Recent STA studies have focused on technology development to optimize performance in order to lower outflow phosphorus concentrations to very low levels (Goforth, 2001; Newman and Lynch, 2001; Dierberg et al., 2002). For th is, treatment efficiency has been commonly compared through manipulation of vegetative co mmunities or hydraulic efficiency. Biota, such as SAV and periphyton, has been spotlighted r ecently instead of traditional EAV systems. According to Dierberg et al. (2002), the native SAV systems ar e more efficient for phosphorus removal at low concentration (14-100 ppb) than the EAV systems. A mesocosm study performed by DeBusk et al. (2004) showed that a periphyton system can generate very low outflow phosphorus concentrations (approximately 10 ppb) On the other hand, negative impact of internal short-circuiting flow on phosphorus re moval performance in STA 1W was reported (Dierberg et al., 2005). System atical phosphorus dynamics mode ling efforts on these wetland ecosystems have been rarely reported.

PAGE 149

149 Therefore, the specific aims of this chapte r are: (1) to develop an ECO Lab phosphorus dynamics model, which is integrated to depthaveraged flow dynamics and solute transport model and incorporates differences in the phos phorus cycle between EAV and SAV systems. This model may be primarily applied to largescaled, subtropical constructed wetland systems similar to South Florida STAs; however, it is ex pected to be applied to various constructed wetland settings because it is eas y to adjust ECO Lab model a ccording to the site-specific characteristics of an ecosystem and (2) to estim ate sensitivities to values (or ranges) of key model parameters commonly used, but not extensiv ely studied in STA-type constructed wetlands. Materials and Methods The ECO Lab phosphorus dynamics model, calib rated and validated with time series concentration profiles of water column phosphorus species observed in the STA 5 northern flowway, incorporates ecosystem dynamics among wa ter column, floc and upper soil layers, and plant communities such as EAV, SAV, and periphyt on (Chapter 3). As a resu lt, a lot of field or laboratory measurement data are basically required to build the ecosystem dynamics. In general, success of a water quality model entirely depe nds on the quality and quantity of collected background data. Data acquisiti on through field survey and measurement, field or lab experiment, and historical literatu re review is one of the most important tasks in developing the model. Although the study area ha s relatively abundant data fo r phosphorus compartments, there are still some unknowns that are not fully supporte d by literature survey. In addition, as the available data are, in many cases, dependent on one-time measurement, they provide no information on their spatio-temporal variations and some assumptions and inferences are inevitable in this study. The linked HD and AD model setups are not revi ewed here again because they were fully covered in the previous chapter. Therefore, th e focus of this section is placed on the ECO Lab

PAGE 150

150 model setup: (1) phosphorus data and assumptions used in the model, (2) initial conditions, (3) model constants used in the model, and (4) model incorporation of differences between phosphorus dynamics of EAV and SAV systems. ECO Lab Model Setup The ECO Lab phosphorus dynamics model consists of 12 state variables, 34 processes, 56 constants, and 3 forcing functions. The Euler met hod was selected to solve the coupled ordinary differential equations defined in Chapter 3 numer ically. Although it is th e simplest numerical integration method, it is favorable on saving the model running time. The numerical solution error incurred by not choosing more sophisticated integration met hods is negligible when the simulation time step is very short. In this st udy, the time simulation time step is 10 minutes and the simulation period for model calibration is from May 1, 2003 to April 30, 2004 (1 year) and for validation is from May 1, 2004 to December 31, 2004 (0.67 years). The model scope addressing specific model f eatures distinguishabl e with historical phosphorus dynamics modeling efforts as well as th e model framework was discussed in Chapter 3. Phosphorus data used in model Like the hydrologic data in Chapter 5, DB HYDRO data for three phosphorus species in water column (SRP, Total Disso lved Reactive Phosphorus (TDRP) and TP) are available during the entire simulation period at hydraulic struct ures through the middle levee separating Cell 1A and 1B (G343B and G343C) as well as at a ll inflow (G342A, G342B, and G349A) and outflow control structures of the nor thern flow-way (G344A and G344B). These phosphorus species were collected as weekly or biweekly grab samples. To generate daily-based phosphorus concentration profiles at each site, weekly or biweekly measured phosphorus data were linearly interpolated in this study. Daily-based DOP a nd PP were calculated as DOP = TDRP SRP and

PAGE 151

151 as PP = TP TDRP, respectively. Phosphorus c oncentration profiles observed at the inflow structures were incorporated into each source grid cell of the HD model domain to generate phosphorus plumes in the study area, and those ob served at the middle and outflow structures were used to calibrate and valid ate the phosphorus dynamics model. Figures 6-1, 6-2, and 6-3 illustrate the in let concentration profiles of observed three phosphorus species (SRPw, DOP, and PP) and the relative po rtion at G342A, G342B, and G349A, respectively. Temporal variation of SRPw concentration follow typical stormwater pattern at the two culvert-induced inflow s ites (G342A and G342B); on the other hand, DOP and PP do not show the stormwater-typed inflow pa ttern (Figure 6-1A and 6-2A). SRPw is the dominant species of inlet phosphorus except during the dry period from April to July; on the other hand, PP is dominant during the four mont hs corresponding to the vegetation growing season (Figure 6-1B and 6-2B). The inlet pattern of phosphorus and the relative portion in seepage return flow (G349A) are different from those of the culver t-induced inflows. Th ere is no relationship between phosphorus levels and typical stormwat er runoff, and PP is predominant year-round at the seepage return flow site (Figure 6-3). Phosphorus accretion to the wetland bottom is the long-term storage mechanism in constructed wetlands. Top soil (0-10 cm) and floc layer samples from STA 5 were collected and analyzed by SFWMD in summer 2003. Refer to Qore Inc. (2004) for the field and laboratory methods. Maps showing spatial variation in bulk density and TP content of the floc and the upper soil layer were provided by Goforth et al. (2005) and were converte d into raster-typed data for model input in this study. The average converted floc and upper soil TP contents for STA 5 northern flow-way are 791 mg/ kg and 492 mg/kg, respectively.

PAGE 152

152 Unfortunately, although PIP and POP were not i ndependently monitored in the field, the dynamics model incorporated to separate the obser ved PP into the two stat e variables because the role of each state variable in phosphorus cycle is totally different. Likewi se, measured floc and soil TP concentration data should be divided into IPf and OPf and IPs and OPs, respectively. In this study, ratios between these inorganic and or ganic phosphorus state variables were assumed to be 0.3:0.7, which originated from the aver age ratio between inorgani c and organic phosphorus concentrations in floc and soil layers of ST A 1W and WCA 1A. Pant and Reddy (2001) reported that the ratio between IP and OP in floc layer of treatment Cell 1 in STA 1W was 0.3:0.7. Also, Corstanje et al. (2006) reported a ratio of 0.25:0.75 in the floc layer of WCA 1A. For soil layer, White et al. (2006) reported a ratio of 0.31:0.69 in top 5 cm soil layer of STA 1W EAV and SAV cells (0.38:0.62 in the entire soil layer), and Cors tanje et al. (2006) reporte d a ratio of 0.2:0.8 in soil layer of WCA 1A. Previously, Reddy et al. (1998b) reported a ratio of 0.33:0.67 in a study regarding forms of soil phosphorus in selected hydrologic units of the Florida Everglades. Although the assumed ratio between inorganic a nd organic floc/soil phosphorus data in these wetland systems that are similar to STA 5 cannot be directly applicable to determine the ratio between water column PIP and POP, this assump tion is not too unreasonabl e in that deposition of water column PP eventually regulates the ratio of floc and soil layer. Initial conditions Phosphorous data used to specify initial condi tions of 12 state variables in the dynamics model are primarily dependent on field measurement in STA 5 northern flow-way or the adjacent wetland systems, such as STA 1W and WCA-2A. For mobile state variables in the water column (SRPw, DOP, PIP, and POP), phosphorus concentrations observed at seven measuremen t points within the nor thern flow-way on the

PAGE 153

153 simulation starting date (May 1, 2003) were linear ly interpolated to se t each initial condition on the entire model domain. Prior to specifying the initi al condition of model doma in for periphyton phosphorus (Pperi), periphyton was assumed to homoge neously exist on only SAV grid cells in Cell 1B due to the lack of field data. Accordi ng to McCormick et al. (1998) areal periphyton phosphorus concentration, mostly stored as a form of epiphyton, was 0.015 g TP/m2 in eutrophic open water areas of WCA 2A during the dry season. This va lue was divided by average water depth of Cell 1B (0.7 m), then a constant va lue of 0.021 mg/L was assigned to all SAV grid cells in Cell 1B. To specify the initial co ndition of the model domain for macrophyte phosphorus (Pmacro), five vegetation types of the northern flow-way (T able 5-2) were first reclassified into three categories: EAV, SAV, and no vegetation. The EAV category includes three vegetation groups: cattail dominant emergent vegetations, shr ub, and floating. SAV and no vegetation categories correspond to the gr oups of open water with Hydrilla and open water with or without vegetation, respectively. Pmacro was set at zero for grid cells corresponding to the no vegetation category. Total macrophyte phosphorus concentrations (including below-ground biomass) in EAV and SAV mesocosms of STA 1W were reported as 0.348 g P/m2 and 0.121 g P/m2, respectively (White et al., 2006). These valu es were divided by average water depths of Cell 1A (0.6 m) and 1B (0.7 m) respectively, and then constant values of 0.58 mg/L and 0.173 mg/L were evenly assigned to all the grid cells corresponding to th e EAV and SAV category, respectively. Spatially-constant SRPf and SRPs concentration were initiall y set at 0.3 mg/L and 0.75 mg/L, respectively, considering the vertical profiles of floc and soil porewater SRP concentrations in STA 1W and the impacted areas of WCA 2A (Koch, 1991; Reddy et al., 1999;

PAGE 154

154 DeBusk and Reddy, 2005). Each raster-typed TP da ta, converted from the TP content maps of the floc and the upper soil laye r (Goforth et al., 2005), was separated using the 0.3:0.7 assumption for the initial condition of IPf, OPf, IPs, and OPs (Figure 6-4). Constants used in model Model constants used in the phosphorus dynamics model were first obtained from field measurements, previous literature data, and inference, then tuned to fit the field data in STA 5 northern flow-way. The finally calibrated valu es used for the phosphorus dynamics model are shown in Table 6-1. In this study, model ca libration was accomplished first by changing the parameters related to sedimentation/resuspension processes and then to th ose related to uptake by biota. Then, model constants asso ciated with the floc and soil la yers were changed to generate near steady state concentration prof iles of the six state variables. Finally, model constants related to phosphorus transformation processes among th e state variables in water column were modified to obtain the least RMSE valu es between measured and simulated SRPw, DOP, and PP concentration profiles at the four monitori ng sites (G343B, G343C, G344A, and G344B). For atmospheric deposition flux of the mobile state variables (SRPw, DOP, PIP, and POP), bulk deposition rates estimated in Chapter 3 were used (Table 3-5). The selected decay and mineralization/degradation rate constants were mostly within the range of literature values (Tables 3-6 and 3-7), and formation rate cons tants were determined by model calibration. Calibrated critical velocity a nd phosphorus sedimentation/resusp ension rate constants are also not highly different with the lite rature values (Table 3-8). The calibrated maximum growth rate cons tants for phosphorus uptake by EAV (mainly Typha spp. ), SAV, and periphyton are sim ilar to the literature values (Table 3-9). In general, sensitivity of macrophytes/periphyton growth to SRP concentration in water column or soil layer is represented by the ratio of SRP to the half saturation constants ( ks_macro ks_root and

PAGE 155

155 ks_peri ). In cases which SRP concentrations are mu ch lower than half saturation constants, growth rate of biota increases almost linearly with increased SRP concentration. On the other hand, if SRP >> the half saturati on constants, the growth rate of biota is almost constant regardless of SRP concentration. In this study, ca librated values of three half saturation constants are greater than average SRP con centrations in water column a nd soil layer. Hence, phosphorus is not saturated with respect to the biota growth, serving as a limiti ng factor for the growth in the ecosystem. An effective diffusion coefficient ( D ) value of 3.41-5 m2/day was used considering a tortuosity factor of 0.5 (Li and Gregory, 1974; Reddy et al., 1996; Fisher and Reddy, 2001; Newman and Pietro, 2001). An average floc laye r depth of 0.088 m was measured in the field (Pietro et al., 2006), and root fractions of the EAVs and the SAVs were set at 0.667 (Davis, 1984; Han, 1985) and 0.05 (Janse, 1998; White et al ., 2006), respectively. Bulk density maps for the floc and the top soil layer (Gof orth et al., 2005) were converted into raster-typed datasets and assigned to the model domain. Porosity of the floc ( poro_f ) and the soil layer ( poro_s ) were set at 0.95 (0.97 in Newman and Pietro (2001); 0.94 in Pant and Reddy (2001); 0.92 in Corstanje et al. (2006)) and 0.74 (White et al., 2006), respectiv ely. Maximum adsorption capacity in the floc ( Smax_f ) and the soil layer ( Smax_s ) were assumed to be 500 mg/kg and 250 mg/kg (98-418 mg/kg in Reddy et al. (1995)), respectivel y. Biodegradable fractions of OPf ( f_excop_f ) and OPs ( f_excop_s ) and exchangeable fractions of IPf ( f_excip_f ) and IPs ( f_excip_s ) were estimated to be 0.25 and 0.1 (7% of TP in Grunwald et al (2006)) and 0.5 (Pant and Reddy, 2001) and 0.01 (less than 0.3% in White et al. (2006)), respectively. Sequestration fluxes of IPf into IPs ( f_seq1 ) and OPf into OPs ( f_seq2 ) were determined by model calibrati on, and sequestration fluxes of IPs

PAGE 156

156 and OPs into deep soil layer ( f_seq3 and f_seq4 ) were set using the values measured by Turner et al. (2006). A minimum water depth ( min_wd ) of 0.03 m was used in the model. In this study, a time-series temperature pr ofile measured at G 343C (Figure 6-5) was specified in the phosphorus dynamics model to re flect temperature dependency of biologically mediated reactions. However, model calibration showed that the temperature effect was not significant in this dynamics modeling study. In other words, model calib ration was better when the temperature effect was eliminated (tempe rature coefficient of Arrhenius equation ( ) = 1.00). Kadlec and Knight (1996) a nd Kadlec and Reddy (2001) repo rted little or no apparent temperature effect on phosphorus removal in marshes. Their findings may be due to the fact that annual temperature variation of subtropical wetla nds is small compared to systems at higher latitudes. Phosphorus dynamics: EAV vs. SAV Figures 6-6 and 6-7 show schematics of phosphorus dynamics in the EAV and SAV system of STA 5 northern flow-way, respectivel y. Differences in phosphorus dynamics models between EAV and SAV wetland ecosystems are pr esented in Table 62 in more detail. Periphyton and direct SRPw uptake by macrophyte in water column are not considered in the EAV phosphorus model. As a result, root uptake of SRPs is dominant in EAV system. On the other hand, direct uptake by m acrophyte in the water column is dominant in SAV system, and the root uptake is minimal due to the small fraction. For macrophyte decomposition processes ( k_decay4 and k_decay6 ) and adsorption of SRPw into PIP, appropriate weighting factors are given to reflect the fact that biomass decompositi on rates are generally fast er in the SAV system, and Ca-P coprecipitation via periphyton does not occur in EAV system. For example, the formation rate constant of PIP from SRPw ( k_form2 ) in SAV system is set at six times higher

