The Geochemical Evolution of Groundwater and SurfaceWater in a Small Glaciated Basin Involving Effluents from
Iron Mining: Hydrologic and Geochemical Constraints
ByEric A. Roth
A Thesis
Submitted toMichigan State University
in partial fulfillment of the requirementsfor the degree of
Masters of Science
Department of Geological Sciences
1992
Abstract
The Geochemical Evolution of Groundwater and SurfaceWater in a Small Glaciated Basin Involving Effluents from
Iron Mining: Hydrologic and Geochemical Constraints
By
Eric Allyn Roth
The impact of iron mining on the Sands Plain aquifer inMarquette County. Michigan allows for a unique opportunity tostudy the geochemical evolution of a groundwater-streamflowsystem in a small glaciated basin. Groundwater and surfacewater chemistry and numerical flow modeling techniques indicatethat effluent water from mining operations affect surface waterand groundwater in the northwestern portion of the study area.Groundwater and surface water indicate elevated concentrationsof calcium, magnesium, sulfate, iron and manganese, which mayresult from dolomite and gypsum dissolution, and pyriteoxidation. Water chemistries observed in the northern portion ofthe 'study area indicate a shallow component of groundwater flowexists. Water chemistry in the southern portion of the study arearesults from feldspar, dolomite, and calcite dissolution andkaolinite precipitation, and has not been impacted by effluentwater.
Acknowledgments
I would like thank my committee members Dr. David T. Long,Dr. Graham Larson, Dr. Mike Velbel, and Norm Grannemann for theirguidance and support. I would like to thank again Dave for giving mea chance to work with him, and for the use of analytical equipmentand materials in the Geochemistry Laboratory at Michigan StateUniversity. I would like to extend special thanks to Norm, not onlyfor his insights in hydrology and flow modeling, but also for hisgenerosity. I would also like to give my deep thanks to Jerry Sunwhose help in groundwater sampling wafi immeasurable.
I would like to thank the faculty and staff at the DepartmentGeological Sciences at Michigan State University, and co-workers atthe U.S Geological Survey in Lansing. I would finally like toacknowledge the following companies and agencies that donatedequipment and funds to this project: Keck instruments, U.S.Geological Survey, Michigan Basin Society, and Chevron Oil.
u
Table of Contents
pageAbstract iiList of Tables v
• List of Figures vi1. Introduction
A. Purpose and Scope 1B. Previous Investigations 3C. Setting 6
1. Historical Development of Water 6Resources in the Sands Plain Region
2. Topography 73. Geology of the Sands Plain Region 74. Hydrology of the Sands Plain Region . 15
a. Surface Water Hydrology 15b. Groundwater and Hydrogeologic Unite 21
2. Hydrologic Simulation of the Sands Plain RegionA. Introduction 22B. Conceptual Model 23C. Mathematical Basis 25D. Numerical Model Design 26E. Model Calibration 29F. Sensitivity Analysis 32G. Potentiometric Surface 32H. MODPATH Simulations 34
1. Mathematical Basis 342. Flow System Analysis 36
I. Volumetric Water Budgets 41J. Zone Budget 41
ui
Table of Contents (cont'd)
page3. Groundwater and Surface Water Chemistry
A. Methods 491. Groundwater Sampling 492. Sediment Collection 513. Field Analysis 514. Analytical Methods 52
a. Calcium and Magnesium 52b. Sodium and Potassium 53c. Chloride 53d. Sulfate 53e. Silica 53f. Trace metals 54
5. Pre-Treatment for Clay Mineralogy Analysis 546. X-Ray Diffraction Analysis 55
B. Groundwater-Surface Water Chemistry 561. Groundwater Isogram Maps 562. Piper Diagrams 76
C. Minerals and Clay Minerals 81D. Chemical Modeling with WATEQ4F 85E. Groundwater-Streamflow System 95
4. Discussion 1045. Conclusion 1086. References 1107. Appendix
A. XRD Figures 119
I
IV
List of Tables
