‘ . “flan?“ an. f. _:‘ 0‘ 551mg.“ qr. .. . I: r 11414.2; ,. Hr“ n~ . .r Ir . 134...“.- rain :flflfldu; a: W .1392? En. .HRU: 2 LIBRARY 6) a: Michigan State 577 7 ‘7‘ 5/92 University This is to certify that the thesis entitled EFFECT OF SMALL-SCALE HETEROGENEITIES ON TRACER TRANSPORT presented by MICHAEL A. BRENNAN has been accepted towards fulfillment of the requirements for the Master of degree in Environmental Geosciences Science KB 40?) MM Maia Professor’s Signature 5/3/07 Date MSU is an Affirmative Action/Equal Opportunity Institution PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 6/01 c:/CIRC/DateDue.p65—p. 1 5 EFFECT OF SMA EFFECT OF SMALL-SCALE HETEROGENEITIES ON TRACER TRANSPORT By Michael A. Brennan A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Geological Sciences 2004 EFFECT 0F 5” A large cc neterogeneous a nerhern East Le of hydraulic con: series of tracer t measurements. this project and the tritium tracer to molecular ma 'ates, as descrit aquifer sedimen D‘SiOsity enablin Comm model The nUlherical s Experimem l in Vv’i L‘ ’ ABSTRACT EFFECT OF SMALL-SCALE HETEROGENEITIES ON TRACER TRANSPORT BY Michael A. Brennan A large column (108 cm in length with an inner diameter of 45 cm) of heterogeneous aquifer sediment collected from a shallow glacial drift deposit in northern East Lansing, MI was characterized to obtain high-resolution estimates of hydraulic conductivity. The characterization was accomplished through a series of tracer transport experiments, visual observations, and physical measurements. Bromide, fluorescein and tritium were used as the tracers for this project and all displayed asymmetrical breakthrough curves (BTC’s). Also, the tritium tracer test had a slightly delayed BTC peak time which was attributed to molecular mass differences of the tracers with associated differential diffusion rates, as described by Ewers (1997). The physical measurements taken of the aquifer sediment provided high-resolution estimates of hydraulic conductivity and porosity enabling a more accurate groundwater flow and solute transport computer model to be created than was possible prior to this characterization. The numerical simulations allowed for the dominant physical processes (i.e. advection and diffusion) to be studied in a carefully controlled laboratory experiment in which the boundary conditions are known. Specifically, this study focuses on the effect of centimeter to decimeter scale heterogeneities on the flow and transport of conservative tracers. Dedic Dedicated to my wife Lisa, daughter Delia and son Patrick. iii limost sincere and t members Gary WeiS ran-stop help and SL lrlrs. Florence Hill an 1: Nathaniel Ostrom. for the use of their la e through the yea any age the meanir mother, may she res :Elar of hard work m ACKNOWLEDGEMENTS A most sincere and eternal thanks to my advisor Dave Hyndman, my committee members Gary Weissmann and Grahame Larson, and Loretta Knutson for their non-stop help and support. Thanks to the Department of Geological Sciences, Mrs. Florence Hill and the Central Lapidary Society for their financial support and to Nathaniel Ostrom, Peggy Ostrom, Will Kopacik, Randy Schaetzl, Xhanda Zhao for the use of their laboratory and equipment. To my sisters, for their support and love through the years. To my father, may he rest in peace, for showing my at an early age the meaning of hard work. Finally, a very special thank you to my mother, may she rest in peace, whose constant support, many sacrifices and pillar of hard work made this goal attainable. iv LISTOFTABLES usror FIGURES. TRODUCTION ..... INBACKGROUND General- ........... GEOLOGY ........... Surficial Geolog Sedimentclogy- METHODS ............. MESOCOSM DEE TRACER............. DISASSEMBLY .. SEDIMENT CORI HYDRAULIC COT Slug Tests- ...... Falling Head Pg Constant Head Kozeny-Carmer EFFECTIVE POP GRAIN SIZE ....... DIGITAL IMAGIN COMPUTER MOI RESULTS’DISCUS. TRACER ............. Tiert (Wells 2- Tier2 (Wells 6~ Tler3 (Samplin Summary ......... DISASSEMBLY .. TABLE OF CONTENTS LIST OF TABLES ................................................................................................ vii LIST OF FIGURES ............................................................................................... ix INTRODUCTION .................................................................................................. 1 BACKGROUND ................................................................................................ 5 General- ........................................................................................................ 5 GEOLOGY ........................................................................................................ 8 Surficial Geology- .......................................................................................... 8 Sedimentology- .................................................. 9 METHODS ......................................................................................................... 11 MESOCOSM DESIGN .................................................................................... 11 TRACER ......................................................................................................... 14 DISASSEMBLY .............................................................................................. 20 SEDIMENT CORING ...................................................................................... 20 HYDRAULIC CONDUCTIVITY 23 Slug Tests- .................................................................................................. 23 Falling Head Permeameter Tests on Small Sediment Cores- ..................... 23 Constant Head Permeameter Tests on Small Sediment Cores- ................. 24 Kozeny-Carmen Equation- .......................................................................... 29 EFFECTIVE POROSITY ................................................................................. 30 GRAIN SIZE ................................................................................................... 31 DIGITAL IMAGING ......................................................................................... 33 COMPUTER MODEL ...................................................................................... 33 RESULTS/DISCUSSION .................................................................................... 38 TRACER ......................................................................................................... 38 Tier 1 (Wells 2-5) ......................................................................................... 45 Tier 2 (Wells 6-10) ....................................................................................... 46 Tier 3 (Sampling Ports 11-15 and Outflow) ................................................. 47 Summary ..................................................................................................... 47 DISASSEMBLY .............................................................................................. 57 HYDRAULIC CONDUCTIVITY ....................................................................... 59 EFFECTIVE POROSITY ................................................................................. 63 GRAIN SIZE ................................................................................................... 64 SEDIMENTOLOGY ......................................................................................... 65 Clayey Silt Unit ............................................................................................ 67 Silty Clay with Trace of Fine Gravel Unit ..................................................... 67 Fine Silty Sand Unit ..................................................................................... 68 Fine to Very Fine Sand Unit ........................................................................ 68 Fine to Medium Sand Unit ........................................................................... 69 COMPUTER SIML SUIAIAARY AND CO APPENDIX A. Sam; APPENDIX 8. Well ( APPENDIX C. FIUOr APPENDIX D. Fluor APPENDIX E. FaIIin APPENDIX F. Cons‘ APPENDIX G. PorO' APPENDIX H. Grair APPENDIXI. Cross APPENDIX J. Mode BZBLIOGRAPHY ..... TABLE OF CONTENTS (CONT’D) COMPUTER SIMULATIONS ...................................................... 70 SU MMARY AND CONCLUSIONS ..................................................................... 82 APPENDIX A. Sample Locations and Lithologic Interpretations ........................ 85 APPENDIX B. Well Casing Volume Calculation ................................................ 96 APPENDIX C. Fluorescein Calibration Curve .................................................... 98 APPENDIX D. Fluorescein BTC’S ................................................................... 101 APPENDIX E. Falling Head Permeameter Data .............................................. 110 APPENDIX F. Constant Head Permemeter Data ............................................ 119 APPENDIX G. Porosity Data .......................................................................... 123 APPENDIX H. Grain Size Data ....................................................................... 124 APPENDIX l. Cross-Sections of Final Lithologies and Model Grid .................. 138 APP ENDIX J. Model Simulation Sensitivities .................................................. 143 BIBLIOGRAPHY ............................................................................................... 144 vi Table 1. Well and 5 Table 2. Samplingl rab'ze 3. Sample Nc Tab’eA. Sediments Tables. Summary .‘E.orescein mass CO 'esults (see earlier c Tania 6. Sampling I Table 7. Summary. Table 8. Water VOIL Table 9. Slug Test l TatrletO. Kozeny-c . Constant “379 12- Hydraulic GOIEIB. SUmman' J able14 S . ummaw 7' H eels. General F rabieie. Final MOC Tablet] easur [FIR-2d er wasw ' m. w e.e subtraN *iIeAt umrr .gQun ' 'a'h .d in later TqLi cue LIST OF TABLES Table 1. Well and Sampling Port Inlet Locations. .............................................. 13 Table 2. Sampling Frequencies During the Tracer Test. ................................... 17 Table 3. Sample Nomenclature ......................................................................... 22 Table 4. Sedimentation Times of 2 micron particles .......................................... 33 Table 5. Summary of Tracer Tests and Mass Conservation. * denotes fluorescein mass conserved, bromide mass was not calculated due to erratic results (see earlier discussion) ........................................................................... 39 Table 6. Sampling Point Lithologic Descriptions ................................................ 45 Table 7. Summary Of Tracer Peak Arrival Times and Concentrations ............... 48 Table 8. Water Volume Extracted From Mesocosm .......................................... 58 Table 9. Slug Test Results ................................................................................ 60 Table 10. Kozeny-Carmen: Summary of Hydraulic Conductivity Estimates ...... 61 Table 11. Constant Head: Summary of Hydraulic Conductivity Estimates ........ 62 Table 12. Hydraulic Conductivity Comparisons ................................................. 62 Table 13. Summary Of Effective Porosity Measurements .................................. 64 Table 14. Summary of Grain Size ..................................................................... 65 Table 15. General Physical Descriptions of Sediment ....................................... 67 Table 16. Final Model Parameters .................................................................... 71 Table 17. Measured Heads. Taken from Ewers, 1997. * denotes simulated heads were subtracted from 170 cm as this was the constant head used at the Oufilow ................................................................................................................ 72 Table A-1. Summary of Sampling Analysis. * denotes method used was Improved on in later tests ................................................................................... 94 Table 0-1. Fluorescein Calibration Curve Data ................................................. 98 vii TableE-t. Falling HI TasIeF-i. Constant Vaa’eG-I. Porosity I Table H-l. Summa' LIST OF TABLES (CONT'D) Table E-1. Falling Head Permeameter Data .................................................... 110 Table F-1. Constant Head Permeameter Data ................................................ 119 Table G-1. Porosity Data ................................................................................. 123 Table H-1. Summary of Grain Size Percentages ............................................. 136 viii Poure 1. ApnroxlmE :risenled in color) -- dare 2. Mesocosr" tell locations at the | left and right. The TI s‘the plates used to trait; in place with i‘. re outer holes. As PVC pipe against TC to prevent leakage a heft-teen the mesoc: A denotes well and dimensions along v. coordinate data wer~ F‘gure 3. Surficial G Fgure 4. Typical Cr =tastiers) denote we cores were taken. type named. Z-dim nnq yfi'de. ‘igure 5. Close-Up r rigure 6. Sediment Tigure 7. erection Schemati LIST OF FIGURES Fig ure 1. Approximate Sample Collection Location. (Images in this thesis are presented in color) ................................................................................................ 6 Fig ure 2. Mesocosm design. Side view with the PVC tube in the center, the we ll locations at the top, sampling ports at the far right and the end plates at far left and right. The hatched areas of these endplates indicate recessed portions of the plates used to create constant head boundaries. The end plates are held firmly in place with four tie rods (not shown) attached to the end plates through the Outer holes. As the tie rods were tightened, the end plates were sealed to the PVC pipe. against rubber gaskets. The gray lines indicate the rubber seals used to prevent leakage at the ends. Stainless steel screens (not Shown) were placed between the mesocosm end plates and the sediment face inside the PVC tube. W denotes well and SP denotes sampling port. Lower left diagram shows dimensions along with approximate reference point location where x, y and z coordinate data were obtained in GMS version 4.0 .............................................. 7 Figure 3. Surficial Geology Deposits. (Source: Puzio and Larson, 1982) ........... 9 Figure 4. Typical Cross-Section of Mesocosm Layer. Small circles (metal Washers) denote well screen location and larger holes are locations where SOil cores were taken. Lines represent interpreted lithologic boundary with the unit tYpe named. Z-dimension is looking into the figure with flow coming out of the page- .................................................................................................................... 7 FIQU re 5. Close-Up of Outflow Sampling Locations. ........................................... 14 F‘QU re 6. Sediment Core and Constant Head Attachment Assembly ................ 27 FIQU re 7. Schematic of Constant Head Set-Up. Small arrows denote water flow direction .............................................................................................................. 29 FIQU re 8. Model Grid .......................................................................................... 35 FISH re 9. Model Grid and Typical Lithology Overlay .......................................... 36 FIQUre 10. Fluorescein Tracer Mass Recovery. August 1998 Fluorescein data shDWn ................................................................................................................. 39 ix figure II- T Awrescein ‘ are not SOON .‘iQUTe I2 ear resents C :inl BSCC TI September Figure13. T Sampling Pt concentratlc data. and Fl Figure 14. ' coitcentratic Autust t199l data taken I ampepon Fr gure 15 .iashers) de Wes were Tillie namec Sane “By P. retire 16. SIIUIW on H lhe media LIST OF FIGURES (CONT’D) Fig ure 11. Tracer Concentration Histories (Well 2 through Well 5). 0/00 rep resents concentrations are normalized by injection concentration (100 ppm). Flu orescein August 1996 data, Tritium January 1997 data, and Fluorescein September 1997 data taken from Ewers (1997). Outliers at Well 4 and Well 5 are not shown ..................................................................................................... 41 Fig ure 12. Tracer Concentration Histories (Well 6 through Well 10). C/Co rep resents concentrations are normalized by injection concentration (100 ppm). F|uorescein August 1996 data, Tritium January 1997 data, and Fluorescein September 1997 data taken from Ewers (1997) ................................................. 42 Figure 13. Tracer Concentration Histories (Sampling Port 11 through Sampling Port 15). C/Co represents concentrations are normalized by injection concentration (100 ppm). Fluorescein August 1996 data, Tritium January 1997 data, and Fluorescein September 1997 data taken from Ewers (1997) ............. 43 Figure 14. Tracer Concentration Histories (Outflow). C/Co represents concentrations are normalized by injection concentration (100 ppm). Fluorescein August 1996 data, Tritium January 1997 data, and Fluorescein September 1997 data taken from Ewers (1997). The outflow represents the combined flow Of sample port 11 through sample port 15 (lower right diagram Of this figure) ........ 44 I:igure 15. Clay Cracks Observed in Outflow Layer. Small circles (metal Washers) denote well screen location and larger holes are locations where soil cores were taken. Lines represent interpreted lithologic boundary with the unit type named. Z-dimension is looking into the figure with flow coming out'of the 93963 ................................................................................................................... 53 FIQLI re 16. Diffusive Retardation. (Ewers, 1997) Three identical particles are Shown on the left and three different identical particles are shown on the right. The media on both sides is identical with the highly conductive layer between two less conductive layers. From top to bottom 5 progressive time steps are rep resented. Notice in time step 2 on the left side that the concentration gradient cau'S-ed the leading particle to exit the highly conductive layer and become 'dormant’ in the less conductive surroundings. Only in time step 5 is the particle Te‘eased back into the highly conductive layer. Also, notice that the non-diffusive lTaCer on the right is exiting in time step 5, while the diffusive tracer is still at least ‘2 Steps behind .................................................................................................... 55 Figure 17. Volume Drained Water ..................................................................... 59 Figure 18. Histogram of Kozeny-Carmen Hydraulic Conductivity Estimates. Plot on the left is fine to very fine sand samples and plot on the right is silty sand samples .............................................................................................................. 61 Figure 19- HIS! on the left is fin samples .......... Figure 20. DIQI lithologic bOUOC Figure 21. MOC Figure 22. Sim {Wells 25) ...... Figure 23. SimI lili’ells 6-1 0) Figure 24. SimI {Sampling Port. Figure 25. SW (Outflow) ......... Rgure A-l. Dig oenote sedimen Screen location. ‘It'Iologic bound; Figure A-2. Digi location. Lines I and underlined. figureAO. i' CCation, ' gl LIST OF FIGURES (CONT’D) Figure 19. Histogram of Constant Head Hydraulic Conductivity Estimates. Plot on the left is fine to very fine sand samples and plot on the right is silty sand samples .............................................................................................................. 62 P i gure 20. Digital Image of Layer N (Inflow End). Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted |ithologic boundary with the unit type named and underlined ............................. 66 F-' igure 21. Model Statistics ................................................................................ 73 Figure 22. Simulated and Observed Tracer Concentration Histories (Wells 2-5) ......................................................................................................... 76 Figure 23. Simulated and Observed Tracer Concentration Histories (Wells 6-10) ....................................................................................................... 77 Figure 24. Simulated and Observed Tracer Concentration Histories (Sampling Ports 11 through 15) ......................................................................... 78 Figure 25. Simulated and Observed Tracer Concentration Histories (Outflow) ............................................................................................................ 79 Figure A-1. Digital Image of Layer A (Outflow End). Medium sized circles denote sediment core sample location and smaller circles (washers) denoted well screen location. Each well is labeled as “W-#”. Lines represent interpreted lithologic boundary with the unit type named and underlined ............................. 86 Fig u re A-2. Digital Image of Layer B. Circles denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named an d underlined .................................................................................................... 86 Fig u re A-3. Digital Image of Layer C. Circles denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined .................................................................................................... 87 Figure A-4. Digital Image of Layer D. Circles (plastic cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the umt type named and underlined ......................................................................... 