PAGE 157

157 than in the EAV system. To incorporate these processes in the model, the related parameter values were determined by model calibration. Results and Discussion As water enters the northern flow-way, it first flows into a cattail-dominant EAV marsh (Cell 1A). Farther along in the treatment cell, additional phosphor us species are removed in a treatment cell (Cell 1B), which primarily consis ts of SAV plus periphy ton that attach to underwater stems, bringing the inlet concentrat ions down to lower levels. Average operating conditions of the northern flow-way for three years (WY 2003 to 2005) were reported by Juston and DeBusk (2006). Hydraulic and phosphorus lo ading rates are 3.5 cm/day and 2.3 g/m2/year, respectively. Average concen trations of input and output phosphorus are 177 and 91 ppb, respectively, and phosphorus removal rate is 46%. Ahead of running the ECO Lab phosphorus dynamics model, conservative SRPw, DOP, and PP transport simulations, based on the chloride transport model setup described in Chapter 5, were used to estimate water column phosphorus retention in STA 5 northern flow-way. This exercise was performed to figure out how different the reactive char acteristics are with respect to the phosphorus species and the we tland vegetation type. Conser vative phosphorus transport model results and implications are first presen ted, followed by the results and discussion of phosphorus dynamics model. Conservative Phosphorus Transport Model Estimation of phosphorus retention In general, the phosphorus budget in the water co lumn of most constructed wetlands is as follows: ISF + P OSF OGW Ret = PWC (6-1)

PAGE 158

158 where ISF, P, OSF, OGW, Ret and PWC denote the phosphorus mass of surface water inflow, atmospheric deposition, surface water outflow, ne t groundwater seepage, retention, and water column storage change, respectively. ISF and OSF are relatively easily monitored in the field; however, it is extremely difficult to measure the other terms directly. Hence, in many cases, P, OGW, and PWC would be often omitted in estimating phosp horus retention if they are very small compared to ISF and OSF. In addition, phosphorus removal effi ciency of a constructed wetland is computed in the same way and compared to similar systems. In this study, the annual phosphorus budget for STA 5 northern flow-way (May 1, 2003 to April 30, 2004) was estimated through comparison of measured and cons ervatively-simulated phosphorus concentration profile s at four monitoring point s (G343B, G343C, G344A, and G344B). If phosphorus was conservative in the entire study area, its mass balance could be expressed as: ISF* + P* OSF* OGW* = PWC* (6-2) where each term is the same as Equation 6-1 excep t denotation of to reflect conservative solute transport condition. To calculate ISF and OSF *, daily measured surface water inflow/outflow and linearly-interpolated, measured inlet and simulate d outlet phosphorus co ncentrations were multiplied to calculate daily-based phosphorus masses in surface water flow, which were summed with respect to the one -year simulation period. For P*, the annual average estimation shown in Table 3-5 was used. PWC was calculated by MIKE 21 Ma ss Budget module, then the only unknown, OGW *, was determined by Equation 6-2. In reality, some amount of inlet phosphorus is physically, biologically, or chemically retained in most constructed wetland systems, following Equation 6-1, not 6-2; otherwise, outlet phosphorus concentrations would be higher. ISF* and P* are the same with ISF and P because

PAGE 159

159 they are given terms. If we assume that OGW* and PWC* are approximately identical to OGW and PWC, respectively, OSF* is divided into OSF and Ret as follows: OSF* = OSF + Ret (6-3) where OSF is calculated by multiplying measured surface water outflow and the linearlyinterpolated, measured outlet phosphorus concentration. OSF is increased as much as the amount of phosphorus retention in the r eal condition, compared to OSF. Therefore, the retained phosphorus mass, Ret, was determined by Equa tion 6-3. For four phosphorus species, SRPw, DOP, PP, and TP, each annual phosphorus budget is summarized in the Table 6-3 and compared to data provided by Pietro et al. (200 6) and Chimney (2007). Except for the ISF term of PP, both show an overall agreement for budget estimation. For SRPw and PP, 45.5% and 50.3% of inlet phosphorus masses were retained in the entire nor thern flow-way, respectively; on the other hand, only 8.2% of inlet DOP mass was retained in the same area. As for TP, 43.7% of inlet TP was removed in the northern flow-way. Figures 6-8, 6-9, and 6-10 pres ent results of conservative solute transport simulation from May 1, 2003 to December 31, 2004 for SRPw, DOP, and PP concentratio n profiles at the four water quality measurement point s in the STA 5 northern flow -way (G343B, G343C, G344A, and G344B), respectively. The uncerta inty of estimating phosphorus rete ntion using the conservative transport model is totally dependent on accuracy of phosphorus concentration profiles simulated at each measurement point. Therefore, the 1 RMSE variations of the simulated SRPw, DOP, and PP concentration profiles, which are base d on RMSEs of chloride transport simulation calculated at the same measurement points, are estimated using the assumption that the coefficient of variation (CV = RMSE/Mean) in chlo ride transport simulation is the same as that in conservative phosphorus tran sport simulation as follow:

PAGE 160

160 P P CL CLMean RMSE Mean RMSE (6-4) where RMSECL and RMSEP are RMSE values of conservative so lute transport model for chloride and phosphorus, respectively, and MeanCL and MeanP are each simulation mean value. The only unknown, RMSEP, was determined from Equation 6-4. Appendix C, D, and E show the 1 RMSE variations of the simulated SRPw, DOP, and PP concentrati on profiles, respectively. Implication of this modeling study Unlike generally expected, the dominan t inlet phosphorus species is SRPw in the wet season and PP in the spring. It is not ed that retention or loss of SRPw is minimal in Cell 1A, EAV dominant treatment cell (Figure 6-8A and B). Loss of SRPw occurs mainly in spring and early summer season and after the storm season in Cell 1B, SAV dominated treatment cell (Figure 68C and D). Generally, as well-known, the SAV wetland is more efficient for phosphorus removal than EAV (Kadlec, 2006). It seems that the period of SRPw loss corresponds to the vegetation growing season. On the other hand, storm (wet) and mid-winter seasons are not likely to be good for SRPw uptake due to storm and temperature effects. For DOP, retention in Cell 1A is also neglig ible because there is almost no difference between measured and simulated DOP profiles (F igure 6-9A and B). Moreover, Cell 1B may have a possibility of acting as a source of DOP (F igure 6-9C and D). This suggests the possibility that the SAV system plays a role as a DOP s ource, but it has not been clearly understood. However, it is likely that there may be no domin ant mechanisms for removal in current STA 5 northern flow-way. Due to low retention ability, it may be difficult with the current system to meet extremely low effluent TP target levels such as 10 ppb. Results of the DOP conservative transport simulation confirm the findi ngs of White et al. (2004 and 2006).

PAGE 161

161 For PP, a relatively constant loss pattern is observed in both Cell 1A and 1B, except during the storm season (Figure 6-10). In the wet seas on, the flow-way may experience mechanical resuspension due to fast currents. Hence, it is noted that PP behavior in the water column is primarily regulated by phosphorus mass transfer m echanisms between water column and the floc layer, such as sedimentation and resuspension, rather than biomass-related biogeochemical processes. These results show that the conservative phosphorus transport model, which is based on calibrated HD and AD models, can pr ovide useful information on spatio-temporal variations of phosphorus retention with a statis tical confidence level with re spect to the species and wetland vegetation. Phosphorus Dynamics Model Phosphorus dynamics simulation The ECO Lab phosphorus dynamics model was calibrated to find the best agreement between simulated and measur ed model components, SRPw, DOP, and PP, through variation of a number of parameters used in the model. Figures 6-11, 6-12, and 6-13 illu strate results of model calibration on SRPw, DOP, and PP concentration profiles at G343B, G343C, G344A, and G344B, respectively. During the calibration period, average phosphorus con centrations between measurement and simulation are almost identical (Table 6-4). The RMSE values of model calibration are shown in Table 6-5. There are spatial variations of model fit with respect to measurement points, but they are not re peated among phosphorus species. For SRPw, DOP, and PP, average RMSE values are 0.031 mg/L, 0. 006 mg/L, and 0.015 mg/L, respectively. Annual variations of SRPw observed in Cell 1A (Figure 6-11 A and B) were better simulated than those in Cell 1B (Figure 6-11 C and D). Except for a couple months correspond ing to the vegetation growing season in the SAV system, sp atio-temporal vari ations of SRPw were well simulated. For

PAGE 162

162 DOP, the simulated profiles agreed well with annually measured concentration profiles (Figure 6-12). However, calibrated PP concentration profiles were not as good as those of SRPw and DOP because they are too flat to catch the fluc tuation of measured concentration profiles. Using a temporally different data set (May 1, 2004 to December 31, 2004) at the same measurement points, the m odel was validated for SRPw, DOP, and PP concentration profiles, respectively (Figures 6-14, 615, and 6-16). During the vali dation period, average phosphorus concentrations, both measured and simula ted, are presented in Table 6-6. SRPw was slightly overestimated on average; PP was underestimated and similar simulation patterns with model calibration were observed during model validation. The RMSE values of model validation are shown in Table 6-7, which are not very different with those of the model calibration (Table 6-5). In this study, deviations between measured and simulated state variables may be caused by the following reasons: (1) HD and AD model errors (2) inaccurate initial concentrations, (3) limitations of the transformation processes de scribed in ECO Lab model, (4) inaccurate calibration of constants/parameters used in the model, and (5) measurement errors. The errors of ECO Lab phosphorus dynamics simulation intrinsically contain HD and AD model errors. The average RMSE of the HD simulation (about 0.09 m) may propagate into concentration errors, due to water volume discrepa ncies. In particular, the simulation error would be greater either during the dry season or at grid cells near high bathymetry elevation areas due to shallow water depth. Interpolation or assumptions made to specify initial conditions in this study may have caused simulation errors. Phosphorus contents of EAV and SAV cells used in the model may be invalid because the data were ba sed on mesocosm scale study, not a field scale. In addition, the

PAGE 163

163 assumption on homogeneous distribution of Pperi, Pmacro, SRPf, and SRPs without considering the spatial variation may contribu te to generate error. The lack of scientific un derstanding on quantitative relationships among the major compartments in the phosphorus cycle, in particul ar, water column-floc laye r interaction, is one of the most significant factors. For example, although the annual average concentrations of measured and simulated PP are almost the same (Table 6-4), the simulation does not effectively catch temporal fluctuation of PP concentration profiles. This is primarily due to the uncertainty of model parameters on spatio-t emporal variations of mass tran sfer mechanisms between water column and floc layer. For instance, the simulati on result would be better if the spatio-temporal variations for non-mechan ical resuspension of IPf ( v_resus2 ) and OPf ( v_resus4 ) were considered rather than using constant mode l parameter values. Actually, it is, however, extremely difficult to obtain relevant information at the field scale. Due to the inter-dependency of state variab les, the ECO Lab model calibration is not a simple task. For example, even though one state va riable is calibrated to fit measured values, it may not fit well after calibrating other state variables. Therefore, the state variable with the least dependencies on other ones should be first calibrate d. In this study, as there is no time series measurement data for Pmacro, Pperi, IPf, OPf, SRPf, IPs, OPs, and SRPs, it was unfortunately not possible to calibrate these stat e variables. As a result, floc/soil phosphorus components were calibrated to maintain almost constant concentr ation profiles by changi ng related parameters. There are always fluctuations of annual concentr ation profiles of these ph osphorus species in real world; however, it was assumed that the extent s were not dramatically changed during the one year simulation period, compared to water co lumn species. Figure 6-17 presents the annual steady state concentration profiles of six state variables in floc and soil layers at six data

PAGE 164

164 monitoring sites. In the case of Pmacro and Pperi, associated growth and decay parameters were determined, considering a general growth pa ttern of vegetation w hose population increases during the wet, high temperature season and decr eases during the dry, low temperature season. Figure 6-18 shows the annual pattern of Pmacro and Pperi concentration profiles at randomly selected twelve grid cells. It is very difficult to simulate phosphorus transport/transformation processes in constructed wetlands accurately largely due to: (1) lack of dynamic characteristics of measured phosphorus data, (2) complexity of spatio -temporal hydrodynamics and solute transport processes, and (3) lack of scientific understanding of quantitative relationships am ong major compartments in the phosphorus cycle. First, lack of spatio-temporal data measured in field and laboratory for phosphorus species in water column, floc, and soil layer severely limits development of more detailed phosphorus dynamics model. For example, the role of PIP and POP in water column in phosphorus dynamics is totally different; however, each portion in part iculate phosphorus is rarely analyzed. The STA system in South Florida is one of the most we ll-studied constructed wetl and areas in the world with the intensive flow and wate r quality monitoring effort. Neve rtheless, a couple of inevitable assumptions on phosphorus compartmentalization, based on measurements from the adjacent wetlands like WCA-2A, of which vegetation and soil physicochemical conditions are similar, were required to develop the phosphorus dynamics model. This shows the need for spatiotemporal monitoring of each phosphorus state vari ables suggested in this study for development of a more robust phosphorus dynamics model. Second, spatio-temporal heterogeneity of fl ow dynamics and the associated solute transport processes make it difficult to predict phosphorus behavior in constructed wetlands. In

PAGE 165

165 this study, in spite of the uncertainty of seepag e inflow/outflow in the STA 5, flow dynamics and solute transport are simulated with a relatively high accuracy (Chapter 5). However, the results of phosphorus dynamics simulations show that ph osphorus species closely related to physical processes controlled by flow dynami cs (i.e. PP) were partially ove restimated or underestimated. This indicates that there may be some remaining factors that are not fully considered. Limitations of this modeling study are discussed in the next chapter. Finally, the lack of certainty of the qu antitative relationships among the major compartments in phosphorus cycle makes it difficult to simulate phosphorus transformation processes in constructed wetlands accurately. In addition, each kinetic pathway in the phosphorus cycle is influenced by various meteor ological or biogeochemical factors, which are still qualitatively or quantitativ ely unknown. For example, there were two hurricanes (Frances and Jeanne in September 2004) during the mode ling period, but such impacts on phosphorus dynamics in constructed wetlands are not well kn own. Moreover, effects of key biogeochemical parameters, such as redox and pH, on phosphorus re lease and retention were not fully considered in this study. All these factors indicate that more extens ive data collection effo rts and/or additional studies on flow and phosphorus dynamics should be undertaken to support development of models that may then be used to design future systems. Sensitivity analysis After obtaining the best mode l fit between simulated and observed phosphorus data by tuning model parameters, the sensitivities of each simulated SRPw, DOP, and PP concentration profile at four different wa ter quality measurement points (G 343B, G343C, G344A, and G344B) to the key model parameters were examined. The model constant values determined by the model calibration process (Table 6-1) were used as baseline values. In each sensitivity model