pageTable 1. Stratigraphic Column 10Table 2. Measured and Simulated Head Values 30Table 3. Measured and Simulated Streamflow Values 31Table 4. Interflow rates between Adjacent Zones 44Table 5. Groundwater Chemistry for Major Species 58Table 6. Groundwater Chemistry for Trace Species 60Table 7. Surface Water Chemistry for Major Species 61Table 8. Surface Water Chemistry for Trace Species 62Table 9. Saturation Indices for Groundwater and Surface 87
Water Samples with Respect to Selected MineralPhases
Table 10. Equilibrium and Enthalpy Values for Selected 93Minerals
List of Figurespage
Figure 1. Site Location Map 2Figure 2. Areal Distribution of Bedrock 9Figure 3. Altitude of Bedrock Top 11Figure 4. Areal Distribution of Glacial Deposits 13Figure 5. Glacial Deposits Thickness 14Figure 6. Drainage Basins 16Figure 7. Flow Duration curve Cherry and Cedar 18Figure 8. Flow Duration Big and Silver 19Figure 9. Flow Duration Goose Lake Outlet 20Figure 10. Grid 27Figure 11. Potentiometric Surface 33Figure 12. Backwards Particle Tracking from 37
Well LocationsFigure 13. Forwards Particle Tracking from 49
Well LocationsFigure 14. Backwards Particle Tracking from 40
Stream Samples SitesFigure 15. Zone budget 42Figure 16. Groundwater and Surface Water 57
Samples Site LocationsFigure 17. Isogram Total Dissolved Solids in mg/L 64Figure 18. Isogram Calcium in mg/L 66Figure 19. Isogram Magnesium in mg/L 67Figure 20. Isogram Sodium + Potassium in mg/L 68Figure 21. Isogram Bicarbonate in mg/L 69Figure 22. Isogram Sulfate in mg/L 70Figure 23. Isogram Chloride in mg/L 72Figure 24. Isogram Barium ug/L 73Figure 25. Isogram Manganese in ug/L 74Figure 26. Isogram Strontium in ug/L ^ 75Figure 27. Groundwater Piper 77
vi
pageFigure 28. Surface Water Piper 79Figure 29. Historical Data Piper 80Figure 30. Cedar Creek headwaters X-ray diffraction 82Figure 31. Silver Creek headwater X-ray diffraction 83Figure 32. Calcium versus Magnesium in mg/L 96Figure 33. Calcium versus Sulfate in mg/L 96Figure 34. Calcium versus Bicarbonate in mg/L 99Figure 35. Calcium versus Total Dissolved Solids in mg/L 99Figure 36. Calcium versus Strontium in mg/L 102Figure 37. Sulfate versus Bicarbonate in mg/L 102Figure 38. Chloride versus Sodium in mg/L 103
Vll
introduction
Purpose and ScopeThe Sands Plain aquifer of Marquette County in the Upper
Peninsula of Michigan represents a freshwater resource thatdisplays a similar sequence of glacial deposits observed in manyparts of the Great Lakes basin. The impact of iron mining on thissmall glaciated basin allows for a unique opportunity to study thegeochemical evolution of a groundwater-streamflow system,because of its well constrained hydrogeologic setting, geochemicalvariability and the existence of a historical data base (see Figure 1)(Wiitala, 1969; Grannemann, 1979, 1984; Doonan and VanAlstine,1982). The first objective of this study is to characterize thehydrogeologic setting of the Sands Plain region, and the distributionand composition of shallow groundwater and stream water within it.The construction of a finite difference flow model for the SandsPlain area proposed by Grannemann (1984) and modified for thisstudy is used to display the potentiometric surface, groundwaterflow path direction and travel time information, as well as aquantitative volumetric water budget data. Water samples fromtwenty two wells and fifteen stream sites were taken over a largeportion of the region to define the chemical composition of thesewaters. Water samples were analyzed for Ca, Mg, Na, K, HCOs, S04,Cl, Si, Al, Ba, Cu, Fe(ll), Fe (III), Mn, Sr, and Zn, and used inconjunction with precipitation chemistry (NADP/NTN, 1992), andLake Superior water chemistry (Water-Resource Data, 1980).
The second objective of this study is to investigate thefactors which control the distribution and composition of thesedilute waters. This is accomplished by relating hydrologic flowmodeling, water composition data, equilibrium modeling with aid ofWATEQ4F (Ball, 1991), and clay mineralogy information by X-raydiffraction.