87 Figure A-5. Digital Image of Layer E. Large and medium sized circles (plastic and aluminum cores) denote sediment core sample location and smaller circles (washers) denoted well screen location. Each well is labeled as “W-#”. Lines represent interpreted lithologic boundary with the unit type named and underlined ........................................................................................................... 88 xi Figure A-6 denote set boundary I Figure A-7 denote set boundary I Figure A-8 cenote set boundary \ Figure A-9 denote sec toundary I Figure At no alumir ih‘ashers) Tepresent i Flgure A-t CSROIB SEC boundary \ figure A4 L, doundary \ Figure A4 denot LIST OF FIGURES (CONT’D) F—" igure A-6. Digital Image of Layer F. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined ............................................ 88 F: igure A-7. Digital Image of Layer G. Circles (plastic and aluminum cores) d enote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined ............................................ 89 F: igure A-8. Digital Image Of Layer H. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined ............................................ 89 Figure A-9. Digital Image of Layer I. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined ............................................ 90 Figure A-10. Digital Image of Layer J. Large and medium sized circles (plastic and aluminum cores) denote sediment core sample location and smaller circles (washers) denoted well screen location. Each well is labeled as “W-#”. Lines represent interpreted lithologic boundary with the unit type named and underlined - - - . . ...................................................................................................................... 90 -Figure A-11. Digital Image of Layer K. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined ............................................ 91 F5 9 we A-12. Digital Image of Layer L. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic bQundary with the unit type named and underlined ............................................ 91 Fig ure A-13. Digital Image of Layer M. Circles (plastic and aluminum cores) de note sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined ............................................ 92 Fig ure A-14. Digital Image Of Layer N (Inflow End). Circles (plastic and é‘uminum cores) denote sediment core sample location. Lines represent “\terpreted lithologic boundary with the unit type named and underlined ........... 92 Figure A-15. Index Figure to Cross-Section Images. Note that this view Tepresents the position of the mesocosm after all the tracer tests were conducted. Tracer tests were conducted with the mesocosm parallel to the ground, and disassembly and sediment coring were conducted with‘the mesocosm perpendicular to the ground (as shown) ........................................... 93 xii Figure 01- FW Figure 02 Hum meCG-Fmor PWEDJ.W@W Figure [3.2. Well- Figure 0-3, Well-i Figure [34- WEI": Figure DFS. WEH’I Figure 0-6. Well-T Rgure 0.7. Well-E FgureD-B. Well-E Figure D-9. Well-1 meDdO.Sam FIgure D1 1. Sam Figure 0-1. Fi gure C-2. Fi gure C-3. Figure 0-1. Figure D-2. Figure 0-3. Figure D-4. Figure D-5. Figure D-6. Figure D-7. Figure D-8. Fig u re D-9. Figure D-10. Fig u re D-11. Figu re 0-12. FiQu re D-13. Figu re D-14. FIQ u re D-15. FIQU re H-1. LIST OF FIGURES (CONT’D) Fluorescein Calibration Curve Plot (0-10 ppm) ............................... 98 Fluorescein Calibration Curve Plot (5-25 ppm) ............................... 99 Fluorescein Calibration Curve Plot (20-100 ppm) ........................... 99 Well-2: Fluorescein BTC. 1996 data ............................................ 101 Well-3: Fluorescein BTC. 1996 data ............................................ 101 Well-4: Fluorescein BTC. 1996 data ............................................ 102 Well-5: Fluorescein BTC. 1996 data ............................................ 102 Well-6: Fluorescein BTC. 1996 data ............................................ 103 Well-7: Fluorescein BTC. 1996 data ............................................ 103 Well-8: Fluorescein BTC. 1996 data ............................................ 104 Well-9: Fluorescein BTC. 1996 data ............................................ 104 Well-10: Fluorescein BTC. 1996 data .......................................... 105 Sampling Port 11: Fluorescein BTC. 1996 data ......................... 105 Sampling Port 12: Fluorescein BTC. 1996 data ......................... 106 Sampling Port 13: Fluorescein BTC. 1996 data ......................... 106 Sampling Port 14: Fluorescein BTC. 1996 data ......................... 107 Sampling Port 15: Fluorescein BTC. 1996 data ......................... 107 Outflow: Fluorescein BTC. 1996 data ........................................ 108 Sample A3a1 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 125 Figure H-2. grain size Figure H-3. grain size Sample A3a2 Grain Size. Dashed line indicates inferred median ....................................................................................................... 125 Sample A3c3 Grain Size. Dashed line indicates inferred median ....................................................................................................... 126 xiii Figure H-4. Sam gdnshe ............ Flgure H-5. Sam gansue ............ Figure H-6. 58ml gamsue ............ Fgure H-7. Saml gamsue ............. RyeH8.Samp gamsue ............. FgmH9.Samp gansue ............. RgdeO.Sam gansue .............. FyeH41.SamI gamsue .............. F‘- '5. i :C V We H‘I2- Sam; ga ”Sue ......... Fayre H43. Sam; Shhsue ...... LIST OF FIGURES (CONT’D) Figure H-4. Sample A2b2 Grain Size. Dashed line indicates inferred median grain Size .......................................................................................................... 126 Figure H-5. Sample Eb1 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 127 Figure H-6. Sample Eb2 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 127 Figure H-7. Sample Eb3 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 128 Figure H- 8. Sample Eb4 Grain Size. Dashed line indicates inferred median grain Size .......................................................................................................... 128 Figure H-9. Sample D1c Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 129 Figure H-10. Sample D2c Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 129 Figure H-11. Sample D1b Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 130 Figure H-12. Sample Gb1 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 130 Figure H-13. Sample Gb2 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 131 Figure H-14. Sample Gb3 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 131 Figure H-15. Sample Hb1 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 132 Figure H-16. Sample Hb2 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 132 Figure H-17. Sample lb1 Grain Size. Dashed line indicates inferred median grain Size .......................................................................................................... 133 Figure H-18. Sample |b3 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 133 xiv Figure H-19. Sa grain size .......... Figure H-20. Sa grain size ........... Figure H-21. Sa grain size ........... Figure H-22. Sa grain size ........... Figure I-l. Three Figure I-2. Thin Figure l-3. TOTEE Figure I-4. Three Figure J-l. Sum dspersivity value medrum sand (0. open area (0.2), values are a” GQI Parameter by On ( LIST OF FIGURES (CONT’D) Figure H-19. Sample Jc1 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 134 Figure H-20. Sample J02 Grain Size. Dashed line indicates inferred median grain Size .......................................................................................................... 134 Figure H-21. Sample Kb1 Grain Size. Dashed line indicates inferred median grain size. ......................................................................................................... 135 Figure H-22. Sample K02 Grain Size. Dashed line indicates inferred median grain size .......................................................................................................... 135 Figure l-1. Three-Dimensional Cross-Section #1 ............................................ 138 Figure I-2. Three-Dimensional Cross-Section #2 ............................................ 139 Figure l-3. Three-Dimensional Cross-Section #3 ............................................ 140 Figure l—4. Three-Dimensional Cross-Section #4 ............................................ 141 Figure J-1. Summary of Model Simulation Sensitivities. Final model lithologic dispersivity values are as follows (units Of cm/day): silty sand (1.0), fine to medium sand (0.2), very fine to fine sand (0.2), clayey silt (1 .0), silty clay (0.2), open area (0.2), and parting (0.7). Final model lithologic molecular diffusion values are all equal to 1.0 x 10’6. Arrows represent increasing or decreasing - parameter by one order of magnitude .............................................................. 143 XV Numerous that heterogeneil ata wide range c and lilackay, 19E Alaskay et al., 19 2001; Berkowitz . elal., 1980; Pass Field-scale transport focus p at, 1992). Field- materials at Scale have been well d and are primarily be directly Used n reflection of the ir Scale increaSe s(l thservation increr studies include th« INTRODUCTION Numerous studies over the past several decades have focused on the role that heterogeneity has on solute transport. These studies have been conducted at a wide range of scales from the field-scale (Burr et al., 1994; Thorbjarnarson and Mackay, 1994; Hess et al., 1992; Gelhar et al., 1992; 80995 et al., 1992; Mackay et al., 1986) to the laboratory-scale (Silliman et al., 2001; Ursino et al., 2001; Berkowitz et al., 2000; Sugita and Gillham, 1995; Grisak et al., 1980; Rao et al., 1980; Passioura, 1971). Field-scale studies studying the influence of heterogeneity on solute transport focus primarily on scales ranging from 10 to 50,000 meters (Gelhar et al., 1992). Field-scale research can be done in minimally disturbed aquifer materials at scales that are applicable to “real” world concerns. Scaling issues have been well documented by Gelhar (1986) and Schulze-Makuch et al. (1999), and are primarily the reason that laboratory measurements of dispersivity cannot be directly used in field simulations. This “scale effect” is thought to be a reflection of the increase in natural heterogeneities that exist as the observation scale increases (Gelhar et al., 1992). Therefore, normally as the scale of observation increases so to do the dispersivity values. Notable field-scale studies include those conducted at the Borden site (Mackay et al., 1986; Thorbjarnarson and Mackay, 1994; Burr et al., 1994), the Cape Cod site (Hess et al., 1992; Garabedian et al., 1991), the MADE site in Mississippi (Boggs et al., 1992), and the Kesterson site (Benson, 1988). Research at these sites have provided detain“ the exception of Gelhar et al. (19' These tiel aquifers with var such field studiet meter scale hete can not effective accurately chara heterogeneities. discharge bound field studies (eg Next. and perha; adequate|y Chara l1986)stating, “s Unknowable Sma methods, in field. it t ' ' eterogeneities i: Additionally, the 1 (if fri ' s.bution of the limitations can b e Laboratory solute transport D provided detailed data sets to help quantify solute transport processes and, with the exception of the Kesterson site, were tabulated and critically reviewed by Gelhar et al. (1992) for their reliability. These field studies have generated a tremendous amount of data for aquifers with varying degrees of heterogeneities (Gelhar, et al. 1992), however such field studies are limited in their ability to characterize the influence of sub- meter scale heterogeneities on transport. Unfortunately, such field-scale studies can not effectively control the field hydraulic conditions, and can not be used to accurately characterize or evaluate the influence of smaller (sub-meter) scale heterogeneities. The inability to control boundary conditions (i.e. recharge or discharge boundaries) has complicated the conservation of tracer mass in some field studies (e.g., Webster, 1996; Mull et al., 1988; Hotzl and Werner, 1992). Next, and perhaps most importantly, the small-scale heterogeneities cannot be adequately characterized using field-scale methods. This is outlined by Gelhar (1986) stating, “stochastic approaches incorporate the effects of practically unknowable small-scale variations in flow...”. This common use of stochastic methods, in field-scale studies, to determine variations in the small-scale heterogeneities is a weakness that can be addressed at the laboratory scale. Additionally, the time and cost required to adequately characterize the spatial distribution of the tracer, makes these field studies rare. The only way that these limitations can be addressed is under laboratory conditions. Laboratory experiments generally concentrate on smaller scale (<1 meter) solute transport processes through the use of sediment columns with varying deglees ‘ and invo!‘ heterOQer Lindsllorr al. 1994? i Howeveh aquifefS- " (1969) ha‘ hydraulic ( Therefore. more extre when tryin Bes representir commonly used rninirr Passioura ‘ of the studii Conditions. As di “till the follt 990’ reDrese .L J‘ degrees of heterogeneity. Most of this work focused on homogeneous media and involved the use of hand packed glass spheres, ceramic spheres or mildly heterogeneous sands (Beard and Way 1971; Rose and Passioura 1971; Lindstrom and Boersma 1971; Rao et al. 1979; Hornberger et al. 1992; Saiers et al. 1994; Sugita and Gilham 1995; Hendry et al. 1999; Silliman et al. 2002). However, these works only evaluate solute transport in fairly homogeneous aquifers, which are relatively uncommon in nature. Greenkorn and Kessler (1969) have argued that if all geologic formations display spatial variations in hydraulic conductivity then there is no such thing as a homogeneous formation. Therefore, as the scale of interest increases, so to does the probability of finding more extreme heterogeneity within the aquifer. This presents many problems when trying to measure and evaluate solute transport within an aquifer. Besides the difficulty past laboratory studies have had in accurately representing geologic heterogeneities (Gelhar et al., 1992), they are also commonly limited to two dimensional analysis (Burr et al., 1994), have rarely used minimally disturbed samples, and have suffered from edge flow (Rose and Passioura 1971). All of these limitations result in data that are not characteristic of the studied porous media, and are an inadequate representation of field flow conditions. As discussed above, past solute transport studies have been conducted with the following weaknesses: poorly characterized small-scale heterogeneities, poor representation of the sub-meter scale, and the inability to effectively control boundary conditions. One approach created to address all of these documented weaknesses inv: heterogeneous 2 Specifically. this 1. Using samp to Stu 2. Using resoh 3. Colle 4.8c mesc horiz 4- EVali Oiai 5. Deve 9f0u aQuit TilisthE RENEW GGDOS weaknesses involves the use of a minimally disturbed, sub-meter, highly heterogeneous aquifer sample within a controlled laboratory environment. Specifically, this research involved the following: 1. Using sub-meter scale (0.5-1 meter), highly heterogeneous aquifer sample [108 centimeter (cm) in length with an inner diameter of 45 cm] to study solute transport. Using conservative tracers to generate BTC’s used to acquire high- resolution data. Collecting sediment cores (7.5 cm in length with an inner diameter of 4.8 cm and 7.5 cm in length with an inner diameter of 6 cm) from the mesocosm to estimate physical properties of the sediments, including horizontal hydraulic conductivity, porosity and grain size. Evaluating digital images of geologic units (used to aid in the creation of a three-dimensional model grid). Developing three-dimensional computer models to simulate groundwater flow and solute transport processes in the heterogeneous aquifer sample. This thesis will present this approach and will evaluate the influence of naturally deposited sub-meter scale heterogeneities on the flow and transport of conservative tracers. W General' In August sediments was c groundwater flow horn a shallow 9’ of Section 4, T4l‘ bottom of the sat fringe (Ewers 19 table it appearec the water table a Caterpillar 375L ‘w‘th a sharpener the collection tut sediment with m each end to seal before it was trai University (MSU BACKGROUND General- In August of 1996, an intact sediment sample of highly heterogeneous sediments was collected to study the effects of small-scale heterogeneity on groundwater flow and solute transport properties. The sample was collected from a shallow glacial drift deposit in northern East Lansing, MI (NW 1A of NW 1A of Section 4, T4N, R1W) at a depth of approximately four meters (Figure 1). The bottom of the sample was approximately 20 cm above the dry season capillary fringe (Ewers 1997). Even though the sample was collected above the water table it appeared as if the sample came from the same geological formation as the water table aquifer (Dave Hyndman, personal communication). A large Caterpillar 375L excavator was used to push the collection tube (PVC sewer pipe with a sharpened tip) into the side of an excavation face (Ewers 1997). This filled the collection tube (108 cm in length with an inner diameter of 45 cm) with sediment with minimal compaction and disturbance. A PVC plate was fitted to each end to seal the collected sediments, hereafter called the “mesocosm”, before it was transported to the hydrogeology laboratory at Michigan State University (MSU). Figure 2 shows the final Mesocosm design. p517," ', I Source- MIDM? Michigan Geographic Lhrary APPROXIMATE COLLECHON SITE AREA 0 0.125 0.25 0.5 Miles Figure 1. Approximate Sample Collection Location. (Images in this thesis are presented in color). lnf ”SP-12 V End View Side View Direction of flow v k z - ‘\" /.::..... Reference Point (0.0) Figure 2. Mesocosm design (Ewers, 97). Side view with the PVC tube in the center, the well locations at the top, sampling ports at the far right and the end plates at far left and right. The hatched areas of these endplates indicate recessed portions of the plates used to create constant head boundaries. The end plates are held firmly in place with four tie rods (not shown) attached to the end plates through the outer holes. As the tie rods were tightened, the end plates were sealed to the PVC pipe against rubber gaskets. The gray lines indicate the rubber seals used to prevent leakage at the ends. Stainless steel screens (not shown) were placed between the mesocosm end plates and the sediment face inside the PVC tube. W denotes well and SP denotes sampling port. Lower left diagram shows dimensions along with approximate reference point location where x, y and z coordinate data were obtained in GMS version 4.0. At tl“. set at a COr gradient of in the mast screen area (Ewers 199 GEOLOGY Surf/1 The 5 shows the a sand enviror sand and sit deoosits. Tr deposits am; The soil Surve OshlemO-HO Charade-rimC drained. ln a g- . .ssooated wi At the laboratory, a peristaltic pump was connected to the inflow tube and set at a constant rate of approximately 5.5 ml/min to establish a groundwater gradient of approximately 0.18 (Ewers, 1997). Steady-state conditions were held in the mesocosm at all times. In addition, the hydraulic conductivities in the well screen area were estimated by administering slug tests to each of the wells (Ewers 1997). GEOLOGY Surficial Geology- The surficial geologic map of Ingham County (Puzio and Larson, 1982) shows the approximate sample collection site to be located in an undifferentiated sand environment (Figure 3). The undifferentiated sand is desoribed as mostly sand and silt of uncertain age, generally associated with collapsed outwash deposits. The area also shows marsh and swamp deposits, undifferentiated till deposits and undifferentiated sand and gravel deposits (Figure 3). According to the soil survey of Ingham County, soils in this area are a member of the Oshtemo-Houghton-Riddles association soils (USDA, 1979). These soils are characterized by sandy, loamy and muck soils that are well to very poorly drained. In addition, this area has a clearly visible hummocky topography associated with kettle lakes. I I \ Sand 8. Gravel Sand Marsh and TH (eskers 8' kames) I (collapsed outwash Stamp . posuts deposns) Sample Collection Site Figure 3. Surficial Geology Deposits. (Source: Puzio and Larson, 1982). Sedimentology- The genesis of the mesocosm sediments is thought to be part of a series of glacial debris-flow deposits (Grahame Larson, personal communication). Figure 4 shows one typical mesocosm cross-sectional image, with increasing depth towards the bottom of the page. Four distinct labeled units are illustrated with a fifth unit of fine to very sand not shown in Figure 4. The upper two units appear to be a debris-flow deposit (i.e. poor to no grading and matrix supported fabric). The deeper units appear to be part of a different depositional event. due to their well-sorted character. These units are discussed in more detail in the results sections, and all cross sectional images are located in Appendix A. Increasing Depth Figure 4. T Washers) de Wes V‘s/Ere