PAGE 166

166 run, only a single model parameter was changed within a certain range. Results of these sensitivity tests are expressed as RMSE, which is defined in Chapter 5. From Table 6-8, the most sensitive model parameters on SRPw are those related to phosphorus uptake by SAV ( kg_macro and ks_macro ). The next sensitive ones are macrophyte decomposition rate ( k_decay4 ), critical velocity ( v_crit ), and SRPw adsorption/precipitation rate ( k_form2 ). For DOP, the most sensitive parameters are critical velocity ( v_crit ), POP deposition rate ( v_dep2 ) and mineralization rate constant ( k_min ) as shown in Table 6-9. Model parameters associated with water column-floc layer in teraction, such as critical velocity ( v_crit ), POP deposition rate ( v_dep2 ), and non-mechanical resuspension of IPf and OPf ( v_resus2 and v_resus4 ), are most sensitive on PP (Table 6-10). The overall results of the sensitivity analysis reveal that model parameters c ontrolling water column-floc layer interactions are most sensitive in this phosphorus dynamics model. Summary To simulate spatio-temporal variations of phosphorus concen trations monitored in STA 5 northern flow-way, an ECO Lab phosphorus dynam ics model incorporating differences between the phosphorus cycles for the EAV and SAV syst ems was developed and linked with the HD and AD models, which are accurately calibrated with the flow dyna mics and solute transport characteristics of study area. Prior to phosphorus dynamics modeling, conservative SRPw, DOP, and PP transport simulations were attempted both to estimate wate r column phosphorus retention and to figure out how different the reactive char acteristics are with respect to phosphorus species and wetland vegetative communities. The extent of phosphorus retention estimated through the conservative transport simulation and the associated phosphoru s behavior in EAV and SAV treatment system confirm the findings of recent STA studies that the EAV system is less efficient for phosphorus

PAGE 167

167 retention, in particular, SRPw, compared to the SAV system. Furthermore, the current STA system is not sufficient for reducing DOP concentration to very low levels. Although phosphorus dynamics in floc and soil layers and vegetative communities were not verifiable due to the lack of spatio-temporal measured field data, and several assumptions for developing the model were adapted in this modeli ng study, the model results show that simulated phosphorus concentration profiles agree reasonably with measured data. The calibrated model better simulated observed annual variations in SRPw and DOP levels than those in the PP level, which is primarily due to uncertainty of model pa rameters on spatio-tempor al variations of mass transfer mechanisms between the water column and floc layers. This shows that extensive data collecting efforts and more in-depth studies on flow and phosphorus dynamics are mandatory to develop more robust phosphorus dynamics models Sensitivity analysis reveals that model parameters related to sedimentation and resuspensi on processes controlled by critical velocity are most sensitive for water column pho sphorus species in the flow-way.

PAGE 168

168 Table 6-1. Calibrated or estimated values of constants used in the ECO Lab phosphorus dynamics model. Symbol Description Unit Value 1. f_atm1 2. f_atm2 3. f_atm3 4. f_atm4 5. k_decay1 6. k_decay2 7. k_decay3 8. k_decay4 9. k_decay5 10. k_decay6 11. k_decay7 12. k_form1 13. k_form2 14. k_form3 15. k_form4 16. k_form5 17. k_min 18. k_deg_f 19. k_deg_s 20. v_crit 21. v_dep1 22. v_dep2 23. v_resus1 24. v_resus2 25. v_resus3 26. v_resus4 27. kg_macro 28. kg_root 29. kg_peri 30. ks_macro 31. ks_root 32. ks_peri 33. D 34. d_diff_f 35. d_diff_s 36. d_ave_f 37. d_ave_s 38. frac_root 39. bd_f 40. bd_s 41. poro_f 42. poro_s Atmospheric deposition flux of SRPw Atmospheric deposition flux of DOP Atmospheric deposition flux of PIP Atmospheric deposition flux of POP Decay rate constant (POP DOP) Decay rate constant (PIP SRPw) Decay rate constant (Pperi POP) Decay rate constant (Pmacro POP) Decay rate constant (IPf SRPf) Root decay rate constant (Pmacro OPs) Decay rate constant (IPs SRPs) Formation rate constant (DOP POP) Formation rate constant (SRPw PIP) Formation rate constant (POP Pperi) Formation rate constant (SRPf IPf) Formation rate constant (SRPs IPs) Mineralization rate constant (DOP SRPw) Degradation rate constant (OPf SRPf) Degradation rate constant (OPs SRPs) Critical velocity of flow Deposition of PIP Deposition of POP Mechanical resuspension of IPf Non-mechanical (biologi cal) resuspension of IPf Mechanical resuspension of OPf Non-mechanical (biological) resuspension of OPf Macrophyte max. growth rate constant Macrophyte root max. growth rate constant Periphyton max. growth rate constant Macrophyte uptake half saturation constant Macrophyte root uptake half saturation constant Periphyton uptake half saturation constant Effective diffusion coefficient Depth of diffusive exchange in floc layer Depth of diffusive exchange in upper soil layer Average depth of floc layer Average depth of upper soil layer Root fraction of macrophyte Bulk density of floc layer Bulk density of upper soil layer Floc layer porosity Upper soil layer porosity g/m2/day g/m2/day g/m2/day g/m2/day 1/day 1/day 1/day 1/day 1/day 1/day 1/day 1/day 1/day 1/day L/mg/day L/mg/day 1/day 1/day 1/day m/s m/day m/day m/day m/day m/day m/day 1/day 1/day 1/day mg/L mg/L mg/L m2/day m m m m kg/L kg/L 1.31-56.38-61.07-41.07-40.25 0.1 0/0.1 0.01/0.02 0.1 0.002/0.004 0.1 0.02 0.02/0.12 0/0.05 0.075 0.0005 0.22 0.0005 0.0005 0.05 0.4 0.4 0.001 0.00005 0.001 0.00008 0/0.06 0.015 0/0.15 0/0.1 1 0/0.05 3.41-50.044 0.05 0.088 0.1 0.667/0.05 Field data Field data 0.95 0.74

PAGE 169

169 Table 6-1. Continued. Symbol Description Unit Value 43. Smax_f 44. Smax_s 45. f_excop_f 46. f_excop_s 47. f_excip_f 48. f_excip_s 49. f_seq1 50. f_seq2 51. f_seq3 52. f_seq4 53. f_leak1 54. f_leak2 55. 56. min_wd Max. adsorption capacity in floc layer Max. adsorption capacity in upper soil layer Biodegradable phosphorus fraction of OPf Biodegradable phosphorus fraction of OPs Exchangeable phosphorus fraction of IPf Exchangeable phosphorus fraction of IPs Sequestration flux of IPf into IPs Sequestration flux of OPf into OPs Sequestration flux of IPs into deep soil layer Sequestration flux of OPs into deep soil layer Leaching flux of SRPs into deep soil layer Leaching flux into SRPs from deep soil layer Temperature coefficient of Arrhenius equation Minimum water depth mg/kg mg/kg g/m2/day g/m2/day g/m2/day g/m2/day g/m2/day g/m2/day m 500 250 0.25 0.1 0.5 0.01 0.0008/0.0001 0.0008/0.0001 0.0044/0.0025 0.0015/0.0003 0 0 1 0.03

PAGE 170

170Table 6-2. Differences of phosphorus dynamics model between EAV and SAV wetland ecosystem. EAV wetland ecosystem SAV wetland ecosystem Source State variable Periphyton Macrophyte Average floc TP Average soil TP Not considered 0.580 g/m3 958.5 mg/kg 466.1 mg/kg 0.021 g/m30.173 g/m3 675.8 mg/kg 510.2 mg/kg Literature Literature Field Field Processes SRPw uptake by macrophyte in water column SRPs uptake by macrophyte root in soil layer Macrophyte root decay SRPw uptake by periphyton in water column Pperi decay (periphyton sloughing) Pperi form (periphyton cohesion) Ca-P coprecipitation via periphyton Not considered Dominant Considered Not considered Not considered Not considered Not considered Dominant Minimal Considered Considered Considered Considered Considered Constants used in phosphorus dynamics model k_decay3 (Pperi POP) k_decay4 (Pmacro POP) k_decay6 (Pmacro OPs) k_form2 (SRPw PIP) k_form3 (POP Pperi) kg_macro (Macrophyte max. growth rate) kg_peri (Periphyton max. growth rate) ks_macro (Macrophyte uptake half saturation constant) ks_peri (Periphyton uptake half saturation constant) frac_root (Root fraction of macrophyte) Average floc bulk density Average soil bulk density f_seq1 (Sequestration flux of IPf into IPs) f_seq2 (Sequestration flux of OPf into OPs) f_seq3 (Sequestration flux of IPs into deep soil) f_seq4 (Sequestration flux of OPs into deep soil) 0 0.01 0.002 0.02 0 0 0 0 0 0.667 0.05 kg/L 0.69 kg/L 0.0008 g/m2/d 0.0008 g/m2/d 0.0044 g/m2/d 0.0015 g/m2/d 0.1 0.02 0.004 0.12 0.05 0.06 0.15 0.1 0.05 0.05 0.11 kg/L 0.40 kg/L 0.0001 g/m2/d 0.0001 g/m2/d 0.0025 g/m2/d 0.0003 g/m2/d Calibration Calibration Calibration Calibration Calibration Calibration Calibration Calibration Calibration Literature Field Field Calibration Calibration Literature Literature

PAGE 171

171 Table 6-3. Annual phosphorus budget for STA 5 northern flow-way in Water Year (WY) 2004 (May 03-April 04) estimated through conservative phosphor us transport simulation. Inputs Outputs WY 04 ISF P IN OSF OGW OUT Ret RMSE [%Ret] PWC 10.088 0.039 10.1275.0540.3455.3994.609 2.079 [45.5%] 0.118 SRP 10.098* 0.039* 10.137*5.166*0.721*5.887*4.250* [41.9%] 1.483 0.019 1.5021.2920.0941.3850.124 0.389 [8.2%] -0.007 DOP 1.484* 1.484*1.293*0.138*1.431*0.053* [3.6%] 4.769 0.638 5.4071.9800.5952.5762.719 1.427 [50.3%] 0.113 PP 9.433* 9.433*1.947*0.428*2.375*7.058* [74.8%] 16.340 0.696 17.0368.3261.0349.3607.452 [43.7%] 0.224 TP 21.015* 0.116** 21.131*8.406*1.287*9.693*11.361* [53.8%] Unit: metric ton Note: Ret = retention (= INOUTPWC); %Ret = percent retention (= Ret/ IN); PWC = change in water column phosphorus concentration. Data provided by Chimney (2007). ** Data provided by Pietro et al. (2006).

PAGE 172

172 Table 6-4. Comparison of the annual average pho sphorus concentrations between measurement and simulation during the model calibrati on period (May 1, 2003 to April 30, 2004). SRP DOP PP Measured 0.054 0.011 0.014 G343B Simulated 0.054 0.010 0.014 Measured 0.064 0.009 0.015 G343C Simulated 0.058 0.011 0.014 Measured 0.028 0.013 0.017 G344A Simulated 0.030 0.011 0.018 Measured 0.044 0.014 0.020 G344B Simulated 0.038 0.013 0.021 Measured 0.047 0.012 0.017 Average Simulated 0.045 0.011 0.017 Unit: mg/L Table 6-5. RMSE values of the phosphorus dyna mics model during the model calibration period (May 1, 2003 to April 30, 2004). SRP DOP PP G343B 0.025 0.005 0.013 G343C 0.035 0.004 0.015 G344A 0.023 0.008 0.014 G344B 0.041 0.006 0.019 Average 0.031 0.006 0.015 Unit: mg/L

PAGE 173

173 Table 6-6. Comparison of the average phosphorus concentrations between measurement and simulation during the model validation period (May 1, 2004 to December 31, 2004). SRP DOP PP Measured 0.048 0.013 0.033 G343B Simulated 0.059 0.012 0.015 Measured 0.052 0.013 0.034 G343C Simulated 0.064 0.013 0.015 Measured 0.019 0.012 0.014 G344A Simulated 0.033 0.012 0.019 Measured 0.037 0.014 0.018 G344B Simulated 0.040 0.014 0.021 Measured 0.039 0.013 0.025 Average Simulated 0.049 0.013 0.017 Unit: mg/L Table 6-7. RMSE values of the phosphorus dyna mics model during the model validation period (May 1, 2004 to December 31, 2004). SRP DOP PP G343B 0.025 0.007 0.024 G343C 0.026 0.005 0.026 G344A 0.023 0.007 0.011 G344B 0.031 0.010 0.013 Average 0.026 0.007 0.019 Unit: mg/L

PAGE 174

174Table 6-8. Results of sensitivity analysis of SRPw on major parameters used in the phosphorus dynamics model. Upper RMSE (RMSEs**RMSEc*) Parameter Baseline value (E/S) Range Lower G343B G343C G344A G344B Average 0.5 day-1 0.001 0.000 0.001 0.000 0.000 k_decay1 0.25 day-1 0.05 day-1 0.000 0.000 0.001 0.002 0.001 0.5 day-1 0.000 0.000 0.002 0.001 0.001 k_decay2 0.1 day-1 0.01 day-1 0.000 0.000 0.000 0.000 0.000 0/0.5 day-1 0.000 0.000 0.001 -0.001 0.000 k_decay3 0/0.1 day-1 0/0.05 day-1 0.004 0.005 0.002 0.010 0.005 0.05/0.1 day-1 0.006 0.001 0.002 0.000 0.002 k_decay4 0.01/0.02 day-1 0.005/0.01 day-1 0.007 0.011 0.002 0.007 0.007 0.1 day-1 0.000 0.000 0.000 0.001 0.000 k_form1 0.02 day-1 0.005 day-1 0.000 0.000 0.000 0.000 0.000 0.1/0.6 day-1 0.004 0.004 0.004 0.007 0.005 k_form2 0.02/0.12 day-1 0.005/0.03 day-1 0.004 0.012 0.005 0.003 0.006 0/0.5 day-1 0.001 0.002 0.002 0.008 0.003 k_form3 0/0.05 day-1 0/0.01 day-1 0.000 0.001 0.001 -0.001 0.000 0.5 day-1 0.001 0.001 0.001 0.001 0.001 k_min 0.22 day-1 0.05 day-1 0.000 0.001 0.000 0.000 0.000 0.5 m/s 0.000 0.000 0.000 0.000 0.000 v_crit 0.05 m/s 0.01 m/s 0.001 0.002 0.011 0.012 0.007 1 m/day 0.000 0.000 0.000 0.000 0.000 v_dep1 0.4 m/day 0.1 m/day 0.001 0.000 0.002 0.001 0.001 1 m/day 0.000 0.000 0.001 0.001 0.001 v_dep2 0.4 m/day 0.1 m/day 0.003 0.001 0.001 0.000 0.001 Unit: mg/L (E/S) represents the baseline value at the emergent and submerged aquatic vege tation areas, respectively. RMSEc is a RMSE between the measured and simulated SRPw concentration profile generated by model calibration. ** RMSEs is a RMSE between the measured and simulated SRPw concentration profile gene rated by sensitivity test.