Lake Erie
Figure 1
3 *Previous investigations
Combining physical flow information, aqueous chemistry, andmineralogic information is essential to the understanding of thechemical evolution of water in the hydrologic cycle. Previousstudies that investigate aqueous chemistry and/or hydrogeologicmodels of small glaciated systems include: Newbury et al. (1971);Cherry (1972); Sklash et al. (1975); Grisak (1976); Desaulniers,Cherry and Fritz (1980); Wallick (1981); Bradbury (1984); Bottomleyand Johnston (1986); Hendry et al. (1986); Anderson (1989); Ophoriand Toth (1989); Kenoyer and Bowser (1992).
Early studies by Wallick (1981) and Cherry (1972) were able tocharacterize the chemical evolution of groundwater in glaciateddrainage basins by observing the influence of water-rockinteractions on major ions in groundwater. Wallick (1981)attributed a Ca-Mg-HCOs type water in the recharge area of aglacial-drift aquifer to the dissolution of calcite and dolomite bycarbonic acid formed by atmospheric COz in the soil zone. He alsoattributed a Ca-Mg-S04 type water to the dissolution andprecipitation of gypsum in the presence of calcite and/or dolomiteunder conditions of partial saturation. In addition, he attributed aNa-HCOs type water from bedrock aquifers to result from theconsumption of H+ by the chemical weathering of feldspars and thedissolution of calcite and siderite minerals.
The degree to which alumino-silicate and clay mineralscontrol groundwater chemistry is not well understood. Nordstrom etal. (1990) observe that mineral groups such as illites, smectites,and micas have never been shown to be the dominant control ofwater composition. This is reflected by a constant ion activityproduct (IAP) for a known silicate mineral composition in an aquiferwhere the water compositions of that aquifer varies (Nordstrom etal.,1990). The alteration of these minerals may still affect thecomposition of natural waters as observed by Kenoyer and Bowser(1992b), Jackson and Patterson (1982), and Garrels and Mackenzie(1969). Kenoyer and Bowser (1992) indicated that changes ingroundwater chemistry along a flow path in a glacial drift aquiferwhich lacked carbonate minerals could be explained through the
4dissolution of feldspars and clay minerals with the aid of thereaction-path model PHREEQE (Parkhurst, Thorstenson and Plummer,1980).
Velocity and direction of groundwater flow are importantfactors in controlling the pattern of chemical evolution of naturalwaters, because they determine the sequence and duration ofspatially distributed chemical and biologic processes (Schwartz andDomenico, 1973). Flow modeling requires determination of hydraulicparameters and boundary conditions, and the use of calibrationtechniques. Anderson (1989) proposed methods to adapt glacial andglacialfluvial sedimentological facies models for conceptualizinglarge scale hydrogeologic trends and delineating hydraulicconductivity for use in numerical models.
Oxidation and reduction reactions may also have an importantrole in dictating water chemistry in glaciated systems. Hendry etal. (1986) indicated that high sulfate concentrations in fracturedweathered till in Alberta was mainly due the oxidation of organicsulfur by bacteria, along with the dissolution of minor amounts ofsulfate-rich bedrock materials, Hendry (1986) was able disprovethe hypothesis of glacial-load squeezing of sulfate-rich brinesproposed by Cherry (1972) for the origin of sulfate-rich watersthrough the use of d^O and 334S data, and through geochemicalmodeling with aid of PHREEQE (Parkhurst, Thorstenson and Plummer,1980). Hendry et al. concluded that concentrations of Ca+2 and Na+were due to cation-exchange and gypsum precipitation, and thatthese processes were in turn governed by S04*2 concentration andcharge balance of the solution.
Stable and radioactive isotopes with major and tracegroundwater species have been used effectively in delineating zonesof recharge and discharge, as well as the effects of mixing betweentwo or more water bodies. The age and origin of groundwater andhydraulic characteristics and porosity of nonfractured clayey till inthe St. Clairs Basin, Ontario was determined by Desaulnier et al.(H)80). The groundwater of this region was characterized by a freshwater source and displayed no evaporation as indicated by themeteoric water line plot of diQ0 versus 32H. Groundwater
5movement and age were derived by 9180 and Ch relationships, andtritium and t4C water content, respectively.
The interaction between groundwater and streamflow inglaciated systems is described by Newbury et al. (1969), Pinder andJones (1969), and Sklash et al. (1975, 1979). The water carried bystreamflow at a'given instant consists of three components: directrunoff, interflow and baseflow. The relative proportions of thesecomponents largely determine the composition of streamflow(Stumm and Morgan, 1981). The concentration of dissolved solids ineach is influenced in turn by the interactions of precipitation withminerals and vegetation, and by evapotranspiration.