Type named page. Increasing Depth ‘ Silty Sand ' ‘ ine to Medium Sand ‘ ’3 y. A M 3"“. a" -H "’21) 4“ v 1'4. ".14” . Y X ”t“°'°9¥ Well Locations Boundaries Figure 4. Typical Cross-Section of Mesocosm Layer. Small circles (metal washers) denote well screen location and larger holes are locations where soil cores were taken. Lines represent interpreted lithologic boundary with the unit type named. Z-dimension is looking into the figure with flow coming out of the page. hedges The was designe together at 6 cm in length constructed edge flow or flow is a cor beer-teen the flow has bee middle of the and visual ol For more inf Once begin Slowly be Quc'irameE basis that gri Fears A C0r GrounGWatar throughout if W as taken irc aCC ; METHODS MESOCOSM DESIGN (modifietflrom Ewers, 1997) The final mesocosm design is illustrated in Figure 2. The collection tube was designed out of V2” thick polyvinyl chloride (PVC) sewer pipe and held together at each end by 3A” thick PVC and plates (Ewers, 1997). The tube is 108 cm in length with an inner diameter of 45 cm. The tube was machined and constructed to minimize edge flow (Ewers, 1997). In many column experiments, edge flow creates a problem that makes it difficult to obtain valid results. Edge flow is a common limitation in column studies and consists of conduit flow between the sample and the collection tube walls (Ewers, 1997). This conduit flow has been reported in some studies to be as high as 45% greater then in the middle of the column (Rose and Passioura, 1971 ). Based on the tracer results and visual observations, edge flow was not a significant effect in this experiment. For more information on the edge flow controls see Ewers (1997). Once the mesocosm sample was collected, a peristaltic pump was used to begin slowly saturating the sediment. Even though complete saturation could not be guaranteed, the mesocosm was assumed to be completely saturated on the basis that groundwater was continuously pumped into the mesocosm for over 2 years. A constant flow of between 5.5 and 5.7 ml/min was used to establish a groundwater gradient of approximately 0.18 to approach steady state conditions throughout the experiment (Ewers, 1997). Groundwater for the entire experiment was taken from a private well in Okemos, Ml, which was chosen for its accessibility and its proximity (<10 miles) to the aquifer sample location. Raw 11 groundwater v ornplicated tl" Fifteen . tihing (Ewers. Figure 2). A 56 some as well St 1997). With the of the experime abandoned due not mentioned ir Specified in Tabl mesocosm. groundwater was used to avoid geochemical reactions that could have complicated the transport through the mesocosm. Fifteen miniature “wells”, made from 1/16” outer diameter stainless steel tubing (Ewers, 1997), were installed in the mesocosm, (locations illustrated in Figure 2). A series of small holes were drilled in the ends of the well tubing to serve as well screens and the end of the tube was crimped to a point (Ewers, 1997). With the exception of well 1, all the wells were used through the duration of the experiments to sample water and associated tracers. Well 1 was abandoned due extremely low production, as described by Ewers (1997), and is not mentioned in the rest of this thesis. The wells were pushed to the locations specified in Table 1, which were verified during the disassembly of the mesocosm . 12 Table 1. W F/ llrlell/SamfJ Port lnle illilllll 10 11 12 13 14 15 Allmeasuren point shown i J. i and K COC ll The ou ”25 Cm2) to r {Ewers 1997) Table 1. Well and Sampling Port inlet Locations. Wen/Sampnng Pthical Measurements Model Grid Egivalents Port Inlet x Y z J I K 2 32 20.6 70 37 23 18 3 20.5 2.03 70 25 23 18 4 6 14.6 70 1 1 29 18 5 20.5 8 70 25 36 18 6 17.6 31 .7 32 23 12 38 7 31.9 21 .3 32 37 23 38 8 17.5 19.1 32 22 25 38 9 6.7 16.6 32 12 28 38 10 20.2 6.9 32 25 37 38 1 1 32.8 28.6 0 38 15 54 12 13 35.2 0 18 8 54 13 18.8 20.9 0 24 23 54 14 27.3 8.8 0 32 35 54 15 5.1 13.6 0 10 30 54 All measurements are in centimeters and measured from the zero reference point shown in Figure 4. X, Y and Z coordinates were physically measured and J, I and K coordinates are the equivalent locations in the model grid system. The outflow endplate was also designed with five equal recessed areas (125 cm2) to provide area averaged samples of the outflow concentrations (Ewers, 1997). To enable tracer samples to be taken at each of these areas, a piece of stainless steel tubing (approximately 1/16” outer diameter) connected each outflow area outlet, or sampling port inlet, to a sampling port. Each piece of tubing was installed so that each sampling port was the same distance (approximately 15 cm) away from the outflow outlet. This ensured that all sampling port tracer results did not have to be corrected to represent different tracer travel distances. In addition, all of these sampling tubes connected into a 6-Way coupler and eventually discharged as a single outflow. See Figure 5 for 13 the location: view of the E the locations of these sampling ports and refer back to Figure 2 for an internal view of the approximate areas these ports are sampling from. PVC Outflow Plate Tubing I Sample Port 11 / ample Port 13 Sample Port 12 Sample Port 15 Figure 5. Close-Up of Outflow Sampling Locations. TRACER Tracer tests were conducted to provide a detailed assessment of solute transport properties of the mesocosm. The tracer tests were conducted by injecting a mixture of groundwater and tracers into the inflow manifold of the mesocosm and then observing concentrations at the down-gradient wells and outflow points. In addition, these laboratory tracer tests were conducted in a manner similar to a natural gradient field test, but in this case we are able to measure the total outflow much more accurately thus allowing for better control of the mass balance. 14 FheWOL fluorescein an {August 1996 {January 1997 vafuable initial These data are mmmhmiwhhl so observation during the sam To acco snmhaneoushi injectingtwotrE transpon could rdmwewecons Mull Stat, 198 Occurring with: FiUOTeSt ‘ i low paths! ”0‘: inFSibiy deteCte , and Laidiaw‘ 1 SpeCilOphotO FT cl viec trOH'a’s‘het. Previous tracer tests were conducted by Ewers 1997 and involved fluorescein and tritium as tracers. Two tests were performed with fluorescein (August 1996 and September 1997) and one test was performed with tritium (January 1997). All these tests were conducted separately and provided valuable initial data on the solute transport processes within the mesocosm. These data are presented in the results section of this report. However, a limitation with these earlier tracer tests was that each test used only one tracer so observations could not be made on how different tracers were transported during the same time period. To account for any tracer transport differences, two tracers were simultaneously injected for this experiment (August 1998). By simultaneously injecting two tracers, sodium bromide and fluorescein, the differences in their transport could be observed. Both of these tracers are considered to be relatively “conservative” as observed in many studies (Smart and Laidlaw, 1977; Mull et al., 1988; NWWA Annual Meeting, 1989; Gelhar et al., 1992). Differences in the observed concentration histories can be used to evaluate processes occurring within the mesocosm. These tests are described in more detail below. Fluorescein dye, also known as uranine, is often used for investigating flow paths, flow velocities, and dispersivities (Hotzl and Werner, 1992). It is visibly detected at low concentrations but has poor stability under sunlight (Smart and Laidlaw, 1977). This allows it to be easily detected with a spectrophotometer, which is a device that can measure the amount of electromagnetic radiation a sample absorbs. This absorbance can then be used 15 to estimate ti LBW is a lines absorbing Sp‘ Equation 1. A = a Where. Sodium of the United 5 with a variety c pH/mV meter. Wall Of a formu The fluo bit dissolving 2 Sodium hydrox purchased fror m an iniection and 100 ppm f G's/abated USll’“ to estimate the fluorescein concentration of a sample using Beer's Law. Beer's Law is a linear relationship between absorbance and concentration of an absorbing species as shown in Equation 1 (Tissue, 2000). Equation 1. Beer’s Law Equation. A = aAbC Where: A: absorbance (no units). a A = wavelength-dependent absorptivity coefficient. b: the path length of the sample. c: concentration of the compound in solution. Sodium bromide has been used as a conservative tracer in many regions of the United States (NWWA Annual Meeting, 1989). It can be analyzed rapidly with a variety of simple procedures including a bromide electrode connected to a pH/mV meter. The electrode potential of each sample is measured, and then by way of a formula converted into concentration values. The fluorescein stock solution [10,000 milligram/liter (ppm)] was prepared by dissolving 250 milligrams of fluorescein powder in 25 milliliters (ml) of 10% sodium hydroxide solution. The bromide stock solution (1,000 ppm) was purchased from Cole-Parmer Inc. The fluorescein and bromide injection solution was made using a magnetic stirrer to mix 22 ml of fluorescein stock solution, 220 ml of bromide stock solution, and 1,958 ml of groundwater. This mixture resulted in an injection solution of 2,200 ml containing approximately 100 ppm bromide and 100 ppm fluorescein. The injected concentration of each tracer was evaluated using Equation 2. 16 Equation 2 W he On A; injected at a injection peri using a magr remaining 20 sampling freq sampling ioca Table 2, Sam \ W denotes fin a.’ DUfingl EXT'aCted fl'Onc flow rate was Equation 2. Injected Tracer Concentration Calculation _ C1V1 V2 Where: C2 C2: injected tracer concentration (units of milligrams/liter). C1: concentration of tracer stock solution (units of milligrams/liter). V1: volume of tracer stock solution (units of liters). V2: total volume of tracer mixture injected (units of liters). On August 19, 1998 at 16:55, the first volume of tracer solution was injected at a flow rate of approximately 5.7 ml per minute. During the entire injection period the tracer solution was mechanically mixed in a glass beaker using a magnetic stirrer and stirring plate. In total 2,000 ml was injected with the remaining 200 ml used for calibration curves. Table 2 shows the general well sampling frequency during the tracer test. The last sample was collected at each sampling location approximately 1,636 hours from the start of the test. Table 2. Sampling Frequencies During the Tracer Test. SAMPLING FREQUENCY (HOURS) ELAPSED TIME (HOURS) 2 0-22 1 23-36 2 37-60 4 61-72 8 73-104 Twice a day (~8-12 hours apart) 104 —> 182 Once a day to once a week 182 —9 528* * denotes final sample collected 1,636 hours from the start of the test. During individual sampling events, approximately 2 ml of sample was extracted from each sample location using a 5 ml plastic syringe. In addition, the flow rate was estimated and other observations (i.e. turbidity, color, etc.) were recorded. During sample extraction, care was taken to minimize artificial 17 gradients 'suckiflQ" contamina water reme Appendix E from the pre significant ir the same ap labeled 10 rr foil. The alur degradation c cabinet and v. A spec absorbance at Dubilshed GXCl any Samples Vs The blank (i.e. SOiUilOn GXCBp‘ Contamir‘Q knc WEESUred to a used to C“State c . gradients by allowing the water sample to slowly fill the syringe rather than “sucking” the sample into the syringe. To evaluate the significance of cross- contamination between sampling events, an estimate of the potential volume of water remaining in each well was estimated. Based on calculations shown in Appendix B, a maximum of 13% of each 2 ml water sample could have originated from the previous sample event. This volume is not expected to cause any significant impact on the tracer because each tracer test was conducted using the same approach. After the sample was collected it was transferred into a labeled 10 ml plastic sample bottle and place in a cabinet covered with aluminum foil. The aluminum foil and darkness of the cabinet helped minimize the degradation of the fluorescein due to UV light. The samples remained in the cabinet and were analyzed within 2 months of being sampled. A spectrophotometer 601 was used to measure the fluorescein absorbance at a wavelength of 496 nanometers (nm), in accordance to a published excitation wavelength of 490 nm (Smart and Laidlaw, 1977). Before any samples were measured a blank was measured to calibrate the instrument. The blank (i.e. deionized water) was a solution that was identical to the sample solution except that it did not contain the solute (i.e. fluorescein) that absorbs light (Blaunch, 2000). After the blank was measured, a range of standards containing known concentrations of fluorescein (0.1 ppm to 100 ppm) were measured to attain a calibration curve. The data from this calibration curve were used to create a linear plot of absorbance versus concentration (Appendix C). From this plot, the equation describing the slope of the line was used to 18 determine mentioned for a solutic absorbance concentratil section for r A 1: meter was u The relation: described by range of star 1.000 ppm) v non-linearity l Curves were c fanges were L Before the 1Ta( SOiUiiOn was a i l determine the fluorescein concentrations of the unknown samples. As mentioned above, this direct relationship between concentration and absorbance for a solution is known as Beer’s Law (Loschiavo, 2003). Finally, the absorbance of the unknown samples were measured and converted into concentrations using Beer’s Law (Equation 1). See the Results/Discussion section for more information. A bromide electrode (Orion Model 290A) in conjunction with a pH/mV meter was used to measure the bromide voltage measured in millivolts (mV). The relationship between the bromide concentration and mV response is described by Equation 3 (Hall, 1996). Before tracer samples were analyzed, a range‘of standards containing known concentrations of bromide (0.01 ppm to 1,000 ppm) were measured to attain a bromide calibration curve. Due to the non-linearity that occurs when concentrations are below 0.2 ppm, two calibration curves were created. Equations describing the slope of the two calibration ranges were used to calculate the bromide concentrations in the tracer samples. Before the tracer samples were measured, 0.04 ml of ionic strength adjuster solution was added to the sample solution as recommended by the manufacturer (Cole Parmer). 19 Equation 3. Bi E (3:10- Where: Betore tr lperpendicular ‘ was used to ch rolling cart. Tt‘ which time the approximate d mesocosm, at the outflow er “We‘d. The pressme that Sediment. A “Wily the Ame. Were steam. Equation 3. Bromide Concentration and Voltage Equation. C=1o-E—‘—l 5 Where: C= bromide concentration (units of milligrams/liter) l= ordinate intercept from a linear plot of mV versus the log of the bromide concentration. E= mV response of the sample. 8: the slope of the linear plot. DISASSEMBLY Before the mesocosm was dissembled, it was placed on its inflow end (perpendicular to the floor) to allow the mesocosm to drain. A front-end loader was used to change the orientation of the mesocosm and place it back on its rolling cart. The mesocosm was then allowed to drain for about 1 week, during which time the drained water was collected in a graduated beaker to evaluate the approximate drainable porosity of the mesocosm. Due to the clay content of the mesocosm, and the clays ability to retain the water, argon gas was attached to the outflow end of the mesocosm one month after the mesocosm was first inverted. The argon gas was regulated to 25 Kilopascals and created a positive pressure that helped to overcome the capillary forces between the water and the sediment. After approximately 33 days, the draining was stopped and the top (formerly the outflow end) end cap was removed. SEDIMENT CORING After the sediment at the outflow end was exposed, distinct sediment units were delineated using the Unified Soil Classification Scheme. Depending on the 20 sediment r (4.8 cm dil sandy unit sample wi the PVC C was place The wood allowing it were remc approxima to minimiz Pipe and ti lapproxim; Used to (33 was also u the Cut We: The bags, While The Steel {3 Of the COre. sediment matrix, either PVC cores or aluminum cores were used. PVC cores (4.8 cm diameter x 7.5 cm length) were used to collect samples within the silty or sandy units, while aluminum cores (6 cm diameter x 7.5 cm length) were used to sample within the silty clay units. The aluminum cores were more durable than the PVC cores and provided more volume to be sampled. A flat piece of wood was placed over the core and then hammered to drive the core into the sediment. The wood helped distribute the force of the hammer over the entire core top, allowing it to sample the sediment more uniformly. Before the sediment cores were removed, a reciprocating saw was used to cut through the PVC pipe approximately 3-inches (7.6 cm) below the exposed sediment. Care was taken to minimize the impact of the cutting on the sediment near the edges of the PVC pipe and the sediment cores. Once the PVC pipe was cut, a flat piece of steel (approximately 50 cm length x 50 cm width x 1/8 cm thick) with sharp edges was used to carefully out through the sediment within the mesocosm. The steel plate was also used to help support the weight of the sediment located above where the cut was taken so minimal sediment loss or redistribution would occur. The exposed sediment was then excavated and placed in labeled plastic bags, while the sediment cores were carefully removed using the steel plate. The steel plate helped minimize the chance of sediment spilling out of the bottom of the core, by providing a hard surface to contain the sediment. Some cores were lost due to refusal (i.e. contact with gravel or very hard clay) or were not used after observing fractures or other artificial disturbance effects. After the 21 cores were coll placed over theI Thelabe for the first stuc nomenclature it There are a tot; representing a; sediment samp Appendix A for the interpreted ‘ location is pres Table 3. Sam; Layers l Nomenclature l m cores were collected and labeled, plastic end caps or parafilm sheets were placed over the sample ends until further testing. The labeling nomenclature is shown in Table 3. The nomenclature used for the first studied layer (Layer A) was improved, thus creating a new nomenclature for the remainder of studied layers (Layer B through Layer N). There are a total of 14 layers (Layer A through Layer N) with each layer representing approximately 7.7 cm in thickness. The analyses conducted on the sediment samples were hydraulic conductivity, porosity, and grain size. See Appendix A for a plan view of the sample locations used for analyses along with the interpreted lithology. Also, a summary of analyses performed at each sample location is presented at the end of Appendix A. Table 3. Sample Nomenclature Layers Layer A Layers B-N Layers B-N Nomenclature A1 a2 Ba1 B1a (Sample used for (Sample used for archive purposes) Hydraulic conductivity, porosity and grain size tests) Components A: Sample B: Sample B: Sample collection collection layer. collection layer. layer. 1: sample number a: interpreted 1: sample number within layer. lithologic unit. within layer. a: interpreted 1: sample number a: interpreted lithologic lithologic unit. within layer. unit. 2=number of duplicate samples taken in lithologic unit. 22 W Slug 79* Slug 195 hydraulic cond removing a VG; level to recove Hv'orslev (195‘ method assurr extent and the limitations with valid, and onl1. tests are pres»; Falling The falf estimate hydré with low Cond, method was ft: the water leve conductivities “”96 for this HYDRAULIC CONDUCTIVITY Slug Tests- Slug tests were conducted by Ewers (1997) to obtain estimates of hydraulic conductivity at all nine wells. The slug tests were performed by removing a volume of water from the well and recording the time for the water level to recover to its initial level. These data were analyzed using the method of Hvorslev (1951) to obtain one type of hydraulic conductivity estimate. This method assumed that the sediment was homogeneous, isotropic, infinite in extent and the soil and water are incompressible (Ewers, 1997). Common limitations with this method include the fact that these assumptions are rarely valid, and only a small volume of sediment is measured. The results of the slug tests are presented in the results section. Falling Head Permeameter Tests on Small Sediment Cores- The falling head permeameter method was used as the first approach to estimate hydraulic conductivity. This method is commonly used for sediments with low conductivities (Fetter, 1994). After running the first five samples, the method was found to be inadequate due to the difficulty of accurately measuring the water level change over time. This difficulty was due to the high hydraulic conductivities that the sediment samples exhibited which appeared to be out of range for this measurement method. 23 Con Con with mom (Freeze ant expected fc chosen as a unit. Const cm in lengtl the mesoco hydraulic cc using a seri. Appn packed usin diameter of end 01 the s.