PAGE 175

175Table 6-8. Continued. Upper RMSE (RMSEs**RMSEc*) Parameter Baseline value (E/S) Range Lower G343B G343C G344A G344B Average 0.01 m/day 0.000 0.000 0.000 0.001 0.000 v_resus1, 3 0.001 m/day 0.0001 m/day 0.000 0.000 0.000 0.000 0.000 1-4/2-4 m/d0.001 0.009 0.001 0.000 0.003 v_resus2, 4 5-5/8-5 m/day 1-5/2-5 m/d 0.000 0.000 0.001 0.001 0.001 0/0.2 day-1 0.029 0.037 0.003 0.011 0.020 kg_macro 0/0.06 day-1 0/0.01 day-1 0.009 0.003 0.002 0.000 0.003 0.05 day-1 0.003 0.000 0.000 0.000 0.001 kg_root 0.015 day-1 0.005 day-1 0.000 0.000 0.000 0.000 0.000 0/0.3 day-1 0.007 0.011 -0.001 0.007 0.006 kg_peri 0/0.15 day-1 0/0.03 day-1 0.000 0.000 0.001 -0.001 0.000 0/0.5 mg/L 0.008 0.002 0.002 -0.001 0.003 ks_macro 0/0.1 mg/L 0/0.02 mg/L 0.019 0.027 0.005 0.013 0.016 3 mg/L 0.000 0.000 0.000 0.000 0.000 ks_root 1 mg/L 0.1 mg/L 0.000 0.000 0.000 0.000 0.000 0/0.25 mg/L 0.000 0.000 0.001 -0.001 0.000 ks_peri 0/0.05 mg/L 0/0.01 mg/L 0.003 0.003 0.003 0.011 0.005 1.1 0.005 0.010 0.000 0.002 0.004 1 Unit: mg/L (E/S) represents the baseline value at the emergent and submerged aquatic vege tation areas, respectively. RMSEc is a RMSE between the measured and simulated SRPw concentration profile generated by model calibration. ** RMSEs is a RMSE between the measured and simulated SRPw concentration profile gene rated by sensitivity test.

PAGE 176

176Table 6-9. Results of sensitivity analysis of DOP on ma jor parameters used in the phosphorus dynamics model. Upper RMSE (RMSEs**RMSEc*) Parameter Baseline value (E/S) Range Lower G343B G343C G344A G344B Average 0.5 day-1 0.002 0.004 0.002 0.003 0.003 k_decay1 0.25 day-1 0.05 day-1 0.003 0.002 0.004 0.005 0.004 0.5 day-1 0.000 0.000 0.000 0.000 0.000 k_decay2 0.1 day-1 0.01 day-1 0.000 0.000 0.000 0.000 0.000 0/0.5 day-1 0.000 0.000 0.000 0.000 0.000 k_decay3 0/0.1 day-1 0/0.05 day-1 0.006 0.018 0.000 0.003 0.007 0.05/0.1 day-1 0.001 0.001 0.000 0.000 0.000 k_decay4 0.01/0.02 day-1 0.005/0.01 day-1 -0.001 0.001 -0.001 0.000 0.000 0.1 day-1 0.000 0.000 0.000 0.001 0.001 k_form1 0.02 day-1 0.005 day-1 0.000 0.000 0.000 0.000 0.000 0.1/0.6 day-1 0.000 0.000 0.000 0.001 0.000 k_form2 0.02/0.12 day-1 0.005/0.03 day-1 0.000 0.000 0.000 0.000 0.000 0/0.5 day-1 0.000 0.000 0.001 0.004 0.001 k_form3 0/0.05 day-1 0/0.01 day-1 0.000 0.000 0.000 0.000 0.000 0.5 day-1 0.001 0.001 0.003 0.004 0.002 k_min 0.22 day-1 0.05 day-1 0.005 0.009 0.009 0.011 0.009 0.5 m/s 0.000 0.000 0.000 0.000 0.000 v_crit 0.05 m/s 0.01 m/s 0.007 0.009 0.032 0.039 0.022 1 m/day 0.000 0.000 0.000 0.000 0.000 v_dep1 0.4 m/day 0.1 m/day 0.000 0.000 0.000 0.000 0.000 1 m/day 0.002 0.001 0.003 0.003 0.002 v_dep2 0.4 m/day 0.1 m/day 0.007 0.012 0.009 0.012 0.010 Unit: mg/L (E/S) represents the baseline value at the emergent and submerged aquatic vege tation areas, respectively. RMSEc is a RMSE between the measured and simulated DOP concentration profile generated by model calibration. ** RMSEs is a RMSE between the measured and simulated DOP concentration profile generated by sensitivity test.

PAGE 177

177Table 6-9. Continued. Upper RMSE (RMSEs**RMSEc*) Parameter Baseline value (E/S) Range Lower G343B G343C G344A G344B Average 0.01 m/day 0.000 0.000 0.000 0.001 0.001 v_resus1, 3 0.001 m/day 0.0001 m/day 0.000 0.000 0.000 0.000 0.000 1-4/2-4 m/d0.001 0.004 0.003 0.004 0.003 v_resus2, 4 5-5/8-5 m/day 1-5/2-5 m/d 0.001 0.001 0.003 0.003 0.002 0/0.2 day-1 0.008 0.006 0.001 0.001 0.004 kg_macro 0/0.06 day-1 0/0.01 day-1 0.001 0.000 0.000 0.000 0.000 0.05 day-1 0.005 0.004 0.000 0.000 0.002 kg_root 0.015 day-1 0.005 day-1 0.000 0.000 0.000 0.000 0.000 0/0.3 day-1 0.004 0.008 0.001 0.004 0.004 kg_peri 0/0.15 day-1 0/0.03 day-1 0.000 0.000 0.000 0.000 0.000 0/0.5 mg/L 0.001 0.000 0.000 0.000 0.000 ks_macro 0/0.1 mg/L 0/0.02 mg/L 0.006 0.006 0.000 0.001 0.003 3 mg/L 0.000 0.000 0.000 0.000 0.000 ks_root 1 mg/L 0.1 mg/L 0.000 0.000 0.000 0.000 0.000 0/0.25 mg/L 0.000 0.000 0.000 0.000 0.000 ks_peri 0/0.05 mg/L 0/0.01 mg/L 0.002 0.004 0.001 0.003 0.003 1.1 0.001 0.002 0.003 0.005 0.003 1 Unit: mg/L (E/S) represents the baseline value at the emergent and submerged aquatic vege tation areas, respectively. RMSEc is a RMSE between the measured and simulated DOP concentration profile generated by model calibration. ** RMSEs is a RMSE between the measured and simulated DOP concentration profile generated by sensitivity test.

PAGE 178

178Table 6-10. Results of sensitivity anal ysis of PP on major parameters used in the phosphorus dynamics model. Upper RMSE (RMSEs**RMSEc*) Parameter Baseline value (E/S) Range Lower G343B G343C G344A G344B Average 0.5 day-1 0.000 0.000 0.000 0.000 0.000 k_decay1 0.25 day-1 0.05 day-1 0.000 0.000 0.001 0.000 0.000 0.5 day-1 0.000 0.000 0.000 0.000 0.000 k_decay2 0.1 day-1 0.01 day-1 0.000 0.000 0.001 0.000 0.000 0/0.5 day-1 0.000 0.000 0.000 0.000 0.000 k_decay3 0/0.1 day-1 0/0.05 day-1 0.012 0.022 0.000 0.000 0.009 0.05/0.1 day-1 0.000 0.000 -0.001 -0.001 -0.001 k_decay4 0.01/0.02 day-1 0.005/0.01 day-1 0.000 0.000 0.000 0.000 0.000 0.1 day-1 0.000 0.000 0.000 0.000 0.000 k_form1 0.02 day-1 0.005 day-1 0.000 0.000 0.000 0.000 0.000 0.1/0.6 day-1 0.001 0.001 0.002 0.002 0.002 k_form2 0.02/0.12 day-1 0.005/0.03 day-1 -0.001 0.000 -0.001 0.000 -0.001 0/0.5 day-1 0.000 0.000 0.001 0.001 0.001 k_form3 0/0.05 day-1 0/0.01 day-1 0.000 0.000 0.000 0.000 0.000 0.5 day-1 0.000 0.000 0.000 0.000 0.000 k_min 0.22 day-1 0.05 day-1 0.000 0.000 0.000 0.000 0.000 0.5 m/s 0.000 0.000 0.000 0.000 0.000 v_crit 0.05 m/s 0.01 m/s 0.029 0.044 0.096 0.099 0.067 1 m/day 0.000 0.000 0.000 0.000 0.000 v_dep1 0.4 m/day 0.1 m/day 0.001 0.001 0.004 0.004 0.002 1 m/day 0.001 0.002 0.001 0.001 0.001 v_dep2 0.4 m/day 0.1 m/day 0.006 0.009 0.011 0.012 0.009 Unit: mg/L (E/S) represents the baseline value at the emergent and submerged aquatic vege tation areas, respectively. RMSEc is a RMSE between the measured and simulated PP concentration profile generated by model calibration. ** RMSEs is a RMSE between the measured and simulated PP c oncentration profile genera ted by sensitivity test.

PAGE 179

179Table 6-10. Continued. Upper RMSE (RMSEs**RMSEc*) Parameter Baseline value (E/S) Range Lower G343B G343C G344A G344B Average 0.01 m/day 0.001 0.003 0.001 0.002 0.002 v_resus1, 3 0.001 m/day 0.0001 m/day 0.000 0.000 0.000 0.000 0.000 1-4/2-4 m/d0.002 0.002 0.007 0.005 0.004 v_resus2, 4 5-5/8-5 m/day 1-5/2-5 m/d 0.001 0.001 0.002 0.002 0.002 0/0.2 day-1 0.016 0.005 0.000 0.000 0.005 kg_macro 0/0.06 day-1 0/0.01 day-1 0.000 0.000 0.000 0.000 0.000 0.05 day-1 0.003 0.000 0.000 0.000 0.001 kg_root 0.015 day-1 0.005 day-1 0.000 0.000 0.000 0.000 0.000 0/0.3 day-1 0.016 0.018 0.000 0.000 0.009 kg_peri 0/0.15 day-1 0/0.03 day-1 0.000 0.000 0.000 0.000 0.000 0/0.5 mg/L 0.000 0.000 0.000 0.000 0.000 ks_macro 0/0.1 mg/L 0/0.02 mg/L 0.011 0.005 0.000 0.000 0.004 3 mg/L 0.000 0.000 0.000 0.000 0.000 ks_root 1 mg/L 0.1 mg/L 0.000 0.000 0.000 0.000 0.000 0/0.25 mg/L 0.000 0.000 0.000 0.000 0.000 ks_peri 0/0.05 mg/L 0/0.01 mg/L 0.005 0.005 0.001 0.001 0.003 1.1 0.002 0.003 0.003 0.001 0.002 1 Unit: mg/L (E/S) represents the baseline value at the emergent and submerged aquatic vege tation areas, respectively. RMSEc is a RMSE between the measured and simulated PP concentration profile generated by model calibration. ** RMSEs is a RMSE between the measured and simulated PP c oncentration profile genera ted by sensitivity test.

PAGE 180

180 Inlet phosphorus concentration profiles at G342A0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Phosphorus concentration (mg/ L G342A_SRP G342A_DOP G342A_PPA) Relative portion of inlet phosphorus species at G342A SRP DOP PP0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 5/16/17/18/19/110/111/112/11/12/13/14/15/16/17/18/19/110/111/112/11/1Time (May 1, 2003 to December 31, 2004)Relative percentage (%)B) Figure 6-1. Temporal variations of inlet phos phorus species at G342A. A) Time-series SRPw, DOP, and PP concentration profiles. B) Relative portion of in let phosphorus species.

PAGE 181

181 Inlet phosphorus concentration profiles at G342B0.000 0.040 0.080 0.120 0.160 0.200 0.240 0.280 0.320 0.360 0.400 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Phosphorus concentration (mg/ L G342B_SRP G342B_DOP G342B_PPA) Relative portion of inlet phosphorus species at G342B SRP DOP PP0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 5/16/17/18/19/110/111/112/11/12/13/14/15/16/17/18/19/110/111/112/11/1Time (May 1, 2003 to December 31, 2004)Relative percentage (%)B) Figure 6-2. Temporal variations of inlet phos phorus species at G342B. A) Time-series SRPw, DOP, and PP concentration profiles. B) Relative portion of in let phosphorus species.

PAGE 182

182 Inlet phosphorus concentration profiles at G349A0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Phosphorus concentration (mg/ L G349A_SRP G349A_DOP G349A_PPA) Relative portion of inlet phosphorus species at G349A SRP DOP PP0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 5/16/17/18/19/110/111/112/11/12/13/14/15/16/17/18/19/110/111/112/11/1Time (May 1, 2003 to December 31, 2004)Relative percentage (%)B) Figure 6-3. Temporal variations of inlet phos phorus species at G349A. A) Time-series SRPw, DOP, and PP concentration profiles. B) Relative portion of in let phosphorus species.

PAGE 183

183 A) B) Figure 6-4. Raster-typed phos phorus content prediction maps in the floc and th e upper soil layer of STA 5 northern flow-way use d in the ECO Lab phosphorus dynamics model. A) IPf. B) OPf. C) IPs. D) OPs.

PAGE 184

184 C) D) Figure 6-4. Continued.

PAGE 185

185 Temperature profile at G343C12.0 14.0 16.0 18.0 20.0 22.0 24.0 26.0 28.0 30.0 32.0 5/1/036/30/038/29/0310/28/0312/27/032/25/044/25/046/24/048/23/0410/22/0412/21/04Time (May 1, 2003 to December 31, 2004)Temperature (oC) Figure 6-5. Temperature profile used in the phosphorus dynamics model, measured at G343C.

PAGE 186

186 Figure 6-6. A schematic of phosphorus dynamics in the EAV system of STA 5.

PAGE 187

187 Figure 6-7. A schematic of phosphorus dynamics in the SAV system of STA 5.

PAGE 188

188 G343B SRP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)A) G343C SRP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)B) Figure 6-8. Results of conservative solute tr ansport simulation from May 1, 2003 to December 31, 2004 on SRPw concentration profiles at the four water quality measurement points in STA 5 northern flow-way. A) G343B B) G343C. C) G3 44A. D) G344B.