The contributions of the groundwater component to the totalstreamflow have been approached by means of graphical separationof hydrographs as seen in Chow (1964). This method of separation isconsidered to be some what arbitrary. Pinder and Jones (1969) andNewbury et al. (1969) used groundwater chemistry to investigate thevariations in composition and to determined the groundwatercomponent of streamflow in small glaciated basins in Nova Scotia,and in Manitoba, Ontario, respectively. Both of these studies found amuch higher contribution to streamflow by groundwater (over 90%)during storm events than previously calculated by stream hydrographseparation. Newbury et al. (1969) indicated that baseflowcontributions to streams could be separated into transient and long-term groundwater types during a storm event.
Sklash (1975, 1979) refined the use of water chemistry indelineating groundwater contributions to streamflow in a glaciatedsystem with the aid of 180 isotopes. The use of these techniquesoffer advantages over those of Pinder and Jones (1969) andNewburry et al. (1969) because the content of 180 isotopes isconsidered to be a conservative property and is not affected bychemical or biological reactions in water. The content of oxygen-18in water can only be altered by the mixing of two or moreisotopically different waters. A disadvantage to this method ofstreamflow separation occurs when the isotopic signature of bothgroundwater and surface water do not differ significantly from oneanother.
6Historical development of water resources in Sands Plain
As the name suggests the Sands Plain is a fairly flat lyingsandy plain, but the area received its name from Jacob Sands whosettled in the Marquette County in 1850fs. The development ofnatural resources in the Marquette County began with the discoveryof iron ore near Lake Teal in 1884 by W.A. Burt while surveying andmapping the region (Wiitala, 1967). By the year 1886 mining forhigh grade ore deposits had started at the Jackson mine nearNegaunee. Mining production surged in this area until the early1930fs, and then declined as high grade ore deposits were exhausted.In the early 1950's mining production resurged owing to thedevelopment of new beneficiation processes to concentrate lowgrade ores deposits, and because of the development of new miningtechniques.
In 1977 iron mining in Marquette County produced almost 20percent of iron ore in the United States (Grannemann, 1979). Twomines are currently (1992) operating. These are the Empire and theTilden mines which lie to the west of the Sands Plain region.
Water is an essential resource to the iron ore industry. In1965 the mining industry consumed over 31.5 million gallons ofwater per day (Wiitala, 1967). The mining of iron ore itself placesonly slight demands upon water resources. Beneficiation andpallatization processes place larger demands on water resources.Local streams and lakes are targeted for these processes because oftheir accessibility and low operational pumping cost. Water is usedprimarily in these processes to aid in the grinding of ore rockmaterial for iron ore concentration, and as a medium fortransporting iron ore throughout plant facilities (Wiitala, 1967).Waste rock from concentration processes is made into a slurry andtransported to large settling basins. These large basins arecomposed of a series of smaller basins in which the effluent iscontinually ponded. Solids contained in the effluent are allowed tosettle out of suspension over time. Clay materials are flocculatedout of solution by the addition of Alum, an aluminium potassiumsulfate clarifier. Excess water used in the clarifying process is
7finally discharged into streams. Wiitala (1967) indicates that theoverall usage of water in the clarifier process is high, but the actualconsumption of water is low. Effluent is presently being dischargedby the Empire mine to Warner and Schewietezer Creeks and by theTilden mine to Goose Lake and Goose Lake Outlet.
Other natural resources developed in Marquette County includesand, gravel and dolomite, along with the production of lumber forpulpwood. Water resources are utilized by these industries, as wellas for the generation of hydroelectric power and reservoir storage.