‘ Scrappfoxint aCOnStant h. had of appr head was me Dre‘weitihed mm. The. _ vii eStimate Constant Head Permeameter Tests on Small Sediment Cores- Constant head permeameter methods are commonly used for sediments with hydraulic conductivities ranging between 10'5 cm/sec and 10'1 cm/sec (Freeze and Cherry, 1979). As this was the range in hydraulic conductivities expected for the mesocosm sediments, constant head permeameter tests were chosen as a method for estimating the hydraulic conductivity of each sediment unit. Constant head permeameter tests were conducted on sediment cores (7.5 cm in length with an inner diameter of 4.8 cm diameter) that were collected as the mesocosm was disassembled. This procedure allowed for the horizontal hydraulic conductivity of each collected sediment core sample to be estimated using a series of measurements and is discussed in more detail below. Approximately 300 grams of oven dried sediment sample was artificially packed using a solid weight into a sample core (6 cm in length with an inner diameter of 6.5 cm). The sample core consisted of two porous stones at each end of the sample core with a rubber O-ring sealing each end and four long screws to help tighten the assembly together. The sample was purged with 002 for approximately 10 minutes and allowed to fully saturate before being placed in a constant head platform. The constant head platform was set-up so a hydraulic head of approximately 9 cm could be induced and maintained. The hydraulic head was maintained with a peristaltic pump, and after approximately 10 minutes pre-weighed bottles were used to measure the volumetric discharge from the outflow of the sediment cores. The hydraulic conductivity of each sample was then estimated using the Darcy’s Law, as shown in Equation 4. Since this 24 method involv new method v sample cores Equat Afier new and imj minimally di heterogenei 4-5 cm dian ring that 39, $2,500 hol served as a methOGS. 1 with the We C:ellentjjng Saturated n saturation v method involved the artificially packing of the sediment into the sample core, a new method was developed to preserve the natural heterogeneities within the sample cores. Equation 4. Darcy’s Law K = :95 AAH WherezQ : volumetric discharge through the system (units of volume/ L: sediment sample length A: cross-sectional area of core AH: constant head differential K: hydraulic conductivity (units of length/time) After conducting measurements using the method discussed above, a new and improved method was created. This method involved using the intact minimally disturbed sediment cores that reasonably preserved the natural heterogeneities. This was accomplished by first connecting the bottom of each 4.5 cm diameter sediment core to a PVC endplate, which contained a rubber O- ring that sealed the sediment core and the endplate together. A fine gage mesh (62,500 holes per square inch) separated the sediment from the endplate and served as a screen to keep the sediment sample from moving. Each sediment core was allowed to saturate from the bottom of the core using one of two methods. The first involved placing the bottom of the core in a container of water with the water level just above the top of the core. The saturation time varied depending on the sediment core lithology. As expected, sandier sediments saturated much faster than silty clay sediments. Visual observations of saturation were used to estimate the appropriate saturation time. The second 25 method in saturating than air, it space. Ont transferrec hose was ( PVC const core tube. the sedime t‘ris mesh. during the t sediment or method involved pumping carbon dioxide through the sediment core before saturating it with water. Since carbon dioxide dissolves in water more readily then air, we expect that this method minimized the amount of air filled void space. Once each core was saturated, the sediment core and PVC endplate were transferred to the constant-head permeameter core platform, where a rubber hose was connected to the bottom of the PVC endplate (Figure 6). Next, the PVC constant head core attachment was connected to the top of the sediment core tube. A heavy-gage mesh (25 holes per square inch) was placed on top of the sediment core and a layer of marbles (1 cm in diameter) was placed on top of this mesh. The marbles were used to prevent any scouring that might occur during the test and the mesh was used to keep the marbles from sinking into the sediment once it was saturated. 26 figure 6. 5 Once 170m 21 10 2 331519an c Beak” A r maintain the . ________ Constant Head Water Level PVC Constant Head Core Attachment /Glass Marbles Screen Mesh PVC Sediment Core . , . Sediment Tube ‘ , j . VI Screen Mesh PVC endplate / PVC Hose Connector Rubber Hose Figure 6. Sediment Core and Constant Head Attachment Assembly. Once the core assembly was complete, a constant head of water ranging from 21 to 22 cm was induced. Figure 7 illustrates the constant head set-up consisting of a core platform, constant head reservoir, pump and collection beaker. A peristaltic pump was used to continuously fill the reservoir and maintain the constant head. A rubber hose was used to siphon water from the 27 constant 1 smhoned reached t. the test. I merubbe: was used discharge both the vi calculated sediment 5 To Emmame water. A h used as the Observatim core, an es CondUCthih “ff finer aHO‘WiF constant head reservoir into the PVC constant head core attachment. The siphoned water continued to rise in the constant head core attachment until it reached the height of the constant head reservoir, where it remained throughout the test. After approximately 10 minutes, a graduated cylinder was placed under the rubber discharge hose and the discharge water was collected. A stopwatch was used to measure the length of time that the graduated cylinder collected the discharge water. This procedure was repeated 4 more times and by recording both the volume of water and length of time, the discharge (Q) of the sample was calculated. All of this data was used to estimate the hydraulic conductivity of the sediment sample based on Darcy’s Law (Equation 4). To estimate the upper hydraulic conductivity limits of the constant head permeameter method, an empty sediment core tube was allowed to fill with water. A hydraulic conductivity estimate of 3.78 x 10'1 cm/sec was obtained and used as the upper hydraulic conductivity estimate limit. Also, based on observations made during measurements conducted on a silty clay sediment core, an estimate of 8.94 x 10.5 was estimated to be the lower hydraulic conductivity limit of the constant head method. This estimate was obtained only after allowing the silty clay sediment core to be measured for 6.5 hours. Due to the length of time it would take to obtain hydraulic conductivity estimates in the silty clay unit, theoretical estimates were assigned to the silty clay unit within the mesocosm . 28 Figure 7. low direc Ko eqUatiOn ( The fluid ‘ constam that is diSl ”hdian gr size at Wh Constant Head Reservoir Rubber Hose —> WW 1 Water Source/ Glass Marbles Constant Head Sediment Core Level ' —> Overflow W/A . Container . ' Collection Beaker (Discharge) Table Core Platform Figure 7. Schematic of Constant Head Set-Up. Small arrows denote water flow direction. Kozeny-Carmen Equation- In addition, to all of the methods described above, the Kozeny-Carmen equation (Equation 5) was also used to provide hydraulic conductivity estimates. The fluid pressure, acceleration due to gravity and dynamic viscosity can be held constant. The porosity was estimated from the core samples using an approach that is discussed in more detail in the Effective Porosity methods section. The median grain size was estimated by using a grain size plot to estimate the grain size at which 50 °/o of the grain sizes were larger and 50 °/o were smaller. 29 Equatio was no lo: Sediment 1 core platfc the marble i'a‘as Weigh the sedime end cap ar irOm The SE 8aWe. Equation 5. Kozeny-Carmen Equation. K=p_9* ”3 *dfii u (1-n)2 180 Where: p = fluid pressure (units of mass/length x timez). g= acceleration due to gravity (length/ time2). u: dynamic viscosity (units of mass/length x time). n: porosity (units of decimal fraction or percent). dm = median grain size (units of length). K= hydraulic conductivity (units of length/time). EFFECTIVE POROSITY After the constant head permeameter analysis was complete and there was no longer any water flowing out of the discharge hose, the fully saturated sediment core and constant head attachment assembly was removed from the core platform. The constant head core attachment was then removed along with the marbles and the rubber discharge hose. Next, the sediment core assembly was weighed and the sediment was transferred into a pre-weighed beaker. After the sediment was completely transferred into the beaker the mesh screen, PVC end cap and sediment core tube were measured. Subtracting these weights from the sediment core assembly weight provides the saturated weight of the sample. Next, the sediment sample and beaker were placed in a drying oven at 105 °C, and allowed to dry for 24 hours. The dried sediment sample and beaker were then taken out of the drying oven and allowed to cool for approximately 1 hour and reweighed. The difference in weight between the fully saturated weight 30 and the dry W sediment sam dimensions of estimate the assuming a w samples was saturation wei Equation 6. He: Where ne = ef Vv= vo Vt: tot. W Grain 1 hegp Speed u mO'OUQth d: the Sedimen grailel from t: and the dry weight represented the total volume of water removed from the sediment sample. Next, the total volume was calculated from the measured dimensions of the sediment core. Finally, all these parameters were used to estimate the effective porosity of each sediment sample by using Equation 6, assuming a water density of 1.0 g/cm3. It should be noted that the VV for some samples was also estimated by calculating the difference between the pre- saturation weight and the fully saturated weight. Equation 6. Effective Porosity Equation n, = & Vt Where: ne = effective porosity (units of percentage) . . 3 W: volume of vord space (units of cm ) Vt: total volume (units of cm3) GRAIN SIZE Grain size analyses were conducted at the MSU Physical Geography laboratory. Due to the extensive amount of clay found in the mesocosm and to help speed up the drying process, the drying oven (105 °C) was used. Once thoroughly dried, the sediment sample was ground up using a mortar and pestle. The sediment sample was then poured through a 2.0—mm screen to separate any gravel from the rest of the sample. Next, 12 grams +/- 0.2 grams of the sieved material was weighed and placed in an 8-ounce glass jar, where it was mixed 31 with a dist)ers separate the C tnrough a 64-.r and collected washed with c the sand size in the graduat . . l Sieve Into a 2: for about 24 r the top of a se sizes enablec sand to coats Shaker and it Sie‘fes were then estimat Of the Sedirr The t seme Until a were C810ui giameCUb.i was D\aCe( estimated D . SrCentah \4 V with a dispersing agent and shaken on a motion table for 11 hours to help separate the clay particles. These dispersed samples were then wet sieved through a 64-micron screen. The silt and clay were separated from the sand, and collected in a 1000 ml graduated cylinder. Each sample was continuously washed with distilled water until there was no sediment left in the glass jar. Thus, the sand size grains remained on the sieve while the clay and silt were collected in the graduated cylinder. The sand was then separated from the 64-micron sieve into a 250 ml glass beaker that was then placed in a drying oven (105 °C) for about 24 hours. After the sand was thoroughly dried it was transferred onto the top of a set of sieves (1 mm, 0.5 mm, 0.1 mm, 75 micron). This range of sieve sizes enabled the sand grains to be classified between the range of very fine sand to coarse sand. The set of sieves was then clamped into a mechanical shaker and left to shake for 12 minutes. After the shaking was completed the sieves were emptied and weighed accordingly. The percentage of sand was then estimated based on the total weight of the sand divided by the total weight of the sediment multiplied by 100. The clay and silt solution was brought to a volume of 1000 ml and left to settle until all particles larger than 2-microns settled (see Table 4). These times were calculated from the Stokes’ equation, assuming a particle density of 2.60 grams/cubic centimeter. After the clay and silt solution had settled, a hydrometer was placed in the solution. The gradation on the hydrometer was then read and estimated to equal the number of grams of clay per liter of solution. The percentage of silt was then estimated by adding the clay and sand together and 32 subiia summ; Table TE J H ____._ l L..— ____._ The pe definiti micror U E mesoc and 53 locatio constn lnaddi and er llii'iolog lye, 'Sfor subtracting from the total sediment weight. See the results section for a summary of these results. Table 4. Sedimentation Times of 2 micron particles TEMPERATURE (Celsius) 2 MICRONS (minutes) 18 503 20 480 21 469 31 374 The particle density value has been chosen to simultaneously satisfy the two definitions of the clay fraction, viz., particles having an effective diameter of 2 microns and a settling velocity of 10 centimeters in 8 hours at 20 Celsius. DIGITAL IMAGING A digital camera was set up directly above each layer of exposed mesocosm sediment and digital images were taken throughout the disassembly and sampling process. Figure 4 shows atypical digital image along with the locations of wells and lithologic boundaries. These images were used to construct the grid used for the computer simulations and to document each layer. In addition, they provided a way to document and record the exposed sediment and enabled the opportunity to go back to any layer and reassess earlier lithologic interpretations. Refer to Appendix A for the entire set of images. COMPUTER MODEL ' A numerical model was used to simulate groundwater flow and solute transport within the mesocosm. The Groundwater Modeling System (GMS) version 4.0 was chosen as the graphical user interface to develop the MODFLOW 2000 (Harbaugh et al., 2000) and RT3D (Clement, 1997) numerical 33 models three'di Using a 5031596 MT 3DM 5.090165 Ignspor modemL soluies (; A equali C is in the F through if registereC mapped ll umngthel FgweQi conductivit edited to re EaCl :igital tmag 3'3, . .. i.0iichgi models. MODFLOW 2000 is a computer program that numerically solves the three-dimensional (3-D) ground-water flow equation for a porous medium by using a finite-difference method (Harbaugh et al., 2000). This solution provides both spatial and temporal groundwater head distributions that can be used by the MT3DMS and RT3D solute transport models. MT3DMS is a modular 3-D multi- species transport model (Zheng and Wang, 1999) and RT3D is a 3-D reactive transport model (Clement, 1997). RTSD was chosen as the tracer transport model due to its flexibility and superior ability to handle transport of multiple solutes (Zheng and Wang, 1999). A 54-layer finite-difference grid with 124,416 cells was created with cells of equal 1 cm by 1 cm by 2 cm dimensions (Figure 8), where the larger dimension is in the primary flow direction. Digital images were imported into GMS and through the use of the GMS map module, the digital image coordinates were registered so the digital image would overlay on top of the model-grid. Next, the mapped lithologies were transferred from the digitai image to the model-grid using the materials feature of the MODFLOW 2000 (Harbaugh et al., 2000) code (Figure 9). In instances, where other data (i.e. grain size or hydraulic conductivity) did not support the original mapped lithologies, the model-grid was edited to reflect these changes. Each digital image represented a thickness of 8 cm. Therefore, each digital image represented 4 model layers with the exception of the inflow and outflow digital images. The outflow and inflow digital images were chosen to 34 represent it“ portions W 48 CF ._LLS = 1 cm/cell Figure 8. Mt represent three model layers or 6 cm to account for the sand pack and recessed portions found at both ends. PLAN VIEW SIDE VIEW 48 CELLS = 1cm/cell SHBAV'I 179 48 CELLS = 1 cm/cell "SO/LUCK TOTAL CELLS = 124,416 (195,432 cm3) Figure 8. Model Grid. 35 Modfi.‘ W" ‘ 0",- 51!? Figure 9. M06 Next, a lie. using injec outflow of the n ingesting water/ which correspo it .e tracer test ( Can of the mod it aile constant 9'0‘ . Lndwater flo 3‘90 ‘189Tadier I"ii 35 solved USIP Model Grid Digital Image of Lithology Boundaries Clayey Silt Silty Clay Model Grid with Assigned Materials Figure 9. Model Gridl and Typical Lithology Overlay. Next, a specified flux boundary was created at the inflow of the mesocosm (i.e. using injection wells) and a constant head boundary was created at the outflow of the mesocosm. The specified flux boundary consisted of 4 wells injecting water/tracer at a flow rate of 2.0520m3/day each to total 8,208 cm3/day, which corresponds to the actual average flow rate of 5.7 ml/min observed during the tracer test (August 1998). These injection wells were located in the central part of the model to correspond with the location of the inflow. The cells making up the constant head boundary were held constant at 170 cm, to create a groundwater flow gradient of approximately 0.18. This gradient corresponds to the 0.18 gradient reported in Ewers (1997). Once the groundwater flow model was solved using MODFLOW 2000, RT3D was then used to simulate non- 36 reactive sol periods. Th concentratic oi the meso< injected thro normalized ti grid and use: stress period diffusion of in period 1) and 68 days, only‘ in the Results SONGI was US criterion set tc Characteristics discus56d in it reactive solute transport. The RTBD simulations were divided into two stress periods. The first stress period simulated the injection of tracer with a normalized concentration (concentration divided by initial concentration) of 1.0 at the inflow of the mesocosm through the each of the four injection wells. The tracer was injected through the injection wells for 0.28 days. Next, these calculated normalized tracer concentrations were transferred to the solute transport model grid and used as the starting normalized tracer concentrations for the second stress period. The second stress period simulated the advection, dispersion and diffusion of the tracer after the entire tracer mass had been injected (stress period 1) and was simulated for 4 days. Even though the tracer test ran for over 68 days, only the first 4 days were modeled. This will be described in more detail in the Results section of this paper. The Generalized Conjugate Gradient (GCG) solver was used with the Jacobi precondtioner with the convergence error criterion set to 104. The advection term was solved using the hybrid method of characteristics (HMOC) was used with the Runge-Kutta tracking algorithm used at or near sources and the Euler tracking algorithm used everywhere else. The dispersion term was solved using a wide range of parameters and will be discussed in the Results section. 37 LEAQEB Theth impact of flov essential to u tracer tests is conditions an calculated for outflow conce liters) that dis‘ obtained by i r events. This tracer (220 m the inlected fl “Onservative. RESULTS/DISCUSSION TRACER The tracers fluorescein, tritium and bromide were used to evaluate the impact of flow and transport through naturally heterogeneous sediments, essential to understanding Solute transport processes. A summary of each of the tracer tests is presented in Table 5. To demonstrate the controlled flow conditions and conservative nature of each tracer, the tracer recovery was calculated for each test. The tracer mass was calculated by multiplying the outflow concentration (units of milligrams/liter) by the volume of water (units of liters) that discharged at the time of sampling. The total mass of tracer was obtained by integrating the tracer mass at the outflow across all of the sampling events. This total observed mass was then divided by the total mass of injected tracer (220 milligrams) and multiplied by 100 to give a percent. The fluorescein tracer recovery for the August 1998 test was 99.9%, which means that nearly all the injected fluorescein tracer was recovered, demonstrating that fluorescein was conservative. See Figure 10 for the tracer mass conservation through time for the August 1998 fluorescein test. 38 Table 5. SUI fluorescein it results (see i / _____,______ M Duration of T (Hoursi lritiai Concent i (Fluorescein i Bromide units: I and Tritium ur DPMVmii Average Pump (mHmn) injected Tra. vmume lifters) Mass Conser Cumu Tracer Recov (9e) FiQUre FluoresCein Table 5. Summary of Tracer Tests and Mass Conservation. * denotes fluorescein mass conserved, bromide mass was not calculated due to erratic results (see earlier discussion). Fluorescein . . . Fluorescein Tracer Tracer Test T232£$§si|12 Tritru1r_ne:tracer Test #3 and Bromide #1 Tracer Test Date of Test 8/23/96 1/23/97 7/12/97 8/19/98 Duration of Test 551 639 559 1636 (Hours) Initial Concentration 100 950,000 100 1 00 (Fluorescein and Bromide units=ppm and Tritium units: DPM/ml) Average Pump Rate 5.3 5.2 5.3 5.7 (mein) Injected Tracer 2 2 2 2 Volume (liters) Mass Conserved 74.1 87.8 80.5 *99.9 (°/o) 100' ‘ " —;:-W -- ~ ‘ . g i .: Cumulative J f Tracer Mass , ., Recovery . i (°/o) ' f ‘ t l l l l + I o..’___ _V ILL- . __,._._.._ LL "we-.- ___ _ _ ________. 0 Time (hours) 1300 Figure 10. Fluoure’scein Tracer Mass Recovery. August 1998 Fluorescein data shown. 