PAGE 189

189 G344A SRP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)C) G344B SRP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)D) Figure 6-8. Continued.

PAGE 190

190 G343B DOP Concentration Profile0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)A) G343C DOP Concentration Profile0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)B) Figure 6-9. Results of conservative solute tr ansport simulation from May 1, 2003 to December 31, 2004 on DOP concentration profiles at th e four water quality measurement points in STA 5 northern flow-way. A) G343B B) G343C. C) G3 44A. D) G344B.

PAGE 191

191 G344A DOP Concentration Profile0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)C) G344B DOP Concentration Profile0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)D) Figure 6-9. Continued.

PAGE 192

192 G343B PP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)A) G343C PP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)B) Figure 6-10. Results of conservative solute tr ansport simulation from May 1, 2003 to December 31, 2004 on PP concentration profiles at the four water quality m easurement points in STA 5 northern flow-way. A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 193

193 G344A PP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)C) G344B PP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Concentration (mg/L) Measured Simulated (Conservative)D) Figure 6-10. Continued.

PAGE 194

194 G343B SRP (RMSE = 0.025 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)A) G343C SRP (RMSE = 0.035 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)B) Figure 6-11. Calibration results of flow and phosphorus dynamics coupled simulation from May 1, 2003 to April 30, 2004 on SRPw concentration profiles at the four water quality measurement points in STA 5 northern fl ow-way. A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 195

195 G344A SRP (RMSE = 0.023 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)C) G344B SRP (RMSE = 0.041 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)D) Figure 6-11. Continued.

PAGE 196

196 G343B DOP (RMSE = 0.005 mg/L)0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)A) G343C DOP (RMSE = 0.004 mg/L)0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)B) Figure 6-12. Calibration results of flow and phosphorus dynamics coupled simulation from May 1, 2003 to April 30, 2004 on DOP concentra tion profiles at the four water quality measurement points in STA 5 northern fl ow-way. A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 197

197 G344A DOP (RMSE = 0.008 mg/L)0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)C) G344B DOP (RMSE = 0.006 mg/L)0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)D) Figure 6-12. Continued.

PAGE 198

198 G343B PP (RMSE = 0.013 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)A) G343C PP (RMSE = 0.015 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)B) Figure 6-13. Calibration results of flow and phosphorus dynamics coupled simulation from May 1, 2003 to April 30, 2004 on PP concentrati on profiles at the four water quality measurement points in STA 5 northern fl ow-way. A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 199

199 G344A PP (RMSE = 0.014 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)C) G344B PP (RMSE = 0.019 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Dynamics)D) Figure 6-13. Continued.

PAGE 200

200 G343B SRP (RMSE = 0.025 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)A) G343C SRP (RMSE = 0.026 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)B) Figure 6-14. Validation results of flow and phosphorus dynamics coupled simulation from May 1, 2004 to December 31, 2004 on SRPw concentration profiles at the four water quality measurement points in STA 5 north ern flow-way. A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 201

201 G344A SRP (RMSE = 0.023 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)C) G344B SRP (RMSE = 0.031 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)D) Figure 6-14. Continued.

PAGE 202

202 G343B DOP (RMSE = 0.007 mg/L)0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)A) G343C DOP (RMSE = 0.005 mg/L)0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)B) Figure 6-15. Validation results of flow and phosphorus dynamics coupled simulation from May 1, 2004 to December 31, 2004 on DOP concentra tion profiles at the four water quality measurement points in STA 5 northern fl ow-way. A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 203

203 G344A DOP (RMSE = 0.007 mg/L)0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)C) G344B DOP (RMSE = 0.010 mg/L)0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)D) Figure 6-15. Continued.

PAGE 204

204 G343B PP (RMSE = 0.024 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)A) G343C PP (RMSE = 0.026 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)B) Figure 6-16. Validation results of flow and phosphorus dynamics coupled simulation from May 1, 2004 to December 31, 2004 on PP concentratio n profiles at the four water quality measurement points in STA 5 northern fl ow-way. A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 205

205 G344A PP (RMSE = 0.011 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)C) G344B PP (RMSE = 0.013 mg/L)0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 5/15/316/307/308/299/2810/2811/2712/27Time (May 1, 2004 to December 31, 2004)Concentration (mg/L) Measured Simulated (Dynamics)D) Figure 6-16. Continued.

PAGE 206

206 A) B) C) D) E) F) Figure 6-17. Results of the ECO Lab phosphorus dyna mics model on six state variables in floc and soil layers at six field monitoring sites. A) SRPf. B) IPf. C) OPf. D) SRPs. E) IPs. F) OPs.

PAGE 207

207 A) B) Figure 6-18. Results of the ECO Lab phosphorus dynamics model on Pmacro and Pperi at randomly chosen twelve model grid cells. A) Pmacro. B) Pperi.

PAGE 208

208 CHAPTER 7 CONCLUSIONS AND RECOMMENDATIONS Conclusions In summary, the following conclusions are drawn in this study: (1) The 2-D, fully-distributed flow dynamics, solute transport, and phosphorus dynamics models, developed using MIKE 21 HD, AD, a nd ECO Lab as the basic framework with modifications, successfully simulated the spatio-tem poral variations of water level fluctuation, tracer (bromide/chloride), and phosphorus concen tration profiles in two constructed wetland systems in Florida (OEW Cell 7 and STA 5 northern flow-way). (2) Hydroperiods of the two c onstructed wetlands were accurately simulated by simply using a source/sink option in MIKE 21 HD mode l, rather than specifying flow boundary conditions in the constructed we tland systems surrounded by levees. (3) Relic ditches or other ditch-shaped landf orms and the associated sparse vegetation along the main flow direction intensify short-ci rcuiting flow patterns in constructed wetlands, considerably affecting 2-D solute transport model results. (4) In terms of hydraulic efficiency, the effect of bathymetry on short-circuiting flow is more sensitive than th e vegetation effect. (5) The Mannings roughness coefficients as me trics of hydraulic resistance, estimated as a function of vegetation type and density and ca librated by the models, ranged from 0.022 to 0.045 s/m1/3 for no or sparse vegetation regions (shortcircuiting flow zones), from 0.67 to 1.0 s/m1/3 for dense EAV regions, and from 0.12 to 0.15 s/m1/3 for dense SAV regions, respectively. (6) Dispersivity, determined by chloride tr ansport model calibration for STA 5 northern flow-way, was anisotropic. Longitudinal dispersiv ity, estimated to be 2 m, was over an order of

PAGE 209

209 magnitude higher than the transverse one. Th e sensitivity test showed that longitudinal dispersivity is a more important parameter than the transverse one. (7) Results of the conservative phosphorus tr ansport model for STA 5 northern flow-way confirm findings of recent STA studies that the EAV system is less efficient for phosphorus retention, in particular, SRPw, compared to the SAV system and that the current STA system is not sufficient for reducing DOP conc entration to very low levels. (8) The STA 5 northern flow-way flow and phosphorus dynamics integrated model incorporating differences of phosphorus cycle between the EAV and SAV systems, entirely better simulated observed annual variations in SR P and DOP levels than those in the PP level, which is primarily due to uncertainty of model pa rameters on spatio-tempor al variations of mass transfer mechanisms between water column and floc layer. (9) Model parameters related to sedimenta tion and resuspension processes, including critical velocity, were most se nsitive for water column phosphor us species in the flow-way. Recommendations Results of this study leave more to be invest igated and answered. Th is section discusses what remains to be determined by future resear ch. Before reviewing topics for further study, some technical shortcomings and model limitati ons found in this study to improve modeling of flow and phosphorus dynamics are discussed as follows: (1) In the current MIKE 21 HD model, there is no option to describe hydraulic resistance as a function of water depth. As a result, it is not possible to give tempor al variation of hydraulic roughness coefficients due to the tr ansient flow and stage pattern in most stormwater treatment wetlands. It was reported by Sutr on Corp. (2005) that water de pth-dependency on Mannings n values has an obvious impact on fl ow distribution. Comprehensive research to address selection

PAGE 210

210 of Mannings coefficient values as a function of vegetation type, de nsity, and water depth has not been performed to date. (2) Hydraulic structures, such as culverts a nd weirs, are not explicitly represented in the current MIKE 21 HD model. Alt hough this study showed that a point source and sink option can be relatively successfully applied in replacemen t of inlet and outlet hydraulic structures, it was still not possible to express the hydraulic structures at the middle levee located between treatment cells. This may be one of the most cr itical reasons that highe r simulation deviations from the measured hydroperiod, tracer and phosphor us concentration profiles were observed in Cell 1B than in Cell 1A. (3) The hydraulic relationship between tr eatment cells and surrounding canals is not considered in this study. Water flow in STA 5 basically relies on a hydrau lic gradient formed by stages in the L-2 Canal (upstream) and those in the Discharge Canal (downstream). According to Liyanage and Huebner (2005), the design of curr ent STA 5 gated culverts is susceptible to backflow or reverse flow under certain unco mmon operation conditions like the storm season. Although the time series water le vel profiles at several intern al measurement points were relatively accurately simulated through the HD model, the change of short-term flow velocity, affected directly by the relationship and hydrau lic structure manipulati ons for a management purpose, is not fully verified in this study. Th erefore, if the relations hips are appropriately incorporated in the current HD model, the simulation results of phosphorus transfer mechanisms between water column and floc layer (sedim entation and resuspensi on), as well as the HD simulation itself, may be improved. (4) Some discrepancies between measured and simulated SRP and PP concentration profiles in this study, pa rticularly in SAV system, indicate the possibility of existence of

PAGE 211

211 unknown factors affecting the dynamics. For exam ple, mechanical resuspension or solute transport by wind (hurricane) is not considered in this study; however, their impact may be significant in open water or open water areas with SAV. Systematic studies on wind effects and the difference between EAV and SAV are rarely accomplished in constructed wetland systems, compared to other aquatic bodies such as lakes and estuaries (Chen, 1994). (5) Historically, the main focus of treatment wetland modeling, even in dynamics modeling studies, has been on the behavior and removal of phosphorus species in the water column. In this study, due to th e lack of time series field da ta sufficient to calibrate the phosphorus species in floc and upper soil layer, porewater and soil phosphorus species in the layers were set almost c onstant level during the mode l calibration period through the manipulation of the related parameter values. In addition, it is impossible to calibrate vegetationrelated phosphorus dynamics parameters without field data to show the annual change in population density (concentration) and distribution. For example, although periphyton is found in only 20% of all SAV areas, but it was assumed in this study that periphyton exists on all SAV areas in Cell 1B because there ar e no data for the distribution. As a result, the role of periphyton may be overestimated in this study. Therefore, it is necessary to develop submodels for floc/soil and vegetation (EAV, SAV, and periphyton) phosphorus dynamics, which are precisely calibrated and validated based on time series fiel d and laboratory data collected at several locations in a constructed wetland. When thes e submodels are coupled together, a phosphorus cycle among water column-soil-vege tation will be more quantitatively understood with smaller uncertainties. (6) In this study, two phosphorus dynamics mo dels for EAVand SAV-dominant treatment wetland cell are suggested. However, research on quantitative estimation of key model

PAGE 212

212 parameters in each kinetic pathway, particularly in SAV and periphyton dominant systems, is still required because of the lack of comple te understanding of all processes involved in phosphorus cycles in constructed wetlands. The more information on phosphorus dynamics in constructed wetlands is develope d, the more quantitative underst anding of the roles of various kinetic pathways in the cycle is increased. Numerical modeling approach es for flow and phosphorus dynamics in constructed wetlands are still in their infancy. None of the existing numerical models for phosphorus dynamics has the real ability to predict an episodic event, particularly a short-term event generated by storm or drought. Mainly due to th e limitations and shortcomings mentioned above, the integrated model presented in this study is also not fully satisfactory fo r its forecas ting ability. Based on achievements and complements of this modeling study, two future research topics for model applications are briefly suggested: (1) The hydrodynamics-based phosphorus dynami cs model presented in this study can provide a useful tool to test various hypotheses for best manage ment of constructed wetlands like the South Florida STAs. For example, effective management of both desirable and non-desirable vegetation within the STAs play s a key role in achieving and sustaining long-term phosphorus reduction goals for the Everglades (Goforth, 2005) A variety of management scenarios can be tested to fulfill a goal for outflow phosphorus levels or to increase treat ment efficiency (i.e. hydraulic/phosphorus loading rate reduction, chan ge of vegetation type and distribution, and internal bathymetry ma nipulations to reduce sh ort-circuiting flow). (2) Regarding applicability of the flow a nd phosphorus dynamics model, the role as a predictive model is one of the most critical con cerns in constructed wetl and systems. To make predictions for long-term treatment efficiency or longevity, more systematic understanding of

PAGE 213

213 phosphorus dynamics in floc/soil layer and phosphor us kinetic pathways between water column and floc/soil layer, which are related to the internal loading of phosphorus, is mandatory. For this, a unit HD model, based on long-term historic al hydrological data (i.e inflow, rainfall, and evapotranspiration), may be necessary to be linked with a phosphorus dynamics model, which will be improved by more in-depth scientif ic knowledge on physical and biogeochemical processes taking place in a constructed wetland ecos ystem, particularly in the floc and top soil layer.

PAGE 214

214 APPENDIX A RECALIBRATION RESULTS OF HYDRODYNAMICS SIMULA TION FROM MAY 1, 2003 TO DECEMBER 31, 2004 G343B_H Water Level (RMSE = 0.084 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured Simulated A) G343B_T Water Level (RMSE = 0.110 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured Simulated B) Figure A-1. Recalibration results of hydrodynami cs simulation from May 1, 2003 to December 31, 2004 on hydroperiod fluctuation at the si x water level measurement points in STA 5 northern flow-way. A) G343B_H. B) G 343B_T. C) G343C_H. D) G343C_T. E) G344A_H. F) G344B_H.

PAGE 215

215 G343C_H Water Level (RMSE = 0.093 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured Simulated C) G343C_T Water Level (RMSE = 0.098 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured Simulated D) Figure A-1. Continued.

PAGE 216

216 G344A Water Level (RMSE = 0.133 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured Simulated E) G344B Water Level (RMSE = 0.166 m)3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5/16/308/2910/2812/272/254/256/248/2310/2212/21Time (May 1, 2003 to December 31, 2004)Water Level (m) Measured Simulated F) Figure A-1. Continued.