TopographyThe Sands Plain region of lower Marquette county has an area
of approximately 40 square miles. The western section of the studyarea is composed of hills and ridges of the Marquette Iron Rangewith an average elevation of 1,500 feet above sea level. TheMarquette Iron Range (Wewe hills) lying in the western portion ofthe study area forms a distinct topographic high. The highest pointin Marquette county is Summit Mountain 3 miles south of Negaunee.The west central section is relatively flat-lying with a meanelevation of 1,220 feet The eastern section of the study area iscomposed of northwest trending moraines that display a hummockytopography. The elevation of this area decreases steadily to theeast from 1200 feet to less than 700 feet
Geology of Sands Plain areaThe Marquette Iron range is underlain by Precambrian gneiss
and middle Precambrian metasedimentary and metaintrusive rockswhich are located in the western portion of the study area (seeFigure 2). A summary of the lithologic units in the study arepresented in Table 1. The major structural/stratigraphic featuresof the area are: 1) The Marquette synclinorium, which trends andplunges to the west and consists of the Marquette Supergroup(middle Precambrian) (Gair, 1975); 2) The Palmer basin whichborders the synclinorium to the south (Gair, 1975); 3) The Palmerfault which is associated with the Marquette synclinorium, and
8bounds the Palmer basin to the north (Gair, 1975); and 4) Two majorand several minor buried valleys in the Sands Plain area which werecarved by streams and glacial ice activity (Huges, 1978) (see Figure3). One major buried valley trends to the northeast towards LakeSuperior paralleling the Palmer fault The second buried valleytrends to northeast following Big Creek (Grannemann, 1984).
Lower Precambrian rocks of this area are comprised of theMona Schist and the Compeau Creek Gneiss. The Mona Schist iscomposed of massive mafic metavolcanics and greenstones and isintruded by the Compeau Creek Gneiss (Gair and Thaden, 1968).
Unconformably overlying the Compeau Creek Gneiss is theMarquette Supergroup which is composed of the Chocolay group,Menominee group, and Baraga group, respectively. Formations of theChocolay group consist of the Enchantment Lake Formation, theMesnard Quartzite, the Kona Dolomite, and the Wewe Slate. TheEnchantment Lake Formation is composed of a basal conglomerateand fine grained graywacke. The Kona dolomite is composed of athinly bedded dolomite and quartzose dolomite, and containingcopper bearing minerals (Gair, 1968); (Grannemann, 1984).
The Menominee group is separated from the Chocolay group byan unconformity, and consists of the Goodrich Quartzite, AjibikQuartzite, Siamo Slate, and the Negaunee Iron Formation. NegauneeIron Formation is composed of siderite, hematite, magnetite, pyrite,chert, K-feldspar, quartz and chlorite, and varies in mineralogy andmineral composition (Gair, 1975).
To the east of the Marquette Supergroup is the JacobsvilleSandstone which is late Precambrian. The Jacobsville is a reddishto light gray lenticular sandstone composed of'quartz and variableamounts of feldspar, and is intercalated with gray conglomerate andreddish shale (Gair and Thaden, 1968).
Proterozoic deposits are in turn overlain by Pleistoceneglacial deposits which cover most of the Sands Plain area (sjeeFigure 4). The glacial history of the Sands Plain area has beendescribed by Hughes (1978), and is briefly summarized: During the
RJ26W. R24W.
T.4TR
Figure 2 Areal Distribution of Bedrock
IL25W R24W. lt23W
T.45R
It23W.
Figure 2 Areal Distribution of Bedrock
10
Table 1 Strati graphic Column
Age
Phen
eroz
oicPr
oter
ozoi
c
Ceno
zoic
Que
tern
ery
Peleo
zoic
Cem
brie
n
Pleis
toce
ne
i
I
i
Arch
een
Geologic Unit
Glacial deposits; tilloutvashanddreinagedeposits, end lacustrinedeposits
TrempaaleeuFormation
MunisingSandstone
JacobsvilleSandstone
Intrusiverock
Metamorphsooddikes and sills
Negeuneeiron Formetion
Metesedi mente r yrocks,u Differentiated
Compeeu CreekGneiss
Older Archeanrocks
lithology
Till is a poorly sorted, nonst ratified mixture of send,silt, cley, gravel, end boulders. Outvesh and drainagechannel deposits ere composed of veil-sorted, stratifiedsand and gravel vith some silt. Lacustrine deposits arestratified mixtures of send, silt, and clay vith somegravel. Maximum knovn thickness of glacial deposits inthe study erea is 459 fleet.
Dolomitic limestone. Meximum thickness is about 300 fleet.
Unit consist of a basal conglomerate overlain by a veil-sorted, medium-grained competent sandstone end en upperpoorly sorted, friable sandstone. Meximum thickness isabout 100 fleet.