39 As i were fluore tracer are F mehnearré curves pres 25 ppm COU linearity cou be used to n wavelength t Therefore, si the spectropl the calibratio Presented in The br lest due to er bromide anal Original VOIUn to Ieligih of ti' aniii/Zed. Til the 903 IIUOre Space for stc' The ”LI past traCEr re As described earlier, the tracers used during this study (August 1998) were fluorescein and bromide. The calibration curves used for the fluorescein tracer are presented in Appendix C. As discussed earlier, Beer’s Law describes the linear relationship between concentration and absorbance. The calibration curves presented in Appendix C show that concentrations ranging between 5 and 25 ppm could not be fit to a linear trend line. According to Sibert (2004) this non- linearity could be attributed to the fact that more than one wavelength band could be used to measure the absorbance of the samples. Even with the optimal wavelength band chosen, deviation from Beer’s law can still exist (Sibert, 2004). Therefore, since a published excitation wavelength of fluorescein was used to set the spectrophotometer, this non-linearity should be at a minimum. In addition, to the calibration curves, the Fluorescein tracer concentration histories or BTCs are presented in Appendix D. The bromide data appear to be significantly less representative for this test due to erratic observations noted during the bromide analysis. During the bromide analysis, most of the sample vials contained only 0.5 to 1.5 ml of the original volume of 2 ml. The missing sample volume most likely evaporated due to length of time they were left (approximately 6 months) before they could be analyzed. The delay in bromide analysis was mainly due to the need to analyze the 903 fluorescein samples first, further complicated by the lack of refrigerator space for storage. Therefore, the bromide data are not presented or discussed. The fluorescein concentration histories (or BTCs) are presented along with past tracer results as reported by Ewers (1997) in Figures 11-14. Many of the 40 sampling locations exhibited asymmetric BTC’s, characterized by a sharp peak followed by a gradual tailing off in concentrations. 0.25 Well 4 6160 0 Time (days) Well1 NO DATA 0.03 L Well 3 C’CO .‘:;{’% 'j. If “wk-:1. .3?“ ; ‘ 7’ .‘i ‘ -,' I , 1.... 3......” [.53va If“; ...".;*-fii -""‘ O fa 10 0 Time (days) 10 0003 Well 5 C/Co ‘l i' i I I N’Jl-fix‘fi '~-r— .F- ~ » O . “wa I u‘ I. 0 TIme (days 10 + August 1996 Fluorescein + January 1997 Tritium ‘- September 1997 Fluorescein + August 1998 Fluorescein 0.03 Well 2 CICo .,/' / II' .* ~ 1' IIII‘ - I III Ipunlp 0 0 Time (days) 10 — \ Direction of flow Figure 11. Tracer Concentration Histories (Well 2 through Well 5). C/Co represents concentrations are normalized by injection concentration (100 ppm).F|uorescein August 1996 data, Tritium January 1997 data, and Fluorescein September 1997 data taken from Ewers (1997). Outliers at Well 4 and Well 5 are not shown. 41 9 Well 6 l. . ”it 0;.J anhiomm 0 Time (days) 10 0.06 0.8 0.03 Well 9 5 Well 8 Well 7 I ;' l . g ',| 1 i g _. . '* °’°° l CICo _ .- CICo J if .‘ ~‘ ' ‘i " _,-,.. (I; . . [r ' _ fir, . 4"” . l "I! E ,4; l §-.\ I; ‘ H) p JIM-«.1 L 0 LLKIAALL'I‘: ‘ ,. _)~—I’A»'r"‘"‘~<~' -/ 0 owJ/‘W “ ‘ ‘~ ' O i waf 10 . . 1 . Time (days) 10 Time (days) 0 Time (days) 0.006 , Well 10 if”, l}. l: ”‘, Iii l» "i 6100 *l ’ a“ l r l. 1]. 3., 7‘ ~, ’- > _ A . ‘ 4 [V l l - , ' 0 - l 0 Time (days) 10 + August 1996 Fluorescein J“ September 1997 Fluorescein + January 1997 Tritium + August 1998 Fluorescein Direction of flow Figure 12. Tracer Concentration Histories (Well 6 through Well 10). C/Co represents concentrations are normalized by injection concentration (100 ppm). Fluorescein August 1996 data, Tritium January 1997 data, and Fluorescein September 1997 data taken from Ewers (1997). 42 Figu Sam CON-Cg iala. 0.25 Time (days) Sampling Port 14 0'25 Sampling Port 11 j}: ! \\~\\~ ten. 0 Lle ‘ A 0 Time (days) 10 0'25 Sampling Port 13 fi‘. CICo f 'l' \l ., l/W‘T\\ 10 0 Time (days) 10 Sampling Port15 A L". CICo ,‘ #1:," “ilk?“ t O :5 ‘11/ 't‘§1ugw 0 Time (days) 1 + August 1996 Fluorescein + January 1997 Tritium 4'“ September 1997 Fluorescein + August 1998 Fluorescein “‘ 10.111 3&9 7.1:.) 0-25 Sampling Port12 CICo iii/‘0‘} 1, \ .‘_ .- i were r"fl ""5 95:22:30“ 0 0 Time (days) 10 PVC Outflow Plate Sample Port 11 Sample Port 14 Outflow Sample Port 13 Sample 90" 12 Sample Port 15 Figure 13. Tracer Concentration Histories (Sampling Port 11 through Sampling Port15). C/Co represents concentrations are normalized by injection concentration (100 ppm). Fluorescein August 1996 data, Tritium January 1997 data, and Fluorescein September 1997 data taken from Ewers (1997). 43 0.25 CICo Outflow ..£.'~ -. 1 Time (days) ~°~~ ~'-‘ '2""0~Aoo'.- + August 1996 Fluorescein + January 1997 Tritium -“~ September 1997 Fluorescein + August 1998 Fluorescein 1 0 PVC Outflow Plate Sample Port 11 Sample Port 14 Outflow Sample Port 13 Sample PO“ 12 Sample Port 15 Figure 14. Tracer Concentration Histories (Outflow). C/Co represents concentrations are normalized by injection concentration (100 ppm). Fluorescein August 1996 data, Tritium January 1997 data, and Fluorescein September 1997 data taken from Ewers (1997). The outflow represents the combined flow of sample port 11 through sample port 15 (lower right diagram of this figure). Lithologic observations made during the disassembly and sediment core extractions were used to assist in interpreting the BTCs. These lithologic observations are presented in Table 6. Seven of the sampling points happened to be set in silty sand, four in silty clay, one in the transition zone between silty 44 sand and silty clay, one in the transition zone between silty sand and fine to medium sand, and one in clayey silt. As would be expected, these observations helped provide the characterization data necessary to explain the asymmetric shape of the BTCs. Each BTC is discussed in more detail below: Table 6. Sampling Point Lithologic Descriptions. WELL/SAMPLING LOCATION Lithology Well 2 Silty Sand Well 3 Silty Sand Well 4 Silty Sand Well 5 Silty Clay Well 6 Silty Sand Well 7 Silty Sand Well 8 Silty Sand Well 9 Silty Sand/Silty Clay Well 10 Silty Clay Sampling Port 11 Silty Clay Sampling Port 12 Clayey Silt Sampling Port 13 Silty Clay Sampling Port 14 Fine to Medium Sand/Silty Sand Sampling Port 15 Silty Sand Tier-1 (Wells 2-5): The first tier of sampling locations (wells 2 through 5) (Figure 11) shows that well 4 consistently had sharp asymmetric normalized concentration peaks and had the earliest tracer arrival times within this tier of wells. Well 4 exhibited normalized tracer concentrations that ranged between 0 and 0.25. These concentrations were approximately one order of magnitude greater than those normalized concentrations observed at wells 2 and 3 and two orders of magnitude greater than those observed at well 5. This difference can be explained by referring to the lithologic units near these well screens (layers J and 45 K) found in Appendix A. Well 4 was screened in silty sand (layer J), however a fine to medium sand unit was located almost directly above it (layer K) most likely causing tracer to show up at this well earlier than the other wells. in addition, well-5 was set in the middle of a low conductivity silty clay unit which would explain why the observed normalized tracer concentrations at this well are approximately one to two orders of magnitude lower than the other wells in this tier. . Tier-2 (Wells 6-10): The second tier of sampling locations (well 6 through well 10) from the inflow (Figure 12) shows that well 6 and well 8 consistently had the highest tracer concentrations and one of the earliest tracer arrival times within this tier. Well 6 was screened in a silty sand unit with normalized tracer concentrations ranging from O to 0.25. Referring to layers E and F in Appendix A, the silty sand unit is observed to have smaller units of fine to very fine sand within it. Also, a larger fine to very fine sand unit is found directly above where well 6 is screened. Well 8 was also screened in the silty sand unit with normalized tracer concentrations ranging from 0 to 0.8. These higher concentrations can be explained by observing that the well screen was located along a parting (layer E). Well 7 and Well 9 were also set in silty sand and had normalized tracer concentrations ranging from O to 0.06 which is approximately one order of magnitude lower than the other wells (well 6 and well 8) screened in this silty sand unit. Well 7 appears to be screened in a section of the silty sand unit with less fine to very fine sand located above. In addition, it appears that silty clay is located directly 46 refs a to Tie COlll Shar Sun above where well 9 is screened. Finally, well 10 exhibited normalized tracer concentrations ranging from 0 to 0.006, which is approximately two orders of magnitude lower than the other wells in this tier. This can be explained by referring to layer E in Appendix A to observe that well 10 was set in the middle of a low conductivity silty clay unit. Tier-3 (Sampling Ports 11-15 and Outflow): The last tier of sampling locations (sampling port 11 through sampling port 15 and outflow) (Figure 13 and Figure 14) enabled the entire outflow volume to be sampled. As discussed earlier, these sampling ports were designed to sample a much larger region consisting of approximately 1/5 of the total outflow at each sampling port (Figure 2). All the sampling ports had normalized tracer concentrations ranging from 0 to 0.25 and have similarly shaped BTCs with sharp asymmetric peaks followed by a gradual tailing off in concentrations. Summary: With the exception of well 8 and well 10, the normalized tracer concentrations at each sampling point increased with each new tracer test. Table 7 summarizes the peak arrival time and its concentration as compared to the injected tracer concentration for each tracer test. Another observed trend was that tritium was consistently the slowest tracer. This is most likely caused by the difference in the molecular weight of the two tracers. This will be described in more detail later in this paper and is described in the earlier study by Ewers (1997). 47 Table 7. Summary of Tracer Peak Arrival Times and Concentrations. :ngreefifgg; Fluorescein Tracer Tritium Tracer Test Fluorescein Tracer (1 996) Test (1997) (1997) Test (1998) Slamp. Peak Conc. (% Peak Conc. (% . Conc. (% Peak Conc. (% 00. Arrival . . Peak Arrival . . Time I _ of Arrival Time . of Time (Total . of Arrival Time ' of (Total njected (Total Injected Hours) Injected (Total Injected Hours) Conc.) Hours) Conc.) Conc.) Hours) Gone.) 2 28.0 6.0% 56 0.8% 559 0.4% 159 3.0% 3 27.0 0.5% 34 3.0% 35 2.0% 1 70 3.0% 4 20.0 10.0% 18 19.0% 31 17.0% 17 31.0% 5 123.5 0.1% 56 4.0% 559 0.0% 196 0.4% 6 51.6 7.0% 38 22.0% 51 1 1 .0% 38 32.0% 7 274.0 0.8% 92 2.0% 85 2.0% 24 3.0% 8 31 .4 69.0% 38 23.0% 45 23.0% 34 33.0% 9 274.0 1 .0% 78 3.0% 130 5.0% 121 7.0% 1 0 57.5 0.5% - - 1 30 0.4% 1 96 0.4% 1 1 43.0 12.0% 46 18.0% 54 15.0% 42 30.0% 12 56.0 5.0% 44 19.0% 52_ 1 1 .0% 44 30.0% 13 60.0 4.0% 44 17.0% 55 12.0% 38 31 .0% 14 45.0 9.0% 46 1 9.0% 57 12.0% 40 32.0% 15 55.0 7.0% 48 15.0% 55 13.0% 44 28.0% Outflow 36.0 1 1 .0% 42 18.0% 55 12.0% 40 30.0% A dash indicates that there was not sufficient data collected for center of mass calculation, Samp. Loc. denotes sampling location, and Conc. denotes concentration. To aid in the interpretation of the BTC results, some fundamentals of solute transport need to be described. As groundwater travels through a porous medium (aquifer), the concentration of each solute is affected by both physical processes (fluid movement, diffusion or dispersion) and chemical processes (Drever, 1997). Fluid movement, or advection, refers to the process by which solutes are transported by moving groundwater (Fetter, 1994). Diffusion is the process by which ionic and molecular species dissolved in water move from areas of higher concentration to lower concentrations (Fetter, 1994). Mechanical dispersion is the process by which a solute flowing through a porous medium will 48 llo W l EDCC flow at different velocities as controlled by the heterogeneities that are encountered. Chemical processes (e.g., radioactive decay, sorption, precipitation, redox and complexation) affect solutes depending on their transport characteristics and ability to transform into other substances. These concepts can all be represented using the one-dimensional advection-dispersion equation, as shown in Equation 7 (Freeze and Cherry, 1979). This equation shows that when groundwater velocities are high, mechanical mixing (i.e. dispersivity multiplied by the average linear velocity) dominates the dispersion of the solute (Freeze and Cherry, 1979). More importantly, the dispersivity term represents all the flow velocity variations that cannot be described nor incorporated into solute transport computer models. In other words, dispersivity represents the uncharacterized heterogeneities of the media and is documented to increase with increasing scales of study (Gelhar, 1992). Conversely, in regions with lower groundwater velocities molecular diffusion dominates the dispersion process (Freeze and Cherry, 1979). 49 Equz Equation 7. One-Dimensional Advective—Dispersion Equation. 8C 82C —aC __ = D ———V —+ — reaction terms a: '812 ’az / ( ) Where: BC . . . . . 3? = change in concentration With time [units of (mass/volume)/time). D. = coefficient of hydrodynamic dispersion along flow path (units of length2/time) and defined by or . v + D*. C: solute concentration (units of mass/volume). V. = average linear velocity (advection). t= time. I: curvilinear coordinate direction taken along the flow line. or .= longitudinal dispersivity. D* = coefficient of molecular diffusion for the solute in the porous medium (units of lengthzltime). +/- (reaction terms) = account for sorption, chemical reactions, biological transformations, and radioactive decay. Relating these solute transport processes and Equation 7 back to the BTCs, if the mesocosm was not filled with any sediment the conservative tracers would generate a BTC characterized by a “plug” of tracer arriving at an observation point at the same time controlled only by the process of advection. However, this “plug” response is not realistic in sediments due to the fact that all natural media contain heterogeneity and thus mechanical dispersion. Therefore, when physical properties are taken into account, BTCs similar to those observed in this experiment are expected. These physical properties are responsible for this observed spreading of the tracer front, and cause the BTCs to become asymmetrical in shape. This asymmetrical shape was observed during all of the tracer tests and is shown in Figure 11 through Figure 14. 50 affect Etvrl aswe WESC cond; $8211.“. and fl from . In addition to small-scale heterogeneities, other things that may have affected the shape of the tracer BTCs through time are presented below: 1. Physical Changes of Mesocosm - Considering the mesocosm sample was collected from an unsaturated environment by necessity, it is quite possible that some grains were redistributed as water saturated it. In fact, Ewers (1997) observed sediment mobility in the mesocosm but noted it as the shift from unsaturated conditions to saturated conditions proceeded. This sediment mobility could change some of the sediment hydraulic properties over time, causing a potential modification to flow and former high conductivity regions. Thus, changes in the shape of the BTC from one test to the next could be due to the sediment moving and not to the aquifer medium. One study (Saiers et. al., 1994) found that the deposition of colloidal silica was at least partially reversible. This could support the claim of sediment or colloid mobility based on the fact that over time the silica could be redistributed. However, this issue of sediment or colloid mobility is not a likely cause for any change in the shape of the BTC since all the tracer tests yielded very similar results and no significant channeling or missing sediment was observed during disassembly. Due to the nature of the collection procedure, it is likely that a significant amount of air was trapped within the mesocosm during the earlier tracer tests. This entrapped air may have affected these earlier tracer tests by closing or opening flow paths as they migrated with the water to the outflow. These 51 Change of the e GOES l) ability I changes inflow paths, do not appear to be significant when comparing the BTCs of the earlier tracer tests with the later tracer tests. Therefore, the entrapped air does not appear to have had a significant impact on the tracer tests. Another physical change that could possibly affect the BTC shapes is the ability of clay particles in the sediment to swell and shrink. This would most likely occur in the early phase of the experiment at the beginning of the sample saturation procedure. During disassembly it was observed that clay at the outflow end was parted possibly due to clay shrinkage (Figure 15). It is hard to know if this cracking occurred prior, during or after the tracer tests were conducted. However, the extent of these cracks was observed to be approximately 2 cm in depth and was not observed anywhere else. Therefore, it is unlikely that sediment mobility or clay swelling or shrinking played a major role in the shape of the BTC’s due to the lack of visual evidence and also keeping in mind the mesocosm was continuously saturated before and during all of the tracer tests. Clayey Silt Silty Clay Silty Sand IA. 1'. Figure 15. Clay Cracks Observed in Outflow Layer. Small circles (metal washers) denote well screen location and larger holes are locations where soil cores were taken. Lines represent interpreted lithologic boundary with the unit type named. Z-dimension is looking into the figure with flow coming out of the page. 2. Molecular Weight of Tracers The effect that tracers with different molecular weights have on transport has not commonly been observed or studied. Ewers (1997) observed on two different occasions that fluorescein arrived at several of the sampling locations significantly sooner than tritium. One of the reasons for this measured difference in arrival time is likely the difference in the two tracer’s molecular weights (Ewers, 1997). Fluorescein has a much larger molecular weight (376 g/mole) than tritium (19 g/mole). Using the equation (Equation 8) presented in Ewers (1997), tritium 53 has a five times greater chance of diffusing into lower conductivity regions or dead end pores, as compared to fluorescein. Stated another way, fluorescein would have more inertia to resist the diffusion into and out of the high conductivity regions. Ewers (1997) designated diffusive retardation to be the process most likely to explain this phenomenon. Figure 16, illustrates the process of diffusive retardation and provides an example. Other studies have shown similar trends with the tracers Ca (molecular weight = 40 g/mole) and Cl (molecular weight: 35 g/mole). Ca is considered to be a reactive tracer and Cl a non-reactive tracer. However, Ca tracer front was observed to pass through the column more rapidly than Cl (Grisak et. al., 1980). A possible explanation for these results could be related to the tracer’s molecular weight, as discussed above. Therefore, these processes could be contributing to the shape of the tracer BTCs. Equation 8. Saturated Porous Media Diffusion Coefficient Equation D: 1.728 x 10’4 (m2/day) x % Where: D= Diffusion Coefficient (units of m2/day). M: Molecular Weight (units of grams). 54 ‘— LowK region _. Time Step 1 T— 1:321: :32): _—: Time Step 2 = F Time Step 3 ¥i Time Step 4 i. * Time Step 5 = q Diffusive Solute Non-diffusive Solute Figure 16. Diffusive Retardation (from Ewers, 1997). Three identical particles are shown on the left and three different identical particles are shown on the right. The media on both sides is identical with the highly conductive layer between two less conductive layers. From top to bottom 5 progressive time steps are represented. Notice in time step 2 on the left side that the concentration gradient caused the leading particle to exit the highly conductive layer and become ‘dorrnant’ in the less conductive surroundings. Only in time step 5 is this particle released back into the highly conductive layer. Also, notice that the non-diffusive tracer on the right is exiting in time step 5, while the diffusive tracer is still at least 2 steps behind. 3. Sorption of Tracers Smart and Laidlaw (1977) extensively reviewed fluorescein and came up with various conclusions to environments suitable for its use. In a comparison of tracer dye adsorption on minerals and organic materials, fluorescein showed a slight affinity for kaolinite (Smart and Laidlaw, 1977). Since the mesocosm contained a silty clay unit it is possible that kaolinite is one of the clay minerals making up this unit. However, substantial adsorption is not likely as the tritium 55 tracer test showed fluorescein to be transported through the mesocosm faster than tritium (Ewers, 1997). The bromide tracer has been used in many studies as a conservative tracer (Levy and Chambers, 1997; Petrich, et. al., 1998 and NWWA, 1989). However, some researchers (Boggs and Adams, 1992) have found that adsorption of anionic tracers can be produced by iron oxides and kaolinite present in combination with low ambient groundwater pH. In their study, approximately 15-20% of the bromide tracer was adsorbed during a natural gradient experiment. The authors concluded that methods of soil sampling and sample preparation for column experiments might significantly affect anionic tracer behavior. Methods causing abrasion, fracturing, or desiccation of soil particles may alter mineral surface chemistry affecting anionic adsorption sites. Column procedures that preserve the original mineral surface chemistry are more likely to produce meaningful results. Since the mesocosm is a minimally disturbed sample and the fact that the pH of the groundwater should not have been low there is no reason to believe bromide had any significant adsorption ’ occurring. Tritium has been used as a conservative tracer for many studies (Ewers, 1997; Webster, 1996 and Netter and Behrens, 1992). Tritium becomes incorporated in the water molecule when dissolved in water and the movement of the tritium tracer front in water is synonymous with water movement (Webster, 1996). For these reasons, there is no reason to believe tritium had any significant adsorption during the tritium tracer test conducted by Ewers (1997). 