PAGE 217

217 APPENDIX B CALIBRATION OF THE CHLORIDE TRANSPORT AD MODEL ON LONGITUDINAL AND TRANSVERSE DISPERSIVITY Table B-1. Calibration results of the chloride transport AD model according to the ratio of longitudinal ( x) vs. transverse dispersivity ( y) in STA 5 northern flow-way. (m) RMSE (mg/L) x : y x y G343B G343C G344A G344B Average 0.01 0.01 11.8219.3022.5223.50 19.28 0.1 0.1 10.4116.1622.2122.43 17.80 0.5 0.5 9.7712.8822.0020.84 16.37 0.8 0.8 10.1612.3521.8120.55 16.22 0.9 0.9 10.3012.3021.7420.51 16.21 1 1 10.4312.2821.6620.48 16.21 1.1 1.1 10.5612.2821.5920.46 16.22 1.2 1.2 10.6812.3021.5220.45 16.24 1.3 1.3 10.7912.3321.4620.44 16.25 1.5 1.5 10.9912.4021.3320.44 16.29 2 2 11.3912.6021.0620.49 16.39 3 3 11.8812.9720.7020.67 16.56 4 4 12.1713.2420.5020.90 16.70 1:1 10 10 12.8014.1420.4821.95 17.34 0.1 0.05 10.1324.3722.2922.59 19.85 0.5 0.25 9.5516.6121.6820.97 17.21 0.8 0.4 9.8913.9321.4720.54 16.46 1 0.5 10.1613.0021.3620.38 16.22 1.3 0.65 10.5612.3821.2120.23 16.09 1.4 0.7 10.6812.3121.1720.20 16.09 1.5 0.75 10.8012.2721.1220.17 16.09 1.8 0.9 11.1212.3521.0020.13 16.15 2 1 11.3112.4820.9220.11 16.21 3 1.5 11.5913.2420.6520.21 16.42 4 2 12.0613.5720.4720.38 16.62 2:1 10 5 13.0814.1820.4021.57 17.31

PAGE 218

218 Table B-1. Continued. (m) RMSE (mg/L) x : y x y G343B G343C G344A G344B Average 0.1 0.02 10.3024.9822.3422.80 20.11 0.5 0.1 9.4517.2221.4521.22 17.33 0.8 0.16 9.5114.4321.1020.73 16.44 1 0.2 9.5412.9820.9820.55 16.01 1.3 0.26 9.7112.7720.8020.32 15.90 1.5 0.3 9.8512.7420.7120.21 15.88 1.6 0.32 9.9212.7420.6820.17 15.88 1.7 0.34 9.9912.7620.6420.13 15.88 1.8 0.36 10.0612.7920.6120.09 15.89 2 0.4 10.2112.8720.5520.03 15.92 2.5 0.5 10.5613.1420.4519.94 16.02 4 0.8 11.4213.9520.2819.94 16.40 5:1 10 2 13.0314.8420.3520.87 17.27 0.1 0.01 10.3925.3122.3722.89 20.24 0.5 0.05 9.4917.5021.4021.35 17.44 0.8 0.08 9.4614.7120.9720.85 16.50 1 0.1 9.4913.2920.8020.66 16.06 1.5 0.15 9.6612.9820.4620.30 15.85 1.8 0.18 9.8112.9920.3220.16 15.82 1.85 0.185 9.8413.0020.3020.14 15.82 1.9 0.19 9.8613.0120.2820.12 15.82 2 0.2 9.9113.0420.2520.08 15.82 2.1 0.21 9.9713.0820.2220.05 15.83 2.2 0.22 10.0213.1220.1920.02 15.84 2.5 0.25 10.1813.2720.1219.94 15.88 3 0.3 10.4513.5620.0419.86 15.98 4 0.4 10.9214.1019.9719.81 16.20 10:1 10 1 12.5715.4420.2320.45 17.17

PAGE 219

219 Table B-1. Continued. (m) RMSE (mg/L) x : y x y G343B G343C G344A G344B Average 0.1 0.005 10.4325.5122.3822.95 20.32 0.5 0.025 9.5417.6921.3921.44 17.52 1 0.05 9.5213.4920.7420.74 16.12 1.5 0.075 9.6413.1720.3320.39 15.88 2 0.1 9.8313.2120.0820.15 15.82 2.05 0.1025 9.8513.2320.0620.13 15.82 2.1 0.105 9.8713.2520.0420.12 15.82 2.15 0.1075 9.8913.2720.0220.10 15.82 2.2 0.11 9.9113.2920.0020.08 15.82 2.5 0.125 10.0413.4319.9120.00 15.84 3 0.15 10.2413.7119.8019.89 15.91 4 0.2 10.6214.2519.6819.79 16.09 20:1 10 0.5 12.0315.8019.9820.19 17.00 0.1 0.002 10.4725.6522.3922.99 20.37 0.5 0.01 9.5917.8321.4021.51 17.58 1 0.02 9.5813.9020.6820.79 16.24 1.5 0.03 9.6913.3120.2920.45 15.94 2 0.04 9.8713.3620.0020.22 15.86 2.1 0.042 9.9113.4019.9620.18 15.86 2.2 0.044 9.9413.4419.9120.15 15.86 2.3 0.046 9.9813.4819.8720.12 15.86 2.5 0.05 10.0513.5819.8020.06 15.87 3 0.06 10.2313.8619.6619.95 15.93 4 0.08 10.5414.4319.4919.82 16.07 50:1 10 0.2 11.5816.0619.6220.02 16.82

PAGE 220

220 Table B-1. Continued. (m) RMSE (mg/L) x : y x y G343B G343C G344A G344B Average 0.1 0.001 10.4825.7022.4023.00 20.39 0.5 0.005 9.6117.8921.4021.53 17.61 1 0.01 9.6013.9620.6820.82 16.27 1.5 0.015 9.7313.3720.2820.47 15.96 2 0.02 9.9113.4219.9920.25 15.89 2.1 0.021 9.9513.4619.9420.21 15.89 2.2 0.022 9.9913.5019.9020.17 15.89 2.3 0.023 10.0313.5419.8620.14 15.89 3 0.03 10.2813.9419.6419.97 15.96 4 0.04 10.5914.5219.4519.84 16.10 100:1 10 0.1 11.4616.2019.4719.98 16.77

PAGE 221

221 APPENDIX C RESULTS OF CONSERVATIVE SOLUTE TRANS PORT SIMULATION ( 1 RMSE) FROM MAY 1, 2003 TO APRIL 30, 2004 ON SRP CONCENTRATION PROFILES G343B SRP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSEA) G343C SRP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSEB) Figure C-1. Results of conservati ve solute transport simulation ( 1 RMSE) from May 1, 2003 to April 30, 2004 on SRPw concentration profiles at the four water quality measurement points in STA 5 northern flow-way. A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 222

222 G344A SRP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSEC) G344B SRP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 0.180 0.200 0.220 0.240 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSED) Figure C-1. Continued.

PAGE 223

223 APPENDIX D RESULTS OF CONSERVATIVE SOLUTE TRANS PORT SIMULATION ( 1 RMSE) FROM MAY 1, 2003 TO APRIL 30, 2004 ON DOP CONCENTRATION PROFILES G343B DOP Concentration Profile0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0.045 0.050 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSEA) G343C DOP Concentration Profile0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0.045 0.050 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSEB) Figure D-1. Results of conservati ve solute transport simulation ( 1 RMSE) from May 1, 2003 to April 30, 2004 on DOP concentr ation profiles at the four water quality measurement points in STA 5 northern flow-way. A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 224

224 G344A DOP Concentration Profile0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0.045 0.050 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSEC) G344B DOP Concentration Profile0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0.045 0.050 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSED) Figure D-1. Continued

PAGE 225

225 APPENDIX E RESULTS OF CONSERVATIVE SOLUTE TRANS PORT SIMULATION ( 1 RMSE) FROM MAY 1, 2003 TO APRIL 30, 2004 ON PP CONCENTRATION PROFILES G343B PP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSEA) G343C PP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSEB) Figure E-1. Results of conservati ve solute transport simulation ( 1 RMSE) from May 1, 2003 to April 30, 2004 on PP concentration profiles at the four water quality measurement points in STA 5 northern flow-way. A) G343B. B) G343C. C) G344A. D) G344B.

PAGE 226

226 G344A PP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSEC) G344B PP Concentration Profile0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 5/15/316/307/308/299/2810/2811/2712/271/262/253/264/25Time (May 1, 2003 to April 30, 2004)Concentration (mg/L) Measured Simulated (Conservative) +1 RMSE -1 RMSED) Figure E-1. Continued.

PAGE 227

227 LIST OF REFERENCES Abtew, W. J. (1996). Evapotranspiration measur ements and modeling for the wetland systems in South Florida. Water Res. Bull. 32(3), 465. lvarez, J. A., and Bcares, E. (2006). Seasonal decomposition of Typha latifolia in a freewater surface constructed wetland. Ecol. Eng. 28(2), 99. Ambrose, R. B., Wool, T. A., Connolly J. P., and Schoanz, R. W. (1988). WASP4, a hydrodynamic and water quality model: mode l theory, users manual, and programmers guide EPA/600/3-87/039, U.S. Environmen tal Protection Agency, Athens, Ga. Appan, A., and Wang, H. (2000). Sorption isotherms and kinetics of sediment phosphorus in a tropical reservoir. J. Environ. Eng. 126(11), 993. Arnold, J. G., Allen, P. M., and Morgan, D. S. (2001). Hydrologic model for design and constructed wetlands. Wetlands 21(2), 167. Arnold, J. G., Williams, J. R., Srinivasan, R., King, K. W., and Griggs, R. H. (1994). SWAT, Soil and water assessment tool USDA Agricultural Research Service, Temple, Tx. Asaeda, T., and Van Bon, T. (1997). Modelling the effects of macrophytes on algal blooming in eutrophic shallow lakes. Ecol. Model. 104(2-3), 261. Barko, J. W., and James, W. F. (1997). Effects of submerged aqua tic macrophytes on nutrient dynamics, sedimentation, and resuspension. In: Jeppesen et al. (Ed.), The structuring role of submerged macrophytes in la kes. Ecological studies, Vol. 131 Springer, New York, 197. Bays, J. S., Knight, R. L., Wenke rt, L., Clarke, R., and Gong, S. (2001). Progress in the research and demonstration of Everglades periphyt on-based stormwater treatment areas. Water Sci. Technol. 44(11-12), 123. Bencala, K. E., and Walters, R. A. (1983). Sim ulation of solute transport in a mountain pooland-riffle stream: a transient storage model. Water Resour. Res. 19(3), 718. Black, C. A., and Wise, W. R. (2003). Evaluation of past and potential phosphorus uptake at the Orland easterly wetland. Ecol. Eng. 21, 277. Blom, G., and Aalderink, R. H. (1998). Ca libration of three resu spension/sedimentation models. Water Sci. Technol. 37(3), 41. Blom, G., and Toet, C. (1993). Modelling sediment transport and sediment quality in a shallow Dutch lake (Lake Ketel). Water Sci. Technol. 28(8-9), 79.

PAGE 228

228 Bowie, G. L., Mills, W. B., Porcella, D. B., Campbell, C. L., Pagenkopf, J. R., Rupp, G. L., Johnson, K. M., Chan, P. W. H., Gherin i, S., and Chamberlin, C. E. (1985). Rates, constants, and kinetics formulati ons in surface water quality modeling 2nd Ed., EPA/600/3-85/040, U.S. Environmental Prot ection Agency, Environmental Research Laboratory, Athens, Ga., 455 pp. Braskerud, B. C. (2001). The influence of ve getation on sedimentation and resuspension of soil particles in small constructed wetlands. J. Environ. Qual. 30, 1447. Braskerud, B. C. (2002). Factor s affecting phosphorus retention in small constructed wetlands treating agricultural nonpoint source pollution. Ecol. Eng. 19, 41. Brezonik, P. L., Hendry, C. D., Edgerton, E. S., Schulze, R., and Crisman, T. L. (1983). Acidity, nutrients, and minerals in atmospheric precip itation over Florida: Deposition patterns, mechanisms and ecological effects EPA-600/S3-83-004, U.S. Environmental Protection Agency, Corvallis, Or. (NTIS document PB 83-165 837). Brix, H. (1994). Use of construc ted wetlands in water pollution c ontrol: historical development, present status, and future perspectives. Water Sci. Technol. 30(8), 209. Carleton, J. N., Grizzard, T. J ., Godrej, A. N., and Post, H. E. (2001). Factors affecting the performance of stormwater treatment wetlands. Water Res. 35(6), 1552. Chao, X., Jia, Y., Cooper, C. M., Shields Jr., F. D., and Wang, S. S. Y. (2006). Development and application of a phosphorus model for a shallow oxbow lake. J. Environ. EngASCE 132(11), 1498. Chapra, S. C. (1997). Surface water-quality modeling McGraw-Hill Publishers, New York, Ny., 844 pp. Chen, X. (1994). Effects of hydrodynamics and sediment transport processes on nutrient dynamics in shallow lakes and estuaries. Ph .D. dissertation, Univ ersity of Florida, Gainesville, Fl. Chen, X., and Sheng, Y. P. (2005). Three-dime nsional modeling of sediment and phosphorus dynamics in Lake Okeechobee, Florida: Spring 1989 simulation. J. Environ. Eng-ASCE, 131(3), 359. Chimney, M. (2007). Appendix 5-15: Nutrient budgets for th e flow-ways and treatment cells in the STAs In: 2007 South Florida Environmental Report South Florida Water Management District (SFWMD), West Palm Beach, Fl. Chimney, M. J., and Goforth, G. F. (2001). Environmental impacts to the Everglades ecosystem: a historical perspectiv e and restoration strategies. Water Sci. Technol. 44(11-12), 93.

PAGE 229

229 Chimney, M. J., and Goforth, G. (2006). Histo ry and description of th e Everglades Nutrient Removal Project, a subtropical construc ted wetland in South Florida (USA). Ecol. Eng. 27, 268. Chimney, M. J., and Pietro, K. C. (2006). D ecomposition of macrophyte litter in a subtropical constructed wetland in South Florida (USA). Ecol. Eng. 27, 301. Chimney, M. J., Wenkert, L., and Pietro, K. C. (2006). Patterns of vertic al stratification in a subtropical constructed wetla nd in South Florida (USA). Ecol. Eng. 27, 322. Chow, Ven Te. (1959). Open channel hydraulics McGraw-Hill, New York. Clement, A., Somlydy, L., and Koncsos, L. (1998). Modeling the phosphorus relation of the Kis-Balaton upper reservoir. Water Sci. Technol. 37(3), 113. Corstanje, R., Grunwald, S., Reddy, K. R., Os borne, T. Z., and Newman, S. (2006). Assessment of the spatial distri bution of soil properties in a northern everglades marsh. J. Environ. Qual. 35(3), 938. Craft, C. B., and Richardson, C. J. (1993). Peat accretion and N, P, and organic C accumulation in nutrient-enriched and unenr iched Everglades peatlands. Ecol. Appl. 3(3), 446. Dadaser-Celik, F., Stefan, H. G., and Brezonik, P. L. (2006). Dynamic hydrologic model of the rtlakar marsh in Turkey. Wetlands 26(4), 1089. Danish Hydraulic Institute (DHI). (2004a). ECO Lab short scie ntific description Danish Hydraulic Institute, Hrsholm, Denmark. Danish Hydraulic Institute (DHI). (2004b). MIKE 21 a modeling system for estuaries, coastal water and seas user guide Danish Hydraulic Institute, Hrsholm, Denmark. Danish Hydraulic Institute (DHI). (2004c). WQ templates scientific description Danish Hydraulic Institute, Hrsholm, Denmark. Danish Hydraulic Institute (DHI). (2005a). MIKE 21 flow model advection-dispersion module scientific documentation Danish Hydraulic Institute, Hrsholm, Denmark. Danish Hydraulic Institute (DHI). (2005b). MIKE 21 flow model hydrodynamics module scientific documentation Danish Hydraulic Institute, Hrsholm, Denmark. Das, B. S., and Kluitenberg, G. J. (1996). M oment analysis to estimate degradation rate constants from leaching experiments. Soil Sci. Soc. Am. J. 60, 1724. Davis, C. B., and van der Valk, A. G. (1978). T he decomposition of standing and fallen litter of Typha glauca and Scirpus fluviatilis Can. J. Bot. 56, 662.