56 4.Carbonate Equilibrium Local groundwater is found to usually be slightly alkaline and contain Ca and Mg as the dominant cations and HC03 as the dominant anion (Freeze and Cherry, 1977). Given these constituents, it is quite possible that precipitation and dissolution of carbonate material could occur. This of course, would change the structure of the flow paths whenever one or the other process occurs. No significant amount of carbonate precipitation was observed nor was any voids signifying dissolution occurred. DISASSEMBLY Approximately 3.32% (6,430 ml) of the total drainable mesocosm volume (193,479 ml) was recovered after 787 hours of inversion (Table 8). Argon gas was then used to help free water for the remainder of the time. The water extraction process was stopped after 864 hours of inversion with approximately 5.70% (11,020 ml) of the total drainable water volume (193,479 ml) recovered. This value corresponds to an effective porosity of 5.70% that is consistent with the value of 6.8% used by Ewers 1997. See Figure 17 for the drained water volume through time. 57 Table 8. 4,521 1' ml tie lml: Table 8. Water Volume Extracted From Mesocosm Water Total Water Water Time Hours . Volume Volume Exotracted as Date (hours) smce Extracted Extracted1 a /o of Total Inversion Mesocosm (ml) (ml) Volume 3/21/98 1330 70 6,045 6,045 3.12% 3/29/98 1200 260.5 80 6,125 3.17% 3/31/98 1000 306.5 25 6,150 3.18% 4/5/98 1230 429 130 6,280 3.25% 4/8/98 1200 500.5 30 6,310 3.26% 4/17/98 1600 720.5 70 6,380 3.30% 4/19/98 1200 764.5 40 6,420 3.32% 4/20/98 1000 786.5 10 6,430 3.32% 341/21/93 2100 822.5 2910 9,340 4.83% 4/21/98 1230 838 710 10,050 5.19% 4/21/98 2100 846.5 810 10,860 5.61% 4/21/98 1400 863.5 160 1 1 ,020 5.70% 1. Total volume of Mesocosm (193,479 cm3) = [Pi x Radius of Mesocosm (23.77 cm2) x Height of Mesocosm (109 cm)]. 2. Draining Began on 3/18/98 at 1530. 3. Argon gas hooked up to induce pressure (approximately 25 Kilopascals) on 4/21//98. ml denotes Milliliters. 1ml= 1cm3 58 Total Drained 1 Water (ml) 0 I,” ,_ 7 2-222--. _ L,4.___.___._. .__._._-.._ __. 4 0 Time (hours) 1.000 Figure 17. Total Drained Water. HYDRAULIC CONDQQIVIW As discussed in the methods section, five different methods were used to estimate the hydraulic conductivity of the mesocosm sediment. The methods used included: the slug test (Ewers, 1997), Darcy’s Law (Ewers, 1997), falling head permeameter, the Kozeny-Carmen estimation, the packed constant head permeameter and the unpacked constant head permeameter. As described earlier (Ewers, 1997) performed slug tests on all the wells. These data were analyzed using the method of Hvorslev (1951) (Table 9). These results appear reasonable, with the possible exception of the estimate at well 8. A hydraulic conductivity this high (59,616 cm/day) falls in the range of well-sorted gravel. See the Sedimentology section for more detail on the likely cause of this difference. 59 pres< app: estln likely the s Table 9. Slug Test Results (from Ewers, 97). WELL K (cm/sec) K (cm/day) K (ft/day) 2 2.50E-03 216 7.09 3 2.70E-03 233 7.65 4 2.80E-02 2,419 79.4 5 1.30E-03 112 3.69 6 1.30E-03 112 3.69 7 1.70E-03 147 4.82 8 6.90E-01 59,616 1,956 9 2.50E-03 216 7.09 10 7.40E-04 64 2.10 Avg. K 1.30E-02 1,123 36.9 1. Avg. K denotes average hydraulic conductivity of entire mesocosm calculated by Ewers, 1997 using Darcy’s Law. 2. K denotes hydraulic conductivity. The estimates obtained using the falling head permeameter method are presented in Appendix E. These estimates were all found to be up to approximately an order of magnitude higher than the hydraulic conductivity estimates obtained using the constant head methods (discussed below). It is likely that these estimates are high due to sample error resulting from disturbing the soil cores. Therefore, these data were not used for any further analyses. The hydraulic conductivity estimates generated using the Kozeny-Carmen equation are summarized in Table 10. Additionally, Figure 18 shows a histogram plot of the fine to very fine sand and silty sand measurements. Keeping in mind there were not enough measurements taken to adequately assess the normal distribution, some general trends were observed. The fine to very fine sand estimates appear to be fairly log-normally distributed while the silty sand estimates appear to have a slightly bimodal distribution. 60 Nutnber Ln (i Figu Plot sam Table 10. Kozeny-Carmen: Summary of Hydraulic Conductivity Estimates. Standard Number of Lithology Median Geomean Deviation Measurements Fine to Medium Sand 1 ,074 1 ,106 84 4 Silty Sand 6 8.0 6.7 10 Fine to Very Fine Sand 160 301 203 8 Clayey Silt NA NA NA 0 Silty Clay NA NA NA 0 Units in cm/day and NA denotes no samples available for estimate. 10 1O 3 8 2 5 E 5 2 5 ‘ -6 -5 -4 -3 -2 -11 -10 -9 -3 -7 Ln (Hydraulic COHdUCtiVitY. in cmlsec) Ln (Hydraulic Conductivity, in cmlsec) Figure 18. Histogram of Kozeny-Carmen Hydraulic Conductivity Estimates. Plot on left is fine to very fine sand samples and plot on the right is silty sand samples. The constant head permeameter hydraulic conductivity estimate results from intact cores collected from the mesocosm are summarized in Table 11. These estimates include both the few packed estimates along with the unpacked estimates. Both sets of data were found to be in good agreement and are presented in Appendix F. Additionally, Figure 19 shows a histogram plot of the fine to very fine sand and silty sand measurements. Both the fine to very fine sand and silty sand estimates appear to be fairly log-normally distributed. Table 12 shows these ranges compared to a published source (Fetter, 1994). 61 Table 11. Constant Head: Summary of Hydraulic Conductivity Estimates Standard Number of Lithology Median Geomean Deviation Measurements Fine to Medium Sand 750 211 832 4 Silty Sand 908 346 715 8 Fine to Very Fine Sand 661 62 101 7 Clayey Silt NA NA NA 0 Silty Clay NA NA NA 0 Units in cm/day and NA denotes no samples available for estimate. 10 10 - -__ _ a; 3 5 ‘5’ 5 g :1 z . . Z . ”.21 0 El :__»— " ":1“- o" “ 1‘12 11* -10 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 - -3 -2 Ln (Hydraulic Conductivity, in cm/sec) Ln (Hydraulic Conductivity, in cmlsec) Figure 19. Histogram of Constant Head Permeameter Hydraulic Conductivity Estimates. Plot on left is fine to very fine sand samples and plot on the right is silty sand samples. Table 12. Hydraulic Conductivity Comparisons Published Range of Calculated Range of Hydraulic Hydraulic Conductivities (Fetter, Lithology Conductivities 1994) Fine to Medium Sand 8 1,601 86.4 8,640 Silty Sand 20 1,724 0.86 8.64 Fine to Very Fine Sand 6 294 8.64 86.4 Clayey Silt NA NA 0.086 8.64 Silty Clay NA NA 8.64E-05 0.086 Units in cm/day and NA denotes no samples available for estimate. All of the constant head hydraulic conductivity estimates appear to be within the acceptable range with the exception of the silty sand estimates. This higher range most likely reflects the smaller scale heterogeneities in some of the silty sand sediment cores. These heterogeneities could have represented fine to medium sand lenses. Another possible explanation might be due to the sample disturbance associated with the sediment coring. For example, during sediment coring it is possible that artificial fractures or other high hydraulic conductivity zones could have been created. In addition to the measurements discussed above, over 15 samples had to be discarded dUe to refusal while coring, excessive leaking during the hydraulic conductivity measurement, or the inability to measure hydraulic conductivity due to high clay contents in the samples. EEECTIVE POROSJTY As described earlier, the effective porosity was estimated by measuring the volume of voids each sediment core contained and dividing by the total volume of the sediment core. The effective porosity results are summarized in Table 13. A correction factor was applied to all the data as a result of observations made during the saturated sediment core weighing. These observations noted that a maximum of 10 ml contained in the PVC endplate was erroneously weighed. Therefore, 10 ml was subtracted from the total water removed measurement to evaluate the effect of this error. See Appendix G for the rest of the effective porosity data. 63 Table 13. Summary of Effective Porosity Measurements Standard Number of Lithology Ragga Geomean Deviation Measurements Fine to Medium Sand 0.26 0.37 0.31 0.08 2 Silty Sand 0.29 0.41 0.35 0.04 10 Fine to Very Fine Sand 0.29 0.48 0.35 0.07 6 Clayey Silt NA NA NA NA 0 Silty Clay NA NA NA NA 0 NA denotes no samples available for estimate. GRAIN SIZE The grain size results are summarized in Table 14 with detailed data provided in Appendix H. A column listing the initial lithologic interpretation is also included to compare to the grain size results to show some of the changes that were made based on the grain size results. Samples showing disagreement between the initial lithologic interpretation and the grain size data are highlighted. For all other calculations and data reporting, the grain size estimate results were assigned as the most accurate way to determine lithologies. 64 Table 14. Summary of Grain Size Data Very Sample C223“ Medium Fine 12:1: 5111 Clay Lithology ID Coarse Sand Sand Sand Interpretation Sand A3a1 0.40 19.26 52.50 22.26 5.59 0.00 Fine to Medium Sand A3a2 0.50 18.22 48.55 26.33 6.41 0.00 Fine to Medium Sand A2b2 5.48 9.95 17.61 17.91 47.56 1.49 Silty Sand A3c3 5.70 9.29 15.48 19.78 47.75 2.00 Silty Clay D1b 3.84 8.17 11.42 18.75 32.83 25.00 Silty Clay D10 1.09 13.25 55.171708 5.08 8.33 Clayey Silt D2c 0.58 10.58 39.08 32.25 9.17 8.33 ClayeySilt Eb1 1.33 12.00 38.08 29.92 10.33 8.33 - Silty Sand Eb2 1.09 3.58 11.33 53.75 17.75 12.50 SiltySand Eb3 1.25 12.50 47.92 25.83 4.17 8.33 Silty Sand Eb4 0.58 9.58 31.67 36.08 5.42 16.67, ' Silty Sand Gb1 0.59 2.75 18.92 51.33 18.08 8.33 .. Silty.Sand' Gb2 1.08 18.08 562515.92 0.33 8.33 , Silty Sand ‘ Gb3 0.17 20.25 0.00 0.00 75.42 4.17 Silty Sand Hb1 1.50 9.17 0.00 0.00 81.00 8.33 Silty Sand Hb2 0.17 17.33 0.00 0.00 82.50 0.00 Silty Sand Ib1 0.83 6.92 0.00 0.00 83.92 8.33 Silty Sand lb3 0.75 0.00 0.00 0.00 95.08 4.17 Silty Sand Jc1 3.75 0.00 0.00 0.00 92.08 4.17 Fineto Medium Sand J02 3.00 0.00 0.00 0.00 97.00 0.00 Fine to Medium Sandi Kb1 1.58 0.00 0.00 0.00 81.751667 Silty Sand Kc2 3.09 0.00 0.00 0.00 96.92 0.00 Fine to Medium Sand Shaded cell represents possible erroneous visual interpretation. All values are in %. SEDIMENTOLOGY Digital images of the mesocosm were collected from each layer of exposed sediment to help identify its main hydrostratigraphic zonation (See Appendix A and Figure 20). Based on these images, five major units (Silty Sand, Silty Clay, Clayey Silt, fine to very fine Sand and fine to medium Sand) are 65 apparent. Table 15 summarizes descriptions and observations for each major lithologic unit. Even though there are lenses or smaller-scale units, these are the main units that appear to have significant continuity in the mesocosm. Grain size and hydraulic conductivity estimates suggest that some lithologies described as one unit, are most likely composed of multiple smaller scale units. For example, silty sand is a major unit, but if you look at some of the grain size results of cores taken in this unit, the results suggest the unit was more like a fine to very fine sand. Minor adjustments were made to initial lithology interpretations based on these types of observations. Therefore, within these five major units smaller sub- units are present. ' Clayye Silt W . nd . F-ne to Med‘um $3 ‘ J . . Silty Sand Silty Clay Figure 20. Digital Image of Layer N (Inflow End). Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. 66 Table 15. General Physical Descriptions of Sediment LITHOLOGY GENERAL DESCRIPTIONS . Light yellowish brown, SILT, some Clay, matrix Clayey 8'" supported, poorly sorted and massive. Mottled (in most areas) by light greenish grey and . . brownish yellow, CLAY, some Silt, trace fine Gravel S'lty Cflgéwfgvgfce 0f and root debris (in some areas), matrix supported, 9 , poorly sorted, massive. Gravel clasts are subangular— angular and Show no preferred orientation. Silty Sand Light brown, very fine to fine SAND, some to little Silt, well rounded to rounded, well sorted, massive. . . Brown to light brown yellow, fine to very fine SAND, Fine to Very Fine Sand well sorted, massive. Sand is well rounded to rounded. Fine to Medium Sand Brown to light brown yellow, fine to medium SAND, well sorted, massive. Sand is well rounded to rounded. Cla ye y Silt Unit This unit is found within all of the mesocosm layers, with a thickness (Y- direction) of 10 to 24 cm, and located nearest to the original collection surface. The unit is brownish yellow clayey Silt and is massive (no structures). Silty Clay with Trace of Fine Gravel Unit This unit is found within all of the mesocosm layers, with a thickness (Y-direction) between 13-31 cm in thickness, and located at varying depths throughout the mesocosm. The unit iS brownish yellow silty clay with a trace of fine gravel and extensive mottling (brownish yellow/light greenish gray). It is massive and 67 contains a trace amount of fine gravel ranging in size from <1 cm to 3 cm, with the dominant Size being <1cm. The clasts are mostly angular to sub-angular and with no apparent preferred orientation. Fine Silty Sand Unit This unit is only found within Layer A where it appears to grade into a fine to medium sand. The unit was ~ 12 cm thick and was found towards the bottom of the mesocosm at a depth of 30-42 cm. It is a dark light yellowish brown fine silty sand unit with well-rounded to rounded grains. In addition, the unit is well- sorted, massive and grain supported with no mud. However, some lenses of silty clay exist at various intervals. AS noted earlier, a parting was observed near well 8. This parting most likely was present before the mesocosm was collected as the slug test performed at well 8 Showed a hydraulic conductivity (53,616 cm/day or 1,956 feet/day) of one to two orders of magnitude higher than the other results. A hydraulic conductivity this high is characteristic of a well-sorted gravel, which was not observed to be present . Therefore, this parting was almost certainly present throughout all of the tracer tests. Fine to Very Fine Sand Unit This unit is found within layers A through D and then appears to pinch out by the Silty clay with a trace of fine gravel unit. The unit is thickest in layers A and B and becomes thinner until it pinches out. In addition, the unit is light 68 yellowish brown fine-to-medium massive, well—rounded-to-rounded, well-sorted sand that is grain-supported. Fine to Medium Sand Unit This unit iS found within layers A through D and appears to be pinched out by the silty clay with a trace of fine gravel unit. The unit is thickest in layers A and B and becomes thinner until it pinches out. In addition, the unit is a light yellowish brown fine-to-medium massive, well-rounded to rounded, well-sorted sand that is grain-supported. The bottom two layers, which are mainly composed of fine-to-medium Silty sand are very similar to each other and could possibly be the part of the same layer. It appears that both are well sorted. These are most likely associated with glaciofluvial processes and were part of debris flow deposit. The top layers appear to be part of a more fluidized depositional period. This iS mainly inferred from the mud matrix that can be found in both layers. Given the surficial geology of the area, and the characteristics of the all the deposits, the layers most likely represent multiple debris-flows. Some work (Lawson 1978; Hartshorn 1958; Boulton 1968) has documented Similar types of deposits, which have been interpreted mostly as sediment-flow deposits overlaying outwash deposits. These sediment-flow deposits appeared to be structureless at the macroscale. As mentioned above, there appear to be some discrepancies between the grain Size results and the visual interpretations that were initially made. Table 14 69 summarizes these results and identifies areas where errors appear to have occurred. The biggest discrepancies were in interpreting fine to very fine sand as clayey Silt or Silty sand. This is somewhat expected, as there is clearly smaller scale heterogeneity than what was characterized. The largest differences in interpretation and measured hydraulic properties were for samples J01, J02 and K02. These samples were interpreted as fine to medium sand, but the grain size Shows that they are more likely silty sand to silt. Looking more closely at the sample locations (Appendix A), it appears that the areas where these samples were taken are darker in color and could be localized Silty sand to silt units within the fine to medium sand unit. COMPUTER SIMULATIONS As discussed earlier, MODFLOW 2000 (Harbaugh, et al., 2000) was used to simulate groundwater flow conditions and RT3D version 2 (Clement, 1997) was used to simulate the tracer transport. Appendix I Shows the final interpretation of the lithologic units within the mesocosm and how they were represented in the model grid. Table 15 shows these final model parameters. These final model parameters were obtained by performing model: calibration through the use of all the collected physical and tracer measurements. 70 Table 16. Final Model Parameters Horizontal and Vertical Longitudinal Model Materials Hydraulic Dispersivity Porosity Conductivity (cm) (cm/day) . 48 Silty Sand (20 _ 1,724) 1.0 0.35 Fine to Medium Sand 1’601 0 2 0 31 (8 — 1,601) ' ' . . 80 Very Fine to Fine Sand (6 _ 294) 0.2 0.35 . 35 Clayey Silt (NA) 1.0 0.35 . 1.75 Silty Clay (NA) 0.2 ' 0.33 10,000 Open Area (NA) 0.2 0.38 . 10,000 Parting (NA) 0.7 0.38 Notes: 9391:8910 .‘l . Ranges below Horizontal and Vertical Hydraulic Conductivity values represent values obtained from constant head permeameter measurements. NA denotes no samples available for estimate. The open area column refers to the recessed portions of the endplates. Parting refers to parting observed in the vicinity of well 8. Specific storage and Specific yield values were all 0. Horizontal and vertical anisotropy values were all 1. Ratio of transverse dispersivity to longitudinal dispersivity and ratio of vertical dispersivity to longitudinal dispersivity were both 0.1. Effective molecular diffusion coefficient was 1.0E-06 cm2/sec. The measured head data and the observed flow gradient of 0.18, both obtained from Ewers (1997), were found to have some discrepancies as shown in Table 17. Based on the numerous flow Simulations using various hydraulic conductivity and porosity values, the value of 0.18 as stated in Ewers (1997) was used. Based on these data, the flow model appears to be simulating the conditions in the mesocosm with a mean error of 1.699 cm, a mean absolute 71 error of 1.935 and a root mean square error of 2.851 cm. In addition to this error summary, Figure 21 shows both the computed versus observed values plot and residual versus observed values plot. Theoretically, if the computed flow model were in perfect agreement with the observed conditions, all of the wells would be located directly on the line observed on both plots. These plots Show that the computed groundwater head values at all the wells, except for well 10, are higher than the observed head values. Wells 5 and 10 appeared to show the biggest disparity and appeared to be the only outliers. A possible explanation for this disparity could be due to the problems that MODFLOW 2000 had with the silty clay unit both of these wells were in. This silty clay unit represented a hydraulic conductivity 1 to 3 orders of magnitude lower than the rest of the lithologic units. Table 17. Measured Heads. Taken from Ewers, 1997. * denotes simulated heads were subtracted from 170 cm as this was the constant head used at the outflow. WELUSAMPLING Horizontal Distance Measured *Simulated PORT Gradient from Inflow Head Head Inflow - 0 100 - 2 0.32 38.5 87.5 93.8 3 0.37 38.5 85.8 92.9 4 0.34 38.5 86.9 93.5 5 0.50 38.5 80.8 90.4 6 0.19 76.5 85.6 92.8 7 0.21 76.5 84 92 8 0.19 76.5 85.7 92.9 9 0.19 76.5 85.7 92.9 10 0.24 76.5 81.3 90.7 11 0.35 108.5 62 81 12 0.35 108.5 62 81 13 0.35 108.5 62 81 14 0.35 108.5 62 81 15 0.35 108.5 62 81 Outflow 0.35 108.5 62 81 72 Computed vs. Observed Values 137 134 , 0W2 '0 o ’ E .3 183 ows a LEI E 132 owe U l : I ‘31 ovum 130 owr 17:9 123 - 1300 1825 issn Observed Residual vs. Observed Values 7 0mm 6 -I mm 5 E] 4 V , 0W2 — - III 3 3 ows “D .5 ° LIL] m 2 . owe ‘5 , III 1 ovum -n ' 0W? -1 -2 131D 1.5 1315 1315 132.0 132.5 Obsened Figure 21. Model Statistics. 73 III owe El DWI-14 III OW“ Ii] 0111.111“;J l3] DEN-15 IE] own [Z] 0W6 III owe El DWI-14 III OW“ [El Gilli-12 III OWIS EEI 0W13 CZ] 0W6 Next, the solute transport model was initialized using the flow model obtained from MODFLOW 2000 and all the input parameters discussed earlier in the methods section and in Table 16. Before each transport Simulation was initialized an observation well file was appended to the basic reactive transport package file (.btn) that allowed for the tracer concentration histories to be quantified. After each transport simulation was complete, this observation well file was imported into excel where the simulated tracer BTCS were plotted and compared to the observed tracer BTCS. These comparisons were made and were used as the basis to iteratively change the model parameters. The primary goal of this iterative exercise was to match the observed outflow tracer BTC to the simulated tracer BTC using realistic model parameters. As discussed earlier in this paper, the outflow endplate was designed so of the five equal recessed areas (125 cm2) provided a means to sample each area separately and also to sample them collectively as the outflow (Ewers, 1997). To ensure that the simulated tracer concentrations at each sampling port reflected the correct amount of concentration, the discharge at each sampling port was obtained from the flow model and plugged into Equation 9. The sum at each sampling port was added together to generate the outflow BTCS. Finally, a secondary goal was to match all of the well and sampling port observed tracer BTCS to the simulated tracer BTCs. The final result of all these iterations iS presented in figures 22-25. 74 Equation 9. Outflow Concentration Calculation (Ci, ---) O = Q# —— C# Qtotal Where: 00.8 = Outflow concentration contribution from each sampling port (units of milligrams/liter). Oct; = Cell discharge obtained from MODFLOW at each sampling port (units of cmS/day). C1: Concentration of tracer through time at each sampling port (units of milligrams/liter). Qtotaj = Sum of each sampling ports cell discharge obtained from MODFLOW (units of cm3/day). 75 Well 1 )00 NO DATA 0-25 0.07 0.03 ' Well 3 L ’ Well 2 t.- .r’lt‘l‘" - '1 ’1‘ ‘ ’ f“ ClCo 3.x ”I" #NJ..'.;I’;-’1e,'.‘~'v~'v. l'.' ...... :0 '. ..... "K 55:..."...."'“ n s ' "/. a-.. f... ’ " r,’.~o\—M~....w.,-. . . 0 O '0 ,t . 100 0:0 I! ‘ s" Time (hours) 100 Time (hours) Time (hours 0.005 Well 5 CICo 1 3 0 . . .l-M’M 4 2 0 Time (hours) 100 5 \ \\ + August1996Fluorescein -- September1997 Fluorescein Inflow .’,) Outflow + January1997 Tritium + August1998 Fluorescein -- Simulated Concentration Direction of flow Figure 22. Simulated and Measured Concentration Histories (BTCS). 76 0'60 Well 6 0 Time (hours) 100 0-09 0.8 0.14 Wel|9 j ”1* ., Well8 1‘5”" l i ' Iili ,’ . 11.-1“ F" 0100 ' 0100 CICo . ' - /; lit/X ‘ I; in“: 0 ‘ 'z‘ xvi-.4; .~ . in J" W 0 0 Time (hours) 100 0 Time (hours) 100 0.012 Well 10 l CICo g . ti . l 6 . . 1" NW 8 0 L, A W. 9 7 0 Time (hours) 100 1 \ \\ + August1996 Fluorescein # September 1997 Fluorescein Inflow Outflow + January 1997 Tritium + August 1998 Fluorescein Direction of flow — Simulated Concentration Figure 23. Simulated and Measured Concentration Histories (BTCs). 77 0'25 Sampling Port 11 0 3. .. .3. ~ 0 Time(hours) 100 0'25 Sampling Port 14 0'30 Sampling Port 13 0-30 Sampling Port 12 ._';., CICo CICo ClCo . _ - .vm\.. \~_' .1 f_' .. LM 0 . . “f 0 m. . ., '- Time (hours) 100 0 Time (hours) 100 Time (hours)100 Sampling Port 15 . ,. CICO . I. .1...\. \ .. '::"S. 6% ‘i.'-\ 4 PVC Outflow Plate 0 -." 0 Time (hours) 100 Sample PM 11 Sample Port 14 + August1996 Fluorescein A September 1997 Fluorescein + January1997 Tritium + August 1998 Fluorescein — Simulated Concentration Outflow Sample Port 13 Sample Port 12 Sample Port 15 Figure 24. Simulated and Measured Concentration Histories (BTCs). 78 0.25 ‘ CICo ‘ . t. o o . .fi‘Q "1 .~ ' 4, r s .0 \L.‘ ~e e ' .9 . ~.. 0 ° . "u. e . . ' .0- e . o 0., o. - e / . VTh o . 9’ '. 0 re ~ fiJ—‘nfi. I 0 Time (hours) 10° PVC Outflow Plate + August1996 Fluorescein -- September1997 Fluorescein + January 1997 Tritium + August 1998 F luorescein Sample Port11 . . Sample Port 14 — Simulated Concentration Outflow Sample Port 13 Sample P0" 12 Sample Port 15 Figure 25. Simulated and Measured Concentration Histories (BTCs). Looking at these plots, wells 8, 11, 12, 13, and the outflow appear to have simulated and observed BTCS that match fairly well. Wells 4, 7 and 10 appear to have simulated and observed BTCS that match the least, while the remaining wells had BTCs that matched fair to moderately well. Most importantly, the model adequately simulates the observed tracer BTC at the outflow. Referring to Table 16, a wide range of parameters can be seen along with the different lithologic units that were discretized into the model. 79 These parameters were the end result of an iterative exercise to obtain a reasonable match between the Simulated and observed tracer BTCS. This table also shows the range of dispersivity values that went into obtaining the final BTC set. It is important to note that all the values are below 1 cm. These higher values of 1 cm represented the silty sand unit and the clayey silt unit. Based on visual observations (Appendix A) these units also appeared to have areas throughout with smaller scale heterogeneities that were not included in the model or could not be identified. This observation would support the idea that was introduced at the beginning of this paper, which focuses on the relationship between heterogeneity and dispersivity. To restate this, Gelhar (1992) noted that the main reason why field scale dispersivities are several orders of magnitude greater than laboratory scale dispersivities for the same material is mainly due the influence of natural heterogeneities. So, dispersivity should be proportional to the degree of heterogeneity, so that a more heterogeneous sample would have a larger dispersivity than a less heterogeneous sample according to work by Gelhar (1992) and others. In this case, the predominant heterogeneity of the mesocosm sample was characterized and put into the model, thus the dispersivity is only representing the smaller scale heterogeneities within the described geologic units. Prior to this study fully characterizing the mesocosm heterogeneity, Ewers (1997) estimated the dispersivity to be 9 cm. His simulations were run using homogenous aquifer properties with the measured and simulated concentration histories matching very well. This value is 9 times the highest value that was 80 Jsed after the heterogeneity was more adequately described. To further Ilustrate the effect of dispersivity and molecular diffusion on the outflow BTC 'efer to Appendix J. 81 SUMMARY/CONCLUSIONS The characterization of small-scale heterogeneities at the centimeter to decimeter scale is an essential part of understanding solute transport. Many methods were used to provide high-resolution estimates of the mesocosm sediment. This characterization involved the following: Successful collection of a sub-meter scale, minimally disturbed, highly heterogeneous sediment sample Four conservative tracer tests with a sampling network of 15 locations under steady state conditions Visual interpretation of sediment heterogeneity Digital imaging of the exposed sediments to be used for computer model-grid creation and for reference Physical measurements (hydraulic conductivity, porosity and grain size) of the sediment resulting in estimates used for computer model calibration Computer model simulations of flow and tracer transport Four conservative tracer tests conducted between August 1996 and August 1998 provided a series of asymmetrically shaped BTCS. These asymmetries are most likely a result of the small-scale heterogeneities that exist within the mesocosm. This was confirmed by observing the different lithologic 82 units that each well was screened in and the overall observation of the complex heterogeneity within the mesocosm. Physical measurements were made on approximately thirty-one different sediment cores to estimate the physical parameters of the mesocosm. These values along with the digital images taken and visual observations made of the exposed sediment were used to create a computer model. The computer model was created with reference to the tracer, visual and physical data that was collected. All of these data were used to calibrate the flow and transport model and produce simulated BTCS. These Simulated BTCS were compared to the actual observed tracer BTCS and the differences, or residuals were calculated to estimate the model Simulations validity. Once the model adequately represented the average groundwater head distribution throughout the mesocosm, estimates of dispersivity and molecular diffusion were made. This approach is the first to: . Successfully collect sub-meter scale minimally disturbed highly heterogeneous sediment sample as demonstrated by tracer data results and visual observations. . Study effects of small-scale (centimeter to decimeter) heterogeneity on solute transport. . Study dispersivity in a highly heterogeneous controlled environment. Finally, this approach could be used to conduct multiple mesocosm studies and further evaluate any scaling effect that might occur with varying scales of study and degrees of heterogeneity.- 83 APPENDICES 84 APPENDIX A Sample Locations and Lithologic Interpretations 85 Clayey Silt a O W-l l Silty Clay A30 " w.14 _ . "3’? Ala I A2b2 i"; II. . ’ ‘A' ‘l '1 ‘9'). ,' I: Fine to Med. Sand --5» ’ . * -— ".0: 5" .2? . . ' '. Figure A-1. Digital Image of Layer A (Outflow End). Medium sized circles denote sediment core sample location and smaller circles (washers) denote well screen location. Each well is labeled as ‘W-#”. Lines represent interpreted lithologic boundary with the unit type named and underlined. B'Zc \\ Silty Sand B'r'b Silty Clay 9 Fineto Medium Sand '5! ‘,I .-.. Figure A-2. Digital Image of Layer B. Circles denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. 86 ' w. Clayey Silt SiltySand Clc I Fine to Very Fine Sand Figure A-3. Digital Image of Layer C. Circles denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. gm ClayeySilt Refusal ‘ Silty Sand " Fine to "Very Refusal I ch Fine-Sand Refusal Silty Clay Refusal Figure A-4. Digital Image of Layer D. Circles (plastic cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. 87 - . Clayey Silt 0W6 O Silty Sand Ebl 01m owe Figure A-5. Digital Image of Layer E. Large and medium circles (plastic and aluminum cores) denote sediment core sample location and smaller circles (washers) denote well screen location. Each well is labeled as “W-#”.. Lines represent interpreted lithologic boundary with the unit type named and underlined. I Fbl Silty Sand __._. -——. Silty Clay Figure A-6. Digital Image of Layer F. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. 88 Figure A-7. Digital Image of Layer G. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. A V. / I I: Figure A-8. Digital Image of Layer H. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. 89 Clayey Silt /’ Silty Sand / r” _- .’ / - 7 - d 3“: 7 . Refusal 81W Sand Silt-y Clay Figure A-9. Digital Image of Layer I. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic ‘16..- ‘ Figure A-10. Digital Image of Layer J. Large and medium circles (plastic and aluminum cores) denote sediment core sample location and smaller circles (washers) denote well screen location. Each well is labeled as ‘W-#".. Lines represent interpreted lithologic boundary with the unit type named and underlined. 90 Figure A-11. Digital Image of Layer K. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. - .-. J. " - ClayeySllt 5;, SiltySand {/l . ~/’ Let I / a q'ud Lc 1 b medium * ‘ _ '1 (1100“3 Refusal Silty Sand 4’4 Figure A-12. Digital Image of Layer L. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. 91 Clayey 3111 I“ ”/e‘ Silo/Sand - Mcl / Fra- - ”~’ 1.1.2 and 7k. “W nu - Figure A-13. Digital Image of Layer M. Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. Clayey Silt (“Id _ _ .-- "' 7 ~#.‘ .S’Gfléf— ” Nc2 . ‘fledtwfi W.— Silty Sand Ncl Silty Clay Figure A-14. Digital Image of Layer N (Inflow End). Circles (plastic and aluminum cores) denote sediment core sample location. Lines represent interpreted lithologic boundary with the unit type named and underlined. 92 Inflow Figure A-15. Index Figure To Cross-Section Images. Note that this view represents the position of the mesocosm after all the tracer tests were conducted. Tracer tests were conducted with the mesocosm parallel to the ground, and disassembly and sediment coring were conducted perpendicular to the ground (as shown). 93 Table A-1. Summary of Sampling Analysis. * denotes method used was improved on in later tests. Hydraulic Sample # Conductivity Porosity Grain Size A1a A2a A3a1 X A332 X ><><><><>< ><>< A20 ><><><><><><>< ><><><><><><><><><><><><><><><><><><><><><><><><><><><><>§><><- ><><><><>< ><><><><><><><.—_‘ ._. -.-. ”.— .-.__-._. -._. 400 0.25 ClCo 4 0 Time (hours) 00 Figure D-5. Well 6: Fluorescein BTC. 1996 data. 0.1 C/Co V\/\/\.‘/\f"\ 0 Time (hours) 400 Figure D-6. Well 7: Fluorescein BTC. 1996 data. 103 0.25 ClCo Time (hours) 400 Figure D-7. Well 8: Fluorescein BTC. 1996 data. 0.25 CICo HM O O!” l Time (hours) 400 Figure D-8. Well 9: Fluorescein BTC. 1996 data. 104 0.05 WW“? 0100 0 Time (hours) 400 Figure D-9. Well 10: Fluorescein BTC. 1996 data. 0.25 ClCo I l ‘1 O: M - 400 Time (hours) Figure D-10. Sampling Port 11: Fluorescein BTC. 1996 data. 105 0.25 f‘"““"’“"”""" -._.-..-.- w--- 0-....- - ._ ,-.- .--..._-_-- .. _ -- -- ._.. . .. ._.___..._,- a - ”WM"--- . _. _ ._ __.-.." _ _ __.- - -. - ClCo I I .... ——._...-—.__W~.—- WM“... .- 0 -_ W. _ 0 Time (hours) 400 Figure D-11. Sampling Port 12: Fluorescein BTC. 1996 data. 0.25 1, ClCo L l r!” 0 Time (hours) 400 Figure D-12. Sampling Port 13: Fluorescein BTC. 1996 data. 106 0.25 .~ «~--~~~ .___..-___,.- 1b 1b ; X x ..\/\ M: ‘ Time (hours) ‘p...--... .._.-. _ ..._._.._ . ......... .. ._ 400 C Figure D-13. Sampling Port 14: Fluorescein BTC. 1996 data. 0.25 t W--- - we? ‘1 , i ‘ O C/Co Time (hours) 400 FIQLI re D-14. Sampling Port 15: Fluorescein‘BTC. 1996 data. 107 0.25 .._.-...-...._--..-,.- W.-- W..- ,W----.-.-..-_ “pm..- __-—m--. C/Co ’ W o—J- ~-~—-- —-- -..- “ ‘ 4 0 Time (hours) 00 Figure D-15. Outflow: Fluorescein BTC. 1996 data. 108 APPENDIX E Falling Head Permeameter Data 109 Table E-1. Falling Head Permeameter Data. Change Sample Riser sage Hydraulic Average Sargple if? in Time Slope Length 0:21:22?” Inside Conductivity 02:33:23: ' (sec.) (cm) (cm.) Di;r;e;er (cm/sec) (cm/sec)“, A_1 a 0.0268 8 0.6 4.1 2.144E-01 2.06E-01 A 4 2.58 8 0.6 4.1 8 2.54 8 0.6 4.1 12 3.18 8 0.6 4.1 16 3.44 8 0.6 4.1 20 3.56 8 0.6 4.1 24 4.03 8 0.6 4.1 28 4.22 8 0.6 4.1 32 5.35 8 0.6 4.1 36 5.4 8 0.6 4.1 40 6.97 8 0.6 4.1 44 8.37 8 0.6 4.1 48 10.25 8 0.6 4.1 0.0265 2.120E-01 B 4 2.52 8 0.6 4.1 8 2.87 8 0.6 4.1 12 3.12 8 0.6 4.1 16 3.3 8 0.6 4.1 20 3.69 8 0.6 4.1 24 4.14 8 0.6 4.1 28 4.48 8 0.6 4.1 32 4.97 8 0.6 4.1 36 5.91 8 0.6 4.1 40 7.18 8 0.6 4.1 44 8.6 8 0.6 4.1 48 10.5 8 0.6 4.1 0.0258 2.064E-01 C 4 2.62 8 0.6 4.1 8 2.94 8 0.6 4.1 12 3.22 8 0.6 4.1 16 3.32 8 0.6 4.1 20 3.75 8 0.6 4.1 24 4.39 8 0.6 4.1 28 4.56 8 0.6 4.1 32 5.1 8 0.6 4.1 36 6 8 0.6 4.1 40 7.41 8 0.6 4.1 44 8.93 8 0.6 4.1 48 10.9 8 0.6 4.1 110 Sample am Ie Riser u H dr ulic Average Sample 339” 31"?33: Slope Eeng’th InSide lfsige C01¥du2tivity Hydraulic # (in.) (sec.) (cm) Diameter Diameter (c m/sec) Conductivity (cm.) (cm/sec) (cm.) 0.0247 1 .976E-01 D 4 2.77 8 0.6 4.1 8 2.91 8 0.6 4.1 12 3.29 8 0.6 4.1 16 3.42 8 0.6 4.1 20 3.81 8 0.6 4.1 24 4.4 8 0.6 4.1 28 4.74 8 0.6 4.1 32 6.14 8 0.6 4.1 36 5.17 8 0.6 4.1 40 7.46 8 0.6 4.1 44 8.99 8 0.6 4.1 48 10.95 8 0.6 4.1 0.025 2.000E-01 E 4 2.72 8 0.6 4.1 8 2.98 8 0.6 4.1 12 3.37 8 0.6 4.1 16 3.5 8 0.6 4.1 20 3.82 8 0.6 4.1 24 4.54 8 0.6 4.1 28 4.81 8 0.6 4.1 32 5.3 8 0.6 4.1 36 6.14 8 0.6 4.1 40 7.72 8 0.6 4.1 44 9.29 8 0.6 4.1 48 11.48 8 0.6 4.1 A_2a 0.0334 2.672E-01 2.165-01 A 4 2.27 8 0.6 4.1 8 2.63 8 0.6 4.1 12 3.09 8 0.6 4.1 16 3.1 8 0.6 4.1 20 3.59 8 0.6 4.1 24 4.19 8 0.6 4.1 28 4.43 8 0.6 4.1 32 5.18 8 0.6 4.1 36 6.21 8 0.6 4.1 40 8.29 8 0.6 4.1 44 12.62 8 0.6 4.1 48 21.78 8 0.6 4.1 0.0271 2.168E-01 B 4 2.47 8 0.6 4.1 111 Sample Chan e Sam Ie Riser Cu H draulic Average 53'2”” if? in Tinge Slope Length 0:23:25” Inslge Coxductivlty 62:33:32,122), (sec.) (cm) (cm.) Diameter (cm/sec) (cm/sec) (cm.) 8 2.97 8 0.6 4.1 12 3.25 8 0.6 4.1 16 3.22 8 0.6 4.1 20 3.89 8 0.6 4.1 24 4.37 8 0.6 4.1 28 4.57 8 0.6 4.1 32 5.14 8 0.6 4.1 36 6.02 8 0.6 4.1 40 7.54 8 0.6 4.1 44 9.24 8 0.6 4.1 48 12.26 8 0.6 4.1 0.0267 2.136E-01 C 4 2.47 8 0.6 4.1 8 2.97 8 0.6 4.1 12 3.47 8 0.6 4.1 16 3.53 8 0.6 4.1 20 4.03 8 0.6 4.1 24 4.62 8 0.6 4.1 28 4.9 8 0.6 4.1 32 5.45 8 0.6 4.1 36 6.35 8 0.6 4.1 40 7.85 8 0.6 4.1 44 9.69 8 0.6 4.1 48 12.64 8 0.6 4.1 0.023 1.840E-01 D 4 2.97 8 0.6 4.1 8 3.2 8 0.6 4.1 12 3.72 8 0.6 4.1 16 3.81 8 0.6 4.1 20 4.17 8 0.6 4.1 24 4.84 8 0.6 4.1 28 5.21 8 0.6 4.1 32 5.7 8 0.6 4.1 36 6.8 8 0.6 4.1 40 8.27 8 0.6 4.1 44 10.15 8 0.6 4.1 48 12.61 8 0.6 4.1 0.0246 1.968E-01 E 4 2.64 8 0.6 4.1 8 3.28 8 0.6 4.1 12 3.73 8 0.6 4.1 112 Sample Riser Av r Sample Step Change Sample Inside CUP Hydraulic Hydelraaugllec # (In.) 11 Time Slope Length Diameter Inside Conductlvlty C on d u ctivity (sec.) (cm) ( cm.) Diameter (cm/sec) (cm/sec) (cm.) 16 3.85 8 0.6 4.1 20 4.24 8 0.6 4.1 24 4.9 8 0.6 4.1 28 5.3 8 0.6 4.1 32 5.73 8 0.6 4.1 36 6.91 8 0.6 4.1 40 8.44 8 0.6 4.1 44 10.42 8 0.6 4.1 48 13.02 8 0.6 4.1 A_1 b 0.0016 1.280E-02 1.34E-02 A 3 32.54 8 0.6 4.1 6 36.14 8 0.6 4.1 9 40.17 8 0.6 4.1 12 43.89 8 0.6 4.1 15 45.26 8 0.6 4.1 18 49.53 8 0.6 4.1 21 56.21 8 0.6 4.1 24 62.9 8 0.6 4.1 27 66.89 8 0.6 4.1 30 72.16 8 0.6 4.1 33 80.02 8 0.6 4.1 36 91.46 8 0.6 4.1 39 108.65 8 0.6 4.1 42 124.66 8 0.6 4.1 45 140.76 8 0.6 4.1 48 172.84 8 0.6 4.1 B 3 38.1 0.0016 8 0.6 4.1 1.280E-02 6 42.32 8 0.6 4.1 9 47.66 8 0.6 4.1 12 54.2 8 0.6 4.1 15 58.91 8 0.6 4.1 18 67.32 8 0.6 4.1 21 77.68 8 0.6 4.1 24 90.74 8 0.6 4.1 27 97.94 8 0.6 4.1 30 1 15.93 8 0.6 4.1 33 144.77 8 0.6 4.1 36 190.9 8 0.6 4.1 39 263.71 8 0.6 4.1 42 323.78 8 0.6 4.1 45 426.04 8 0.6 4.1 113 Sample RI er Avera e Sample Step Change Sample Insslde CUP Hydraulic Hydraugllc # (ln.) nTlme Slope Length Diameter Inside Conductlvlty C on du ctl vIty (sec.) (cm) ( cm.) Diameter (cm/sec) (cm/sec) (cm.) 48 420.25 8 0.6 4.1 C 3 36.25 0.0017 8 0.6 4.1 1.360E-02 6 39.89 8 0.6 4.1 9 45.43 8 0.6 4.1 12 50.59 8 0.6 4.1 15 53.52 8 0.6 4.1 18 62.78 8 0.6 4.1 21 73.12 8 0.6 4.1 24 76.92 8 0.6 4.1 27 89.52 8 0.6 4.1 30 106.86 8 0.6 4.1 33 135.72 8 0.6 4.1 36 175.23 8 0.6 4.1 39 246.82 8 0.6 4.1 42 341.14 8 0.6 4.1 45 386.72 8 0.6 4.1 48 539.97 8 0.6 4.1 D 3 36.55 0.0018 8 0.6 4.1 1.440E-02 6 39.25 8 0.6 4.1 9 43.58 8 0.6 4.1 12 47.02 8 0.6 4.1 15 48.28 8 0.6 4.1 18 54.44 8 0.6 4.1 21 62.15 8 0.6 4.1 24 71.38 8 0.6 4.1 27 78.3 8 0.6 4.1 30 91.49 8 0.6 4.1 33 111.92 8 0.6 4.1 36 150.42 8 0.6 4.1 39 213.53 8 0.6 4.1 42 302.75 8 0.6 4.1 45 426.08 8 0.6 4.1 48 563 8 0.6 4.1 A_2c 0.0043 3.440E-02 2.53E-02 A 3 11 8 0.6 4.1 6 13.79 8 0.6 4.1 9 16.31 8 0.6 4.1 12 16.37 8 0.6 4.1 15 17.03 8 0.6 4.1 18 20.05 8 0.6 4.1 21 22.53 8 0.6 4.1 114 Sample Riser Avera e Sample Step Change Sample Inside CUP Hydraulic Hydraugllc # (I n.) nTIme Slope Length Diameter Inside Conductivity Con ductIvIty (sec.) (cm) ( 0 m.) Diameter (cm/sec) (cm/sec) (cm.) 24 25.51 8 0.6 4.1 27 26.39 8 0.6 4.1 30 27.76 8 0.6 4.1 33 30.73 8 0.6 4.1 36 35.58 8 0.6 4.1 39 41.48 8 0.6 4.1 42 48.12 8 0.6 4.1 45 55.08 8 0.6 4.1 48 65.21 8 0.6 4.1 B 3 15.15 0.0034 8 0.6 4.1 2.720E-02 6 16.59 8 0.6 4.1 9 18.63 8 0.6 4.1 12 19.91 8 0.6 4.1 15 19.14 8 0.6 4.1 18 21.48 8 0.6 4.1 21 24.3 8 0.6 4.1 24 27.42 8 0.6 4.1 27 28.81 8 0.6 4.1 30 31.37 8 0.6 4.1 33 34.18 8 0.6 4.1 36 39.74 8 0.6 4.1 39 46.18 8 0.6 4.1 42 52.39 8 0.6 4.1 45 60.16 8 0.6 4.1 48 72.4 8 0.6 4.1 C 3 18 0.0027 8 0.6 4.1 2.160E-02 6 20.2 8 0.6 4.1 9 22.09 8 0.6 4.1 12 23.37 8 0.6 4.1 15 23.78 8 0.6 4.1 18 25.8 8 0.6 4.1 21 29.07 8 0.6 4.1 24 32.34 8 0.6 4.1 27 34.05 8 0.6 4.1 30 36.24 8 0.6 4.1 33 39.2 8 0.6 4.1 36 44.47 8 0.6 4.1 39 52.13 8 0.6 4.1 42 59.69 8 0.6 4.1 45 67.01 8 0.6 4.1 48 80.85 8 0.6 4.1 115 Sample Riser Avera 0 Sample Step Change Sample Inside CUP Hydraulic Hydraugllc # (I n.) n Tlme Slope Length Diameter Inside Conductlvlty C on du ctlvlty (sec.) (cm) (cm.) Diameter (cm/sec) (cm/sec) (cm.) D 3 18.14 0.0027 8 0.6 4.1 2.160E-02 6 19.83 8 0.6 4.1 9 22.25 8 0.6 4.1 12 24.03 8 0.6 4.1 15 24.32 8 0.6 4.1 18 26.67 8 0.6 4.1 21 29.83 8 0.6 4.1 24 33 8 0.6 4.1 27 34.8 8 0.6 4.1 30 37.03 8 0.6 4.1 33 41.01 8 0.6 4.1 36 46.24 8 0.6 4.1 39 54.3 8 0.6 4.1 42 60.62 8 0.6 4.1 45 68.37 8 0.6 4.1 48 80.79 8 0.6 4.1 E 3 18.87 0.0027 8 0.6 4.1 2.160E-02 6 20.63 8 0.6 4.1 9 22.67 8 0.6 4.1 12 24.01 8 0.6 4.1 15 24.84 8 0.6 4.1 18 26.88 8 0.6 4.1 21 30.14 8 0.6 4.1 24 33.86 8 0.6 4.1 27 35.26 8 0.6 4.1 30 37.16 8 0.6 4.1 33 41.45 8 0.6 4.1 36 46.93 8 0.6 4.1 39 55.27 8 0.6 4.1 42 62 8 0.6 4.1 45 71.32 8 0.6 4.1 48 85.58 8 0.6 4.1 A_3c 0.0005 4.000E-03 3.47E-03 A 3 108.4 8 0.6 4.1 6 114.51 8 0.6 4.1 9 125.19 8 0.6 4.1 12 132.79 8 0.6 4.1 15 135.61 8 0.6 4.1 18 147.4 8 0.6 4.1 21 164.23 8 0.6 4.1 24 174.57 8 0.6 4.1 116 Sample Riser Avera e Sample Step Change Sample Inside CUP Hydraulic Hydralfilc # (In-I nTIme Slope Length Diameter D:nsldte Congsctlvlty Con ductlvlty (sec.) (cm) ( c m.) (acmnefr (0 sec) (cm/sec) 27 202 8 0.6 4.1 30 204.74 8 0.6 4.1 33 225.75 8 0.6 4.1 36 244.94 8 0.6 4.1 39 300.02 8 0.6 4.1 42 300.44 8 0.6 4.1 45 419.81 8 0.6 4.1 48 480.13 8 0.6 4.1 B 3 150.57 0.0004 8 0.6 4.1 3.200E-03 6 162.72 8 0.6 4.1 9 181.41 8 0.6 4.1 12 198.32 8 0.6 4.1 15 201.02 8 0.6 4.1 18 212.97 8 0.6 4.1 21 261.85 8 0.6 4.1 24 282.39 8 0.6 4.1 27 271.02 8 0.6 4.1 30 365.91 8 0.6 4.1 33 416.94 8 0.6 4.1 36 477.97 8 0.6 4.1 39 537.46 8 0.6 4.1 42 719.72 8 0.6 4.1 45 840.03 8 0.6 4.1 48 