PAGE 230

230 Davis, S. M. (1984). Cattail leaf production, mortality, and nutrient flux in Water Conservation Area 2A Techn. Publ. #84-8, South Florida Wa ter Management District, West Palm Beach, Fl. Davis, S. M. (1991). Growth, decomposition, and nutrient retention of Cladium jamaicense Crantz and Typha domingensis Pers. in the Florida Everglades. Aquat. Bot. 40(3), 203 224. Davis, S. M. (1994). Phosphorus inputs and vegetation sensitivity in the Everglades. In: Davis, S. M., and Ogden, J. C. (Ed.), Everglades-the ecosystem and its restoration St. Lucie Press, Fl, 357. DeAngelis, D. L., Loreau, M., Neergaard, D., Mu lholland, P. J., and Marzolf, E. R. (1995). Modelling nutrient-periphyton dynamics in stream s: the importance of transient storage zone. Ecol. Model. 80, 149. DeBusk, T. A., Grace, K. A., Dierberg, F. E., Jack son, S. D., Chimney, M. J., and Gu, B. (2004). An investigation of the lim its of phosphorus removal in wetlands: a mesocosm study of a shallow periphyton-dominated treatment system. Ecol. Eng. 23, 1. DeBusk, W. F., and Reddy, K. R. (2005). Litter decomposition and nutrient dynamics in a phosphorus enriched everglades marsh. Biogeochemistry 75(2), 217. Dennison, W. C., Orth, R. J., Moore, K. A., Steven son, J. C., Carter, V., Kollar, S., Bergstrom, P. W., and Batiuk, R. A. (1993). Assessing water quality with submersed aquatic vegetation. BioScience 43(2), 86. Dierberg, F. E., DeBusk, T. A., Jackson, S. D., Chimney, M. J., and Pietro, K. (2002). Submerged aquatic vegetation-based trea tment wetlands for removing phosphorus from agricultural runoff: response to hydraulic and nutr ient loading. Water Res. 36, 1409 1422. Dierberg, F. E., Juston, J. J., DeBusk, T. A ., Pietro, K., and Gu, B. (2005). Relationship between hydraulic efficiency and phosphorus removal in a submerged aquatic vegetationdominated treatment wetland. Ecol. Eng. 25, 9. Di Toro, D. M., and Fitzpatrick, J. J. (1993). Chesapeake bay sediment flux model Report to U.S. Army Corp. Engineer, waterways experiment station, Vicksburg, Ms. Donigian, A. S., and Huber, W. C. (1990). Modeling of nonpoint sour ce water quality in urban and non-urban areas U.S. Environmental Protection Agency, Athens, Ga. Doren, R. F., Armentano, T. V., Whiteaker, L. D., and Jones, R. D. (1997). Marsh vegetation patterns and soil phosphoru s gradients in the Everglades ecosystem. Aquat. Bot. 56(199), 145.

PAGE 231

231 Ekebjrg, L., and Justesen, P. (1991). An explic it scheme for advection-dispersion modeling in two dimensions. Comput. Meth. Appl. Mech. Eng. 88, 287. Environmental Lab. (1990). CE-QUAL-RIV1: a dynamic one-dim ensional (longitudinal) water quality model for streams Instruction report E-90-1, U.S. Army Corp. Engineer waterways experiment station, Vicksburg, Ms. Feng, K., and Molz, F. J. (1997). A 2D, diffusion-based, wetland flow model. J. Hydrol. 196, 230. Fernandez, G. P., Chescheir, G. M., Skaggs R. W., and Amatya, D. M. (2006). DRAINMODGIS: a lumped parameter watershed scale drainage and water quality model. Agric. Water Manage. 81(1-2), 77. Findlay, S., Howe, K., and Austin, H. K. (1990). Comparison of detritus dynamics in two tidal freshwater wetlands. Ecology 71, 288. Fisher, M. M., and Reddy, K. R. (2001). Phos phorus flux from wetland soils affected by longterm nutrient loading. J. Environ. Qual. 30, 261. Gchter, R., and Meyer, J. S. (1993). The role of microorganisms in mobilization and fixation of phosphorus in sediments. Hydrobiologia 253(1-3), 103. Giraud, F., Faure, J. B., Zimmer, D., Lefeuvre, J. C., and Skaggs, R. W. (1997). Hydrologic modeling of a complex wetland. J. Irrig. Drainage Eng-ASCE 123(5), 344. Goforth, G. F. (2001). Surmounting the engine ering challenges of Everglades restoration. Water Sci. Technol. 44(11-12), 295. Goforth, G. F. (2005). Summary of STA vegetation management practices prepared for the South Florida Water Management District, February 2005. Goforth, G., Pietro, K., Chimney, M., Newman, J ., Bechtel, T., Germain, G., and Iricanin, N. (2005). Appendix 4-13: Vegetation coverage m aps for STA-2, STA-5, and STA-6 December 2003 In: 2005 South Florida Environmental Report South Florida Water Management District (SFWMD), West Palm Beach, Fl. Gopal, B. (1999). Natural and constructed wetl ands for wastewater tr eatment: potentials and problems. Water Sci. Technol. 40(3), 27. Grace, K. A. (2003). Phosphorus removal and so il stability within emergent and submerged vegetation communities in treatment wetlands. M.S. Thesis, University of Florida, Gainesville, Fl.

PAGE 232

232 Grimshaw, H. J., and Dolske, D. A. (2002). R ainfall concentrations and wet atmospheric deposition of phosphorus and other c onstituents in Florida, U.S.A. Water Air Soil Pollut. 137, 117. Grunwald, S., Corstanje, R., Weinrich, B. E., and Reddy, K. R. (2006). Spatial patterns of labile forms of phosphorus in a subtropical wetland. J. Environ. Qual. 35(1), 378. Gu, B., DeBusk, T. A., Dierberg, F. E., Chimney, M. J., Pietro, K. C., and Aziz, T. (2001). Phosphorus removal from Everglades agri cultural area runoff by submerged aquatic vegetation/limerock treatment tec hnology: an overview of research. Water Sci. Technol. 44(11-12), 101. Haberl, R. (1999). Constructed wetlands: a chan ce to solve wastewater problems in developing countries. Water Sci. Technol. 40(3), 11. Hammer, D. A. (1989). Constructed wetlands for wastewater treatment Lewis Publishers, Chelsea, Mi, USA. Han, J. S. (1985). Net pr imary production in a marsh. Michigan Bot. 24, 55. Harvey, J. W., Saiers, J. E., and Newlin, J. T. (2005). Solute transport and storage mechanisms in wetlands of the Everglades, South Florida. Water Resour. Res. 41, W05009, doi:10.1029/2004WR003507. Harvey, J. W., Wagner, B. J., and Bencala, K. E. (1996). Evaluating the re liability of the stream tracer approach to characterize stream-subsurface water exchange. Water Resour. Res. 32(8), 2441. Hendry, C. D., Edgerton, E. S., and Brezonik, P. L. (1981). Atmospheric deposition of nitrogen and phosphorus in Florida In: Eisenreich, S. J. (Ed.), Atmospheric pollutants in natural waters Ann Arbor Sci. Publ., Ann Arbor, Mi, 199. Holland, J. F., Martin, J. F., Granata, T., B ouchard, V., Quigley, M., and Brown, L. (2004). Effects of wetland depth and flow rate on re sidence time distribution characteristics. Ecol. Eng. 23, 189. Horppila, J., and Nurminen, L. (2003). Eff ects of submerged macrophytes on sediment resuspension and internal phosphorus loading in Lake Hiidenvesi (southern Finland). Water Res. 37, 4468. Howell, C. J., Crohn, D. M., and Omary, M. ( 2005). Simulating nutrient cycling and removal through treatment wetlands in arid/semiarid environments. Ecol. Eng. 25, 25. Huber, W. C., and Dickinson, R. E. (1988). Storm Water Management Model Users Manual, v.4 EPA/600/3-88/001a, U.S. Environmenta l Protection Agency, Cincinnati, Oh.

PAGE 233

233 HydroQual. (1995). SFWMD wetlands model: model and state-variable selection Report to South Florida Water Management District, West Palm Beach, Fl. Janse, J. H. (1998). A model of ditch vegetation in relation to eutrophication. Water Sci. Technol. 37(3), 139. Jenkins, G. A., and Greenway, M. (2005). The h ydraulic efficiency of fringing versus banded vegetation in constructed wetlands. Ecol. Eng. 25, 61. Jrgensen, S. E., and Bendoricchio, G. (2001). Fundamentals of ecological modeling 3rd Ed., Elsevier, Oxford, UK, 530 pp. Juston, J., and DeBusk, T. A. (2006). Phosphorus mass load and outflow concentration relationships in stormwater treatment areas for Everglades restoration. Ecol. Eng. 26, 206. Kadlec, R. H. (1994a). Detention and mixing in free water wetlands. Ecol. Eng. 3(4), 345 380. Kadlec, R. H. (1994b). Phosphorus uptake in Florida marshes. Water Sci. Technol. 30(8), 225. Kadlec, R. H. (1997). An autobiotic wetland phosphorus model. Ecol. Eng. 8, 145. Kadlec, R. H. (1999a). The limits of phosphorus removal in wetlands. Wetlands Ecol. Manage. 7, 165. Kadlec, R. H. (1999b). Transport of phosphorus in wetlands In: Reddy et al. (Ed.), Phosphorus biogeochemistry in s ubtropical ecosystems Lewis Publishers, Boca Raton, Fl, 355. Kadlec, R. H. (2000). The inadequacy of first-order treatmen t wetland models. Ecol. Eng. 15(1), 105. Kadlec, R. H. (2005). Wetland to pond treatment gradients. Water Sci. Technol. 51(9), 291 298. Kadlec, R. H. (2006). Free surface wetlands for phosphorus removal: The position of the Everglades Nutrient Removal Project. Ecol. Eng. 27, 361. Kadlec, R. H., and Hammer, D. E. (1988). M odeling nutrient behavior in wetlands. Ecol. Model. 40(1), 37. Kadlec, R. H., and Knight, R. L. (1996). Treatment wetlands Lewis Publishers, Boca Raton, Fl, 893 pp.

PAGE 234

234 Kadlec, R. H., and Newman, S. (1992). Everglades protection Stor mwater Treatment Area design support Phosphorus removal in we tland treatment areas Principles and data DOR 106. Kadlec, R. H., and Reddy, K. R. (2001). Temperature effects in treatment marshes. Water Environ. Res. 73(5), 543. Kamp-Nielsen, L. (1975). A kinetic approach to the aerobic sediment -water exchange of phosphorus in Lake Esrom. Ecol. Model. 1, 153. Keefe, S. H., Barber, L. B., Runkel, R. L., Ry an, J. N., McKnight, D. M., and Wass, R. D. (2004). Conservative and reactive solute transport in constructed wetlands. Water Resour. Res. 40, W01201, doi:10.1029/2003WR002130. Knight, R. L., Gu, B., Clarke, R. A., and Newman, J. M. (2003). Long-term phosphorus removal in Florida aquatic systems dominated by submerged aquatic vegetation. Ecol. Eng. 20, 45. Koch, M. S. (1991). Soil and surface water nutrients in the Everglades Nutrient Removal Project Techn. Publ. #91-04, South Florida Wa ter Management District, West Palm Beach, Fl. Konyha, K. D., Shaw, D. T., and Weiler, K. W. (1995). Hydrologic design of a wetland: advantages of continuous modeling. Ecol. Eng. 4, 99. Koskiaho, J. (2003). Flow velocity retardati on and sediment retenti on in two constructed wetland-ponds. Ecol. Eng. 19(5), 325. Larsen, D. P., Mercier, H. T., and Malveg, K. W. (1974). Modeling algal growth dynamics in Shagawa Lake, Minnesota. In: Middlebrooke, E. J., Falke nberg, D. H., and Maloney, T. E. (Ed.), Modeling eutrophication process Ann Arbor Science, Michigan, 15. Lee, H., and Shih, S. (2004). Impacts of ve getation changes on the hydraulic and sediment transport characteristics in Guandu mangrove wetland. Ecol. Eng. 23, 85. Lee, J. K., Roig, L. C., Jenter, H. L., and Visse r, H. M. (2004). Drag coefficients for modeling flow through emergent vegetation in the Florida Everglades. Ecol. Eng. 22, 237. Leon, L. F., Soulis, E. D., Kouwen, N., and Fa rquhar, G. J. (2001). Nonpoint source pollution: a distributed water quality modeling approach. Water Res. 35(4), 997. Levenspiel, O. (1972). Chemical reaction engineering 2nd Ed., John Wiley, New York, 253 325. Li, Y. H., and Gregory, S. (1974) Diffusion of ions in sea wate r and in deep sea sediments. Geochim. Cosmochim. Acta 38, 703.