1080.14 8 0.6 4.1 C 3 127.53 0.0004 8 0.6 4.1 3.200E-03 6 136.37 8 0.6 4.1 9 150.61 8 0.6 4.1 12 156.09 8 0.6 4.1 15 168.59 8 0.6 4.1 18 179.36 8 0.6 4.1 21 202.12 8 0.6 4.1 24 225.04 8 0.6 4.1 27 235.63 8 0.6 4.1 30 219 8 0.6 4.1 33 299.8 8 0.6 4.1 36 300.23 8 0.6 4.1 39 360.2 8 0.6 4.1 42 479.58 8 0.6 4.1 45 480.3 8 0.6 4.1 48 659.65 8 0.6 4.1 117 APPENDIX F Constant Head Permeameter Data 118 Table F-1. Constant Head Permeameter Data. Avera 0 Sample # if: Discharge Area Length Height chgfi‘t'my Hydraugllc ( sec.) (ml/sec) (cm"2) (cm) (cm) ( cm] s e c) Conductivity (cm/sec) A_3a1 167.36 9.02E-01 33.18 6 9 1.81 E-02 1.85E-02 178.76 9.45E-01 33.18 6 9 1.90E-02 207.84 9.22E-01 33.18 6 9 1.85E-02 171.38 9.51 E-01 33.18 6 9 1.91 E-02 183.59 8.91 E-01 33.18 6 9 1.79E-02 A_3a2 157.54 7.92E-01 33.18 6 9 1.59E-02 1.60E-02 170.61 7.92E-01 33.18 6 9 1.59E-02 167.42 7.97E-01 33.18 6 9 1.60E-02 158.4 8.02E-01 33.18 6 9 1.61 E-02 164.45 8.02E-01 33.18 6 9 1.61E-02 Eb1 327 2.1 1 E-02 17.95 7.58 21 .1 4.22E-04 3.78E-04 426 1.85E-02 17.95 7.58 21.1 3.71 E-04 307 1.92E-02 17.95 7.58 21.1 3.85E-04 440 1.70E-02 17.95 7.58 21.1 3.41 E-04 304 1.84E-02 17.95 7.58 21.1 3.69E-04 Eb2 26047 3.16E-03 17.95 7.63 21 .6 6.22E-05 6.92E-05 3253 3.06E-03 17.95 7.63 21.6 6.02E-05 2083 4.32E-03 17.95 7.63 21.6 8.51 E-05 Eb3 564 8.69E-02 18.02 7.7 21 .3 1.74E-03 1.73E-03 587 8.67E-02 18.02 7.7 21.3 1.74E-03 615 8.60E-02 18.02 7.7 21.3 1.73E-03 573 8.59E-02 18.02 7.7 21.3 1.72E-03 555 8.56E-02 18.02 7.7 21.3 1.72E-03 Eb4 575 3.37E-02 18.28 5.18 21.8 4.39E-04 5.03E-04 347 4.03E-02 18.28 5.18 21.8 5.24E—04 288 4.17E-02 18.28 5.18 21.8 5.42E-04 485 3.90E-02 18.28 5.18 21.8 5.07E-04 622 3.86E-02 18.28 5.18 21 .8 5.02 E-04 D10 555 7.01 E-02 17.72 7.65 21.7 1.39E-03 1.34E-03 553 6.85E-02 17.72 7.65 21.7 1.36E-03 574 6.71 E02 17.72 7.65 21.7 1.33E-03 574 6.53E-02 17.72 7.65 21.7 1.30E-03 528 6.63E-02 17.72 7.65 21.7 1.32E-03 D20 598 3.68E-02 18.32 7.63 21 .8 7.03E-04 7.05E-04 339 3.83E-02 18.32 7.63 21.8 7.33E-04 505 3.66E-02 18.32 7.63 21.8 7.00E-04 544 3.58E-02 18.32 7.63 21.8 6.85E-04 D1b Too much clay; impermeable mo water after 7 hours) Fb1 Excessive leaking Fb2 Excessive leaking 119 Average Sample # :3: Discharge Area Length Height 02:35:23” Hydraulic (sec.) (ml/sec) (cm"2) (cm) (cm) (cm/sec) Conductlvlty (cm/sec) CM 451 1.86E-01 18.47 7.58 21.7 3525-03 3.40E-03 467 1.83E-01 18.47 7.58 21.7 3.47E-03 547 1.81E-01 18.47 7.58 21.7 3.42E-03 521 1.76E-01 18.47 7.58 21.7 3.32E-03 345 1.74E-01 18.47 7.58 21 .7 3.29E-03 Gb2 2351 4.34E-03 18.10 7.58 21.7 8.37E-05 8.94E-05 1549 5.04E-03 18.10 7.58 21.7 9.72E-05 1987 4.53E-03 18.10 7.58 21.7 8.74E-05 Hb1 1023 1 .17E-02 18.10 7.6 21.8 2.26E-04 2.26E-04 Hb2 477 243502 18.10 7.7 24 4.31 E-04 2.97E-04 522 2.22E-02 18.10 7.7 24 3.94E-04 664 1.87E-02 18.10 7.7 24 3.31 E-04 642 1.32E-02 18.10 7.7 24 2.35E-04 1166 1.07E-02 18.10 7.7 24 1.90E-04 680 1 .15E-02 18.10 7.7 24 2.03E-04 Gb3 236 3.60E-01 17.95 7.7 24 6.44E-03 6455-03 226 3.61 E01 17.95 7.7 24 6455-03 240 3.56E-01 17.95 7.7 24 6.37E-03 264 3.63E-01 17.95 7.7 24 6.49E-03 198 3.64E-01 17.95 7.7 24 6.50E-03 Ib3 109 8.62E-01 18.10 7.4 22.6 1.56E-02 1.56E-02 105 8.62E-01 18.10 7.4 22.6 1.56E-02 98 8.57E-01 18.10 7.4 22.6 1.55E-02 99 8.64E-01 18.10 7.4 22.6 1.56E-02 55 8.73E-01 18.10 7.4 22.6 1.58E-02 Jb1 573 1.05E-01 18.10 7.65 21.8 2.03E-03 1.77E-03 562 9.96E-02 18.10 7.65 21.8 1.93E—03 580 9.43E-02 18.10 7.65 21.8 1.83E-03 595 8.66E-02 18.10 7.65 21.8 1.68E-03 587 8.09E-02 18.10 7.65 21.8 1.57E-03 584 8.05E-02 18.10 7.65 21.8 1.56E-03 Jb2 Excessive leaking J01 88 1.04E+00 18.10 7.63 21.8 2.01 E-02 2.00E-02 90 1.03E+00 18.10 7.63 21.8 1.99E-02 93 1.03E+00 18.10 7.63 21.8 2.00E-02 85 1.03E+00 18.10 7.63 21.8 1.99E-02 83 1.03E+00 18.10 7.63 21.8 1.99E-02 Jc2 98 9.29E-01 17.80 7.6 21.9 1.81E-02 1.78E-02 102 9.17E-01 17.80 7.6 21.9 1.79E-02 89 9.16E-01 17.80 7.6 21.9 1.79E-02 98 9.08E-01 17.80 7.6 21.9 1.77E-02 69 9.06E-01 17.80 7.6 21.9 1.77E-02 120 Average Sample # 1.7:; Discharge Area Length Height nggtriactt'lltlrfity Hydraulic ( s e c.) (ml/sec) (cm"2) (cm) (cm) ( cm] s e c) Conductlvlty (cm/sec) Blank 4 1.75E+01 17.80 7.6 21.9 3.41E-01 3.78E-01 4 1.90E+01 17.80 7.6 21.9 3.70E-01 2 2.18E+01 17.80 7.6 21.9 4.24E-01 2 2.23E+01 17.80 7.6 21.9 4.34E-01 3 1.65E+01 17.80 7.6 21.9 3.22E-01 Kb1 503 1.11E-01 17.80 7.6 22 2.16E-03 1.90E-03 529 1.03E-01 17.80 7.6 22 2.00E-03 494 9.62 E02 17.80 7.6 22 1 .87E-03 587 9.1 1 E-02 17.80 7.6 22 1.77E-03 439 8.88 E-02 17.80 7.6 22 1 .72E-03 K02 1 18 7.58E-01 18.10 7.6 22 1.45E-02 1.46E-02 87 7.59E-01 18.10 7.6 22 1.45E-02 69 7.68E-01 18.10 7.6 22 1.47E-02 123 7.56E-01 18.10 7.6 22 1.44E-02 38 7.76E-01 18.10 7.6 22 1.48E-02 121 APPENDIX G Porosity Data 122 Table G-1. Porosity Data. Saturated Oven Dried Total in m: 3123'; 33351: 223.521.." 3113;: 1%? LEE. nag. 1,513.13. . .1311); (am) * 1:311“ (am) ”132,? (gm) 9 (°m“3l ° Eb1 220.7 466.62 245.92 413.61 192.91 53.01 7.58 2.39 136.02 0.39 Eb2 238.55 482.93 244.38 423 184.45 59.93 7.63 2.39 136.92 0.44 Eb3 221.16 472.3 251.14 414.05 192.89 58.25 7.7 2.4 139.34 0.42 Eb4 227.71 394.53 166.82 339.28 111.57 55.25 5.18 2.41 94.52 0.58 le1 183.8 445.7 261.9 396.53 212.73 49.17 7.58 2.39 136.02 0.36 le2 165.88 411.88 246 366.19 200.31 45.69 7.7 2.4 139.34 0.33 [Hb1 154.9 417.4 262.5 365.59 210.69 51.81 7.65 2.4 138.43 0.37 le2 151.2 413.2 262 360.94 209.74 52.26 7.65 2.38 136.13 0.38 16133 169.2 423.7 254.5 374.07 204.87 49.63 7.7 2.39 138.18 0.36 lb3 169.3 431 261.7 374.99 205.69 56.01 7.4 2.4 133.91 0.42 Jb1 155 418.5 263.5 352.78 197.78 65.72 7.65 2.4 138.43 0.47 Kb1 165.9 420.4 254.5 358.39 192.49 62.01 7.6 2.38 135.24 0.46 DI b 225 502.58 277.58 445.07 220.07 57.51 7.65 2.4 138.43 0.42 D1 0 231.83 468.49 236.66 1’ 407.5 175.67 60.99 7.65 2.38 136.13 0.45 D20 215.65 451.22 235.57 390.34 174.69 60.88 7.63 2.42 140.38 0.43 J01 151.2 411.7 260.5 349.93 198.73 61.77 7.63 2.4 138.07 0.45 J02 183.8 452.9 269.1 387.95 204.15 64.95 7.6 2.38 135.24 0.48 K62 103.3 368.4 265.1 310.29 206.99 58.1 1 7.6 2.4 137.53 0.42 123 APPENDIX H Grain Size Data 124 Cumulative percent finer than (%) o 1 2 3 4 5 6 7 a 9 10 Grain Size (phi mlts) Figure H-1. Sample A3a1 Grain Size. Dashed line indicates inferred median grain size. 100 Cumulative percent finer than (%) o 1 2 3 4 6 6 7 8 5 10 Grain Size (phi mlts) Figure H-2. Sample A3a2 Grain Size. Dashed line indicates inferred median grain size. 125 f8‘ g \ 5 \ 5 \ is \ c l: ‘5 a 60 -------------- f. a, 2 > i =1 . .9 . ~. 3 t E : :I 9 ° : o 4'00: \\\\\ R; o 1 2 3 4 5 6 7 8 9 10 Grain Size (phi units) Figure H-3. Sample A363 Grain Size. Dashed line indicates inferred median grain size. A 100. 22‘; 5 5 is C u: ‘5 g. 50 . I \ I ‘~ 0 r 5 ' ...... .m i 3 r E . 3 I o r I o 3.97: N. ; 0 1 2 3 4 5 6 7 8 9 10 Grain Size (phi units) Figure H-4. Sample A2b2 Grain Size. Dashed line indicates inferred median grain size. 126 100 v ——9. \“ Cumulative percent flner than (96) o 1 2 3 4 5 6 7 8 9 10 Grain Size (phl units) Figure H-5. Sample Eb1 Grain Size. Dashed line indicates inferred median grain size. A 100 \\ €91. \\ \ 5 5 as g . .5. \1 2 5° __________ . l g E \ '0 . 1! . a : a. 3 I o : R‘fi~x : . o l 2.32: = o 1 2 3 4 5 6 7 8 9 10 Grain Slze (phi units) Figure H-6. Sample Eb2 Grain Size. Dashed line indicates inferred median grain size. 127 § 1 ‘\ x \ ‘\ \ Cumulative percent finer than (%) OI O 3 4 5 6 7 8 9 Grain Size (phi unlts) Figure H-7. Sample Eb3 Grain Size. Dashed line indicates inferred median grain size. 100 Cumulative percent finer than (%) 13.20 4 5 6 7 8 9 Grain Size (phi units) 10 Figure H-8. Sample Eb4 Grain Size. Dashed line indicates inferred median grain size. 128 Cumulative percent finer than (%) o 1 2 3 4 5 6 7 3 9 10 Grain Size (phi units) Figure H-9. Sample D1c Grain Size. Dashed line indicates inferred median grain size. § Cumulative percent finer than (%) 8 3.00 w ---------- 4 5 6 7 8 9 10 Grain Size (phi mits) Figure H-10. Sample D2c Grain Size. Dashed line indicates inferred median grain size. 129 A 100. 2“; 5 5 is C I: E i 50 """"""""""" I o I > I a I .1! I 3 I E l 3 I 0 I I O ”5.00 0 1 2 3 4 5 6 7 8 9 10 Grain Size (phi units) Figure H-11. Sample D1b Grain Size. Dashed line indicates inferred median grain size. “1(1) 5‘; 5 5 5 C t): '5 3 5° """"""" . I 0 I g I E I I 3 I E i 3 I O l I I O 1 2 3 4 5 6 7 8 9 10 Grain Size (phi units) Figure H-12. Sample Gb1 Grain Size. Dashed line indicates inferred median grain size. 130 100 A E 5 £ 25 I: t g m --------- I g 1 a: I _g I 3 l l 5 : \ o : \\ - 0 12.57 \_ o 1 2 3 4 5 6 7 8 9 10 Grain Size (phi mits) Figure H-13. Sample Gb2 Grain Size. Dashed line indicates inferred median grain size. ,. \ i» i Cumulative percent finer than (%) O 5.60 0 1 2 3 4 5 6 7 8 9 10 Grain Size (phi mits) Figure H-14. Sample Gb3 Grain Size. Dashed line indicates inferred median grain size. 131 § 5 \ - 5 5 5 c I: ‘5 i 50 I— -------------------- I o i .3 . E I = i E . 3 I o I I I o 5.98I _ o 1 2 3 4 5 6 7 8 9 10 Grain Size (phi mits) Figure H-15. Sample Hb1 Grain Size. Dashed line indicates inferred median grain size. A 100 5‘3. 5 5 as I: II: E i 60 -------------------- E .1! :I E a 0 O 5.59 0 1 2 3 4 5 6 7 8 9 10 Grain Size (phi mits) Figure H-16. Sample Hb2 Grain Size. Dashed line indicates inferred median grain size. 132 1w 25’ 5 5 B C C ‘5 3 5° """"""""""" 1 o i E i E i 3 i E . 0 i I 0 6.001 _ o 1 2 3 4 5 6 7 8 9 10 Grain Size (phl mits) Figure H-17. Sample lb1 Grain Size. Dashed line indicates inferred median grain size. 1oo - - - Cumulative percent finer than (%) 6.05 0 1 2 3 4 5 6 7 8 9 10 Grain Size (phi unlts) Figure H-18. Sample Ib3 Grain Size. Dashed line indicates inferred median grain size. 133 100 ..x‘ Cumulative percent finer than (%) -‘-----—-- I 0 6.001 0 1 2 3 4 5 6 7 8 Grain Size (phl units) Figure H-19. Sample Jc1 Grain Size. Dashed line indicates inferred median grain size. 100 ‘H—A - - . Cumulative percent finer than (%) O 5.97 0 1 2 3 4 5 Grain Size (phi units) Figure H-20. Sample Jc2 Grain Size. Dashed line indicates inferred median grain size. 134 0) q G) 100 Cumulative percent finer than (%) 0 1 2 3 4 5 6 7 Grain Size (phi units) Figure H-21. Sample Kb1 Grain Size. Dashed line indicates inferred median grain size. ..-. 100 Hi. . g - 53 I: III 5 5 I: a '52 i 50 --------------------- I I 0 I .3 I E I a I E I I 3 i O I I 0 5.99: _ . 0 1 2 3 4 5 6 7 8 9 10 Grain Size (phi units) Figure H-22. Sample Kc2 Grain Size. Dashed line indicates inferred median grain size. 135 Table H-1. Summary of Grain Size Percentages. Sand Percentagfi Total VCS CS MS FS VFS Sand Silt Clay Sam pie °/o °/o °/o °/o °/o % % °/o A3a1 0 0.40 19.26 52.50 22.26 94.41 5.59 0.00 A3a2 0.10 0.40 18.22 48.55 26.33 93.59 6.41 0.00 A3c3 2.60 3.10 9.29 15.48 19.78 50.25 47.75 2.00 A2b2 2.69 2.79 9.95 17.61 17.91 50.95 47.56 1.49 Eb1 0.33 1.00 12.00 38.08 29.92 81.33 10.33 8.33 Eb2 0.17 0.92 3.58 11.33 . 53.75 69.75 17.75 12.50 Eb3 0.42 0.83 12.50 47.92 25.83 87.50 4.17 8.33 Eb4 0.08 0.50 9.58 31.67 36.08 77.92 5.42 16.67 D1 0 0.67 0.42 13.25 55.17 17.08 86.58 5.08 8.33 D2c 0.08 0.50 10.58 39.08 32.25 82.50 9.17 8.33 D1 b 1.67 2.17 8.17 11.42 18.75 42.17 32.83 25.00 Gbi 0.17 0.42 2.75 18.92 51 .33 73.58 18.08 8.33 Gb2 0.50 0.58 18.08 56.25 15.92 91.33 0.33 8.33 Gb3 0.00 0.17 20.25 0.00 0.00 20.42 75.42 4.17 Hb1 0.42 1.08 9.17 0.00 0.00 10.67 81.00 8.33 Hb2 0.17 0.00 17.33 0.00 0.00 17.50 82.50 0.00 Ib1 0.00 0.83 6.92 0.00 0.00 7.75 83.92 8.33 lb3 0.17 0.58 0.00 0.00 0.00 0.75 95.08 4.17 Jc1 1.50 2.25 0.00 0.00 0.00 3.75 92.08 4.17 Jc2 1.67 1.33 0.00 0.00 0.00 3.00 97.00 0.00 Kb1 1.08 0.50 0.00 0.00 0.00 1.58 81.75 16.67 K02 1.42 1.67 0.00 0.00 0.00 3.08 96.92 0.00 VCS denotes very coarse, CS denotes coarse sand, MS denotes medium sand, FS denotes fine sand and VFS denotes very fine sand. 136 APPENDIX I Cross-Sections of Final Lithologies and Model Grid 137 .1. Legend Y I Silty Sand fl Silty Clay [I Fine to Medium Sand I Clayey Silt I Open Area Fine to Very Fine Sand Figure l-1. Three-Dimensional Cross-Section #1. 138 J Legend I Silty Sand Silty Clay I:| Fine to Medium Sand I Clayey Silt I Open Area Fine to Very Fine Sand Figure l-2. Three-Dimensional Cross-Section #2. 139 .1; X Legend I Silty Sand n Silty Clay I: Fine to Medium Sand I Clayey Silt I Open Area 1:] Fine to Very Fine Sand Figure l-3. Three-Dimensional Cross-Section #3. 140 Ti... Legend I Silty Sand n Silty Clay L—J Fine to Medium Sand I Clayey Silt I Open Area Fine to Very Fine Sand Figure I-4. Three-Dimensional Cross-Section #4. 141 APPENDIX J Model Simulation Sensitivities 142 +Dispersivity *Dispersivity Final Model * Molecular Molecular Diffusion Diffusion Figure J-1. Summary of Model Simulation Sensitivities. Final model lithologic dispersivity values are as follows (units of cm/day): silty sand (1 .0), fine to medium sand (0.2), very fine to fine sand (0.2), clayey silt (1 .0), silty clay (0.2), open area (0.2) and parting (0.7). Final model lithologic molecular diffusion values are all equal to 1.0 x 10 . Arrows represent increasing or decreasing parameter by one order of magnitude. 143 REFERENCES Benson, 8., 1988, Characterization of the Hydrologic and Transport Properties of the Shallow Aquifer under Kesterson Reservoir, Merced County, California [Ph. D. thesis]: University of California-Berkeley, 341 p. Berkowitz, B, Scher, H and Silliman, S., 2000, Anomalous transport in laboratory- Scale, heterogeneous porous media: Water Resources Research, v. 36(1), p. 149-158. Blaunch, D.N., 2000, Beer-Lambert Law, Mpzl/wwwchmgavidson.edu_/ChemistrvAppIets/spectrophotometrv/Unkno wnSolution.html Boggs, J.M. and Adams, BE, 1992, Field Study of Dispersion in a Heterogeneous Aquifer: 4. Investigation of Adsorption and Sampling Bias: Water Resources Research, v. 28, p 3325-3336. Boulton, G., 1968, Flow tills and related deposits on some Vestsitsbergen Glaciers: Journal of Glaciology, v. 7, p. 391-412. Burr, D.T., Sudicky, EA. and Naff, R.L., 1994, Nonreactive and reactive solute transport in three-dimensional heterogeneous porous media: Mean displacement, plume spreading, and uncertainty, Water Resources Research, v. 30(3), p. 791-815. Clement, T.P., 1997, RT3D (version-2) Update Document: Battelle Pacific Northwest National Laboratory, 59 p. Clement, T.P., 1997, RTsD (version-1) A Modular Computer Code for Simulating Reactive Multispecies Transport in 3-Dimensional Groundwater Systems: Battelle Pacific Northwest National Laboratory, 10 p. Drever, J.l., 1997, The Geochemistry of Natural Waters-Surface and Groundwater Environments-Third Edition: New Jersey, Prentice- Hall, lnc.. Ewers, K., 1997, Three Dimensional Groundwater Flow and Contaminant Transport in a Medium Scale Highly Heterogeneous Environment [MS thesis]: Michigan State University, 76 p. Fetter, C.W., 1994, Applied Hydrogeology—Third Edition: New Jersey, Prentice- Hall, lnc.. 144 Fetter, C.W., 1999, Contaminant Hydrogeology-Second Edition: Upper Saddle River, New Jersey, Prentice Hall. Freeze, RA. and J.A Cherry, 1979, Groundwater: Englewood Cliffs, New Jersey, Prentice-Hall. Garabedian, SP, 1989, Changes in Skewness and Kurtosis Over Time for a Natural-Gradient Tracer Test on Cape Cod, Massachusetts: Oral Presentation: Tracers in Hydrogeology: Principles, Problems, And Practical Applications-NWWA Annual Meeting and Exposition Educational Program— Houston, Texas- October 30-November 1. Gelhar, L.W., Welty, C. and Rehfeldt, KR, 1992, A Critical Review of Data on Field-Scale Dispersion in Aquifers: Water Resources Research, v. 28(7), p. 1955-1974. Gelhar, L.W., 1986, Stochastic Subsurface Hydrology From Theory to Application: Water Resources Research, v. 22(9), p. 1358-1458. Grisak, G.E., Pickens, J.F. and Cherry, J.A., 1980, Solute Transport Through Fractured Media - 2. Column Study of Fractured Till: Water Resources Research, v. 16(4), p. 731-739. Greenkorn, RA and DP. Kessler, 1969, Dispersion in heterogeneous nonuniform anisotropic porous media: Ind. Eng. Chem., v. 61, pp. 14-32. Hall, S.H., 1996, Practical Single-Well Tracer Methods for Aquifer Testing, Workshop Notebook: Tenth National Outdoor Action Conference and Exposition, May 13-15, 1996, Las Vegas, Nevada. National Ground Water Association, Columbus, Ohio, USA. Harbaugh, A.W., Banta, E.R., Hill, MO. and McDonald, MG, 2000, Modflow- 2000, The US. Geological Survey Modular Ground-Water Model - User Guide to Modularization Concepts and the Ground-Water Flow Process: US. Geological Survey Open-Report 00-92, 121 p. Hartshorn, M., 1958, Flowtill in southeastern Massachusetts: Geological Society of America Bulletin, v. 69, p. 477-482. Harvey, R.W., Voss, CL, and Souza, W.R., 1989, Comparison of Bacterial and Solute Transport Through Structured and Unstructured Porous Material: Oral Presentation: Tracers in Hydrogeology: Principles, Problems, And Practical Applications-NWWA Annual Meeting and Exposition Educational Program- Houston, Texas- October 30-November 1. 145 Hess, K.M., Wolf, SH. and MA. Celia, 1992, Large-Scale Natural Gradient Tracer Test in Sand and Gravel, Cape Cod, Massachusetts. 3. Hydraulic Conductivity Variability and Calculated Macrodispersivities, v. 28(8), p. 2011-2027. Hornberger, G.M., Mills, AL. and Herman, J.L, 1992, Bacterial Transport in Porous Media: Evaluation of a Model Using Laboratory Observations: Water Resources Research, v. 28 (3), p. 915-938. Hotzl, H. and Werner, A., 1992, Tracer Hydrology: proceedings of the 6th International Symposium on Water Tracing, Karlsruhe, Germany, 21-26 September: Brookfield, VT, Rotterdam, 464 p. Hvorslev, M.J., 1951, Time Lag and Soil Permeability in Groundwater Observations: WatenNays Exp. Sta. Bull-Vicksburg, Mississippi, US. Army Corps of Engineers. Levy, BS. and Chambers, RM, 1997, Bromide as a Conservative Tracer for Soil-Water Studies: Hydrological Processes, v. 1, p. 385-389. Lindstrom, FT. and Boersma, L., 1971, A Theory on the Mass Transport of Previously Distributed Chemicals in a Water Saturated Sorbing Porous Medium: Soil Science, v. 111 (2), p. 192-199. Loschiavo, T.,2003. Beer-Lambert Law, mpzl/www.holyghostmorg/facuIN/Los/HonghenVLAbs/beer.htm Mackay, D.M., Fryberg, D.L., Roberts, RV. and Cherry, J.A., 1986, A Natural Gradient Experiment on Solute Transport in a Sand Aquifer. 1. Approach and Overview of Plume Movement, Water Resources Research, v. 22(13), p. 2017-2029. Mull, D.S., Smoot, J.L. and Liebermann, TD, 1988, Dye Tracing Techniques Used To Determine Ground-Water Flow In A Carbonate Aquifer System Near Elizabethtown, Kentucky, US. Geological Survey: Water-Resources Investigations Report 87-4174, 95 p. Netter, R. and Behrens, H., 1992, Application of Different Water Tracers for Investigation of Constructed Wetlands Hydraulics: Tracer Hydrology, p 125-129. Passioura, J.B., 1971, Hydrodynamic Dispersion in Aggregated Media: 1. Theory: Soil Science, v. 111 (6), p. 339-344. 146 Petrich, CR, 1998, Encapsulated Cell Bioremediation: Evaluation on the basis of Particle Tracer Tests, Groundwater, v. 36(5), p. 771-778. Puzio, M. and Larson, 6., 1982, Map of Surficial Geology of Ingham County. Rao, P.S.C, Rolston, D.E., Jessup, RE. and Davidson, J.M., 1980, Solute Transport in Aggregated Porous Media: Theoretical and Experimental Evaluation: Soil Science Society of America Journal, v. 44, p. 1139-1145. Rose, DA. and Passioura, J.B., 1971, The Analysis of Experiments on Hydrodynamic Dispersion: Soil Science, v. 111 (4), p. 252-257. Saiers, J.E., Hornberger, GM. and Liang, L., 1994, First— and second-order kinetics approaches for modeling the transport of colloidal particles in porous media: Water Resources Research, v. 30 (9), p. 2499-. Schulze-Makuch, D., Carlson, D.A., Cherkauer, 0.8. and Malik, P. , 1999, Scale Dependency of Hydraulic Conductivity in Heterogeneous Media: Groundwater, v. 37(6), p. 904-919. Smart, RA. and l.M.S Laidlaw, 1977, An Evaluation of Some Fluorescent Dyes for Water Tracing: Water Resources Research, v. 13(1),15-33. Silliman, SE, 2001, Laboratory study of chemical transport to wells within heterogeneous porous media: Water Resources Research, v. 37 (7), p. 1883-1892. Silliman, S.E., Dunlap, R., Fletcher, M. and Schneegurt, MA, 2001, Bacterial transport in heterogeneous porous media: Observations from laboratory experiments: Water Resources Research, v. 37(11), p. 2699-2707. Sibert, G.,2004. Chemistry: Experiment 11—Beer’s Law, mpz/lwwwchem.vt.edu/RVGS/ACT/lab/Experiments/Exp 1 1- _Beer’s_l__aw.html. Sugita, F. and Gillham, R.W., 1995, Pore scale variation in retardation factor as a cause of nonideal reactive breakthrough curves: Water Resources Research, v. 31(1), 121-128. Thorbjarnarson, K.W. and Mackay, OM, 1994, A forced-gradient experiment on solute transport in the Borden aquifer. 2. Transport and dispersion on the conservative tracer, Water Resources Research, v. 30(2), 385-399. Tissue, B.,2000. Beer-Lambert Law, _h_ttp://www.chem.vt.edu/chem-ed/spec/beerslaw.html. 147 It. ' Ursino, N., Gimmi, T. and Fluhler, H., 2001, Combined effects of heterogeneity, anisotropy, and saturation on steady state flow and transport: A laboratory sand tank experiment: Water Resources Research, v. 37(2), p. 201-208. US. Department of Agriculture, 1979, Soil Survey of Ingham County, Michigan. Webster, DA, 1996, Results of Ground-Water Tracer Tests Using Tritiated Water at Oak Ridge National Laboratory, Tennessee, US. Geological Survey: Water-Resources Investigations Report 95-4182, 49 p. Zheng, C. and Wang, PP, 1999, MT3DMS: A Modular Three-Dimensional Multispecies Transport Model for Simulation of Advection, Dispersion, and Chemical Reactions of Contaminants in Groundwater Systems; Documentation and User’s Guide: Contract Report SERDP-99-1,U.S. Army Engineer Research and Development Center, Vicksburg, MS, 169 p. 148 I111111111111111111»