PAGE 235

235 Lijklema, L. (1993). Considerations in modeli ng the sediment-water exchange of phosphorus. Hydrobiologia 253, 219. Liyanage, T. R., and Huebner, R. S. (2005). Water budget analysis for Stormwater Treatment Area 5 (Water Year 2003-2004) Techn. Publ. ERA #427, South Florida Water Management District, West Palm Beach, Fl. Loftin, C. S., Kitchens, W. M., and Ansay, N. (2001). Development and a pplication of a spatial hydrology model of Okefenokee swamp, Georgia. J. Am. Water Resour. Assoc. 37(4), 935. Lung, W. (2001). Water quality modeling for wa steload allocations and TMDLs John Wiley, New York, 71. Mansell, R. S., Bloom, S. A., and Sun, G. (2000). A model for wetla nd hydrology: description and validation. Soil Sci. 165(5), 384. Martinez, C. J., and Wise, W. R. (2003a). A nalysis of constructed treatment wetland hydraulic with the transient storage model OTIS. Ecol. Eng. 20, 211. Martinez, C. J., and Wise, W. R. (2003b). Hydr aulic analysis of the Orlando easterly wetland. J. Environ. Eng-ASCE 129(6), 553. McCormick, P. V., Shuford III, R. B. E., Bac kus, J. G., and Kennedy, W. C. (1998). Spatial and seasonal patterns of periphyton biomass and productivity in the northern Everglades, Florida, U.S.A. Hydrobiologia 362, 185. McCormick, P. V., Shuford III, R. B. E., and Chimney, M. J. (2006). P eriphyton as a potential phosphorus sink in the Everglades Nutrient Removal Project. Ecol. Eng. 27, 279. Mitsch, W. J. (1994). Global wetlands old world and new Elsevier, Amsterdam, The Netherlands, 967 pp. Mitsch, W. J., Cronk, J. K., Wu, X., and Nair n, R. W. (1995). Phosphorus retention in constructed freshwater riparian marshes. Ecol. Appl. 5(3), 830. Mitsch, W. J., and Gosselink, J. G. (1993). Wetlands 2nd Ed., Van Nostrand Reinhold, New York, 722 pp. Mitsch, W. J., Horne, A. J., and Nairn, R. W. (2000). Nitrogen and phosphorus retention in wetlandsecological approaches to solving excess nutrient problems. Ecol. Eng. 14, 1 7.

PAGE 236

236 Mitsch, W. J., and Jrgensen, S. E. (1990). Modelling and management In: Patten et al. (Ed.), Wetlands and shallow contin ental water bodies. Vol. I. SPB Academic Publ., The Hague, The Netherlands, 727. Nepf, H. M. (1999). Drag, turbulence, and di ffusion in flow through emergent vegetation. Water Resour. Res. 35(2), 479. Newman, J. M., and Lynch, T. (2001). The Ever glades Nutrient Removal Project test cells: STA optimization status of the research at the north site. Water Sci. Technol. 44(1112), 117. Newman, S., and Pietro, K. (2001). Phosphorus st orage and release in response to flooding: implications for Everglades stormwater treatment areas. Ecol. Eng. 18, 23. Nick Miller, Inc. (2006). STA 5 vegetation map. (Nov. 17, 2006). Noe, G. B., and Childers, D. L. (2007). Phosph orus budgets in Everglades wetland ecosystems: the effects of hydrology and nutrient enrichment. Wetlands Ecol. Manage. 15, 189. Nungesser, M. K., and Chimney, M. J. (2001). Evaluation of phosphorus retention in a South Florida treatment wetland. Water Sci. Technol. 44(11-12), 109. Pant, H. K., and Reddy, K. R. (2001). Hydrologi c influence on stability of organic phosphorus in wetland detritus. J. Environ. Qual. 30(2), 668. Parrish, D. M., and Huebner, R. S. (2004). Water budget analysis for Stormwater Treatment Area 5 (May 1, 2000 to April 30, 2003) Techn. Publ. EMA #418, South Florida Water Management District (SFWMD), West Palm Beach, Fl. Persson, J. (2000). The hydraulic perf ormance of ponds of various layouts. Urban Water 2, 243. Persson, J., Somes, N. L. G., and Wong, T. H. F. (1999). Hydraulics effi ciency of constructed wetlands and ponds. Water Sci. Technol. 40(3), 291. Persson, J., and Wittgren, H. B. (2003). H ow hydrological and hydrau lic conditions affect performance of ponds. Ecol. Eng. 21, 259. Pietro, K., Bearzotti, R., Chimney, M., Germain, G., Iricanin, N., Piccone, T., and Samfilippo, K. (2006). Chapter 4: STA performance, compliance, and optimization In: 2006 South Florida Environmental Report South Florida Water Management District (SFWMD), West Palm Beach, Fl. Pietro, K., Bearzotti, R., Chimney, M., Germ ain, G., Iricanin, N., and Piccone, T. (2007). Chapter 5: STA performance, compliance, and optimization In: 2007 South Florida

PAGE 237

237 Environmental Report South Florida Water Management District (SFWMD), West Palm Beach, Fl. Pietro, K. C., Chimney, M. J., and Steinm an, A. D. (2006). Phosphorus removal by the Ceratophyllum /periphyton complex in a South Fl orida (USA) freshwater marsh. Ecol. Eng. 27, 290. Poiani, K. A., and Johnson, W. C. (1993). A spatial simulation model of hydrology and vegetation dynamics in semi-permanent prairie wetlands. Ecol. Appl. 3(2), 279. Post, Buckley, Schuh & Jernigan, Inc. (1985). Construction plans for Or lando easterly wetlands reclamation wildlife management and conservation area Qi, C., and Grunwald, S. (2005). GIS-based hydrologic modeling in the Sandusky Watershed using SWAT. Trans. ASAE 48(1), 169. Qore, Inc. (2004). Appendix 4B-1: Scope of work for inte nsive spatial monitoring of sediment, vegetation, and water quality in th e Stormwater Treatment Areas In: 2004 Everglades Consolidated Report South Florida Water Management District (SFWMD), West Palm Beach, Fl. Raghunathan, R., Slawecki, T., Fontaine, T. D., Chen Z. Q., Dilks, D. W., Bierman, V. J., and Wade, S. (2001). Exploring the dynamics and fate of total phosphorus in the Florida Everglades using a calibrated mass balance model. Ecol. Model. 142(3), 247. Reddy, K. R., DeLaune, R. D., DeBusk, W. F ., and Koch, M. S. (1993). Long-term nutrient accumulation rates in the Everglades. Soil Sci. Soc. Am. J. 57, 1147. Reddy, K. R., Diaz, O. A., Scinto, L. J., a nd Agami, M. (1995). Phosphorus dynamics in selected wetlands and streams of the lake Okeechobee Basin. Ecol. Eng. 5, 183. Reddy, K. R., Fisher, M. M., and Ivanoff, D. (1996). Resuspension and diffusive flux of nitrogen and phosphorus in a hypereutrophic lake. J. Environ. Qual. 25, 363. Reddy, K. R., Kadlec, R. H., Flaig, E., and Gale P. M. (1999). Phosphorus retention in streams and wetlands: a review. Crit. Rev. Environ. Sci. Technol. 29(1), 83. Reddy, K. R., OConnor, G. A., and Gale, P. M. (1998a). Phosphorus sorption capacities of wetland soils and stream sediment s impacted by dairy effluent. J. Environ. Qual. 27, 438. Reddy, K. R., Wang, Y., DeBusk, W. F., Fisher, M. M., and Newman, S. (1998b). Forms of soil phosphorus in selected hydrologic un its of the Florida everglades. Soil Sci. Soc. Am. J. 62(4), 1134.

PAGE 238

238 Redfield, G. W. (1998). Quantifying atmospheric deposition of phosphorus: A conceptual model and literature review for environmental management Techn. Publ. WRE #60, South Florida Water Management District West Palm Beach, Fl, 35 pp. Restrepo, J. I., Montoya, A. M., and Obeyseke ra, J. (1998). A wetland simulation module for the MODFLOW ground water model. Ground Water 36(5), 764. Richardson, C. J., and Craft, C. B. (1993). Effective phosphorus retenti on in wetlands: fact or fiction? Chapter 28 In: Moshiri, G. A. (Ed.), Constructed wetlands for water quality improvement CRC Press, Inc., Boca Raton, Fl. Rodgers, J. H., McKevitt, M. E., Hammerlund, D. O., Dickson, K. L., and Cairns, J. (1983). Primary production and decomposition of s ubmergent and emergent aquatic plants of two Appalachian rivers In: Fontaine, T. D., and Bartell, S. M. (Ed.), Dynamics of lotic systems Ann Arbor Science, Ann Arbor, Mi, 283. Rung, M., and Olesen, K. W. (2003). Com bined 1and 2-dimensional flood modeling. 4th Iranian Hydraulic Conference October 21-23, 2003, Shiraz, Iran. Sand-Jensen, K., and Borum, J. (1991). Int eractions among phytoplankton, periphyton, and macrophytes in temperate fresh-waters and estuaries. Aquat. Bot. 41(1-3), 137. Sardin, M., Schweich, D., Leij, F. L., a nd van Genuchten, M. Th. (1991). Modeling the nonequilibrium transport of linearly interac ting solutes in porous media: A review. Water Resour. Res. 27, 2287. Schaafsma, J. A., Baldwin, A. H., and Streb, C. A. (2000). An evaluation of a constructed wetland to treat wastewater from a dairy farm in Maryland, USA. Ecol. Eng. 14, 199 206. Scinto, L. J., and Reddy, K. R. (2003). Biotic and abiotic uptake of phosphorus by periphyton in a subtropical freshwater wetland. Aquat. Bot. 77, 203. Smith, M. B. (1993). A GIS-based distributed parameter hydrologic model for urban areas. Hydrol. Process. 7(1), 45. Sndergaard, M., Jensen, J. P., and Jeppesen, E. (2001). Retention and internal loading of phosphorus in shallow, eutrophic lakes. TheScientificWorld 1, 427. South Florida Water Management District (SFWMD). (2000). Operation plan Stormwater Treatment Area 5 (Revised) SFWMD, West Palm Beach, Fl. South Florida Water Management District (SFWMD). (2005a). Documentation of the Natural System Model (NSM) version 4.2. SFWMD, West Palm Beach, Fl. (May 5, 2007).

PAGE 239

239 South Florida Water Management District (SFWMD). (2005b). Final documentation for the South Florida Water Management Model ( SFWMM) v5.5. SFWMD, West Palm Beach, Fl. (May 18, 2007). Su, M., Stolte, W. J., and van der Kamp, G. (2000). Modelling Ca nadian prairie wetland hydrology using a semi-distri buted streamflow model. Hydrol. Process. 14, 2405. Sutron Corp. (2005). STA-5 2-D Hydraulic modeling (Linked cells model) Task 4.3 Final report, South Florida Water Management District, West Palm Beach, Fl. Thackston, E. L., Shields Jr., F. D., and Schroede r, P. R. (1987). Residence time distributions of shallow basins. J. Environ. Eng-ASCE 113, 1319. Tsanis, I. K., and Boyle, S. (2001). A 2D hydrodynamic/pollutant transport GIS model. Adv. Eng. Softw. 32, 353. Turner, B. L., Newman, S., and Newman, J. M. (2006). Organic phosphorus sequestration in subtropical treatment wetlands. Environ. Sci. Technol. 40, 727. van der Peijl, M. J., and Verhoeven, J. T. A. (1999). A model of carbon, nitrogen, and phosphorus dynamics and their interacti ons in river marginal wetlands. Ecol. Model. 118, 95. Walker, W. W. (1993). A mass-balance model for estimati ng phosphorus settling rate in Everglades Water Conservation Area 2A Report to U.S. Department of Justice, Environment and Natura l Resources Division. Walker, W. W. (1995). Design basis for Ev erglades Stormwater Treatment Areas. Water Resour. Bull. 31(4), 671. Walton, R., Chapman, R. S., and Davis, J. E. (1996). Development and application of the wetlands dynamic water budget model. Wetlands 16(3), 347. Wang, H., Appan, A., and Gulliver, J. S. (2003a ). Modeling of phosphorus dynamics in aquatic sediments: I-model development. Water Res. 37, 3928. Wang, H., Appan, A., and Gulliver, J. S. ( 2003b). Modeling of phosphorus dynamics in aquatic sediments: II-examination of model performance. Water Res. 37, 3939. Wang, N., and Mitsch, W. J. (2000). A detail ed ecosystem model of phosphorus dynamics in created riparian wetlands. Ecol. Model. 126, 101. Wantman Group, Inc. (2005). Stormwater Treatment Area 5 (STA 5) topographic survey

PAGE 240

240 Warwick, J. J., and Haness, S. J. (1994). Efficiency of ARC/INFO GIS application to hydrologic modeling. J. Water Resour. Plan. Manage-ASCE 120(3), 366. Werner, T. M., and Kadlec, R. H. (1996). Appl ication of residence time distributions to stormwater treatment systems. Ecol. Eng. 7, 213. Werner, T. M., and Kadlec, R. H. (2000). W etland residence time distribution modeling. Ecol. Eng. 15, 77. White, J. R., Reddy, K. R., DeBusk, W. F., Wise, W. R., and Crisman, T. L. (2002). Phosphorus removal capacity of the Orlando Easterly Wetland treatment system Final report to the City of Orlando, March 2002. White, J. R., Reddy, K. R., and Moustafa, M. Z. (2004). Influence of hydrologic regime and vegetation on phosphorus retention in Everglad es stormwater treatment area wetlands. Hydrol. Process. 18(2), 343. White, J. R., Reddy, K. R., and Newman, J. (200 6). Hydrologic and vegetative effects on water column phosphorus in wetland mesocosms. Soil Sci. Soc. Am. J. 70(4), 1242. Wilsnack, M. M., Welter, D. E., Montoya, A. M ., Restrepo, J. I., and Obeysekera, J. (2001). Simulation flow in regional wetlands with the MODFLOW wetlands package. J. Am. Water Resour. Assoc. 37(3), 655. Wrman, A., and Kronns, V. (2005). Effect of pond shape and vegetation heterogeneity on flow and treatment performance of constructed wetlands. J. Hydrol. 301, 123. Xu, Z. X., Ito, K., Schultz, G. A., and Li, J. Y. (2001). Integrated hydrologic modeling and GIS in water resources management. J. Comput. Civil Eng. 15(3), 217.

PAGE 241

241 BIOGRAPHICAL SKETCH Joong Hyuk Min was born on October 3, 1974 in Se oul, Korea. The eldest of two sons, he grew up mostly in Seoul, graduating from Dae -Il Foreign Language High School in 1993. He earned his Bachelor of Science in Earth and Envi ronmental Sciences and his Master of Science in Geochemistry from Korea University in 1999 and 2002, respectivel y. During his master program, he studied a variety of areas in hydrogeochemistry, mainly conducting groundwater quality assessment and interpretation on the impacts of agricultural activities on shallow aquifers. In August 2003, he began his doctoral st udy in the Department of Environmental Engineering Sciences at the Univ ersity of Florida in Gainesville as an alumni fellow. He has pursued a Ph.D. degree under the guidance of Prof William R. Wise and mainly conducted 2-D modeling studies for flow dynamics, solute trans port, and phosphorus dynamics in constructed wetlands. He earned his Doctor of Philosophy in Hydrologic Sciences from the University of Florida in August 2007. Upon completion of his Ph.D. program, Joong Hyuk plans to work as a post-doctoral research associate at the University of Louisian a/U.S. Fish and Wildlife Service in Lafayette, Louisiana. Joong Hyuk has been married to Moon Jung for 4 years. They have a son, Toby, age 1.5.