A mottled red or reddish- brovn feldspatic sandstonecontaining lenses of red or gray conglomerete end some redshale. Four lithelogic units are recognized: a basalconglomerate, a lenticular sandstone, a massive sandstone,end en upper red siltstone. Thickness varies from 1 to 1 00feet.
Mostly diabase dikes. Massive, derk gray, medium to finegrained; in pieces extensively argillizad.
Mostly metadiabese dikes. Ranges from thin, fine-grained .intrusions that may be greatly altered to thick, coarsegrained intrusions that ere reletivelitresistent toalteration; variable in appaa ranee and composition.
Iron rich metesedimentery rock in places extensivelyoxidized. Near Palmer, thickness ranges from 450 to1,300 feet.
Includes Goodrich Quertzite, Siemo Slate, AjiMk Quertzite,Wove Slate, Kona Dolomite, Mesnerd Quertzite, endenchantment Lake Formation.
Mostly gneiss composed of lightly colored, quartzite feldsparvith some pegmetite, end locally, some layered rock.
Mostly medium grai ned chlorite -quartz- muscovite schist,q ua rtz - pi eglocl we -chlorite schist, and actinolite schistcut by thin dikes.
11
t>
-1000- BEDROCK CX)NTOUK —Shows mtotud*of bedtx>ck »orf«ce. Contour inurvml 100 feetNGVDof 1929
BUKIED VALLEY — Aitm potato downj rmdient
Figure 3 Altitude of Bedrock Surface
11
,ita&.w
Figure 3 Altitude of Bedrock Surface
12last Wisconsinan stage, ice advanced from the northeast alongvalleys covering a large portion of the Sands Plain area with ice. Iceadvancement was hindered by elevation and resistant rock allowingfor the formation of a till ridge known as the Outer MarquetteMoraine. The Outer Marquette Moraine is composed of basal tilldeposits containing a mixed assortment of boulders, gravel, sand,silt, and clay. Deposits both underlying and to the rear of the OuterMarquette Moraine are composed of ablation till. Abalation till iscomposed of coarse grained to fine grained sand with lesser amountsof clay (approximately 11%) and is roughly 175 feet thick(Grannemann, 1984). Outwash is deposited in front of this morainallobe to the south and to the west by stream systems associated withmelt water. These outwash deposits are composed mainly of coarseto medium sand and gravel and range from 0 to 150 feet in thickness.Transitional till deposits also occur between the Outer MarquetteMoraine and outwash deposits. These deposits represent thetransition between unstratified ablation till and stratified outwashdeposits (Grannemann, 1984).
Glacial ice readvancement from the northeast generated asecond ridge of till, the Inner Marquette Moraine. Outwash was againdeposited in a narrow plain between the inner and outer moraines bymelt water. Ancestral lakes produced during the final glacial icerecession deposited lacustrine sediments in the northeasternsection of the study area. Lacustrine deposits consist of silt, finesand and clay (up to 17%) (Grannemann, 1984). In total, glacialdeposits range roughly from 0 to 450 ft, and are thickest alongburied valleys as observed in Figure 5.
13
&26W. R2SW* 8J4W.
T.4TN.
T.46N.I
0 1
Figure 4 Areal Distribution of Glacial Deposits
R25W. &24W. &23W.
T.47R
T.4SR
R23W.
figure 4 Areal Distribution of Glacial Deposits
14
HMW.
figure 5 Glacial Deposits Thickness
14
R»W. IL14W.
T.4SR
IU4W. R23W.
, nOMETEBS
Figure 5 Glacial Deposits Thickness
15
Hydrology of the Sands Plain areaWater in the Sands Plain area is derived almost exclusively by
precipitation, which is stored temporarily in streams, lakes andgroundwater (Grannemann, 1984). The Sands Plain area is drainedprimarily by the Chocolay River basin. This basin is surrounded tothe north by the Carp River basin, and to the south and west by theEscanaba River basin (see Figure 6).
Surface WaterThe Sands Plain area contains three major rivers, the
Chocolay, the Carp, and the East Branch Escanaba. The Carp and theChocolay Rivers flow eastward and northeastward, respectively, anddischarge to Lake Superior. The East Branch Escanaba River joinswith the Middle Branch Escanaba River at the town of Gwin, tobecome the Escanaba River which flows south into Lake Michigan.
Tributaries of the Chocolay River are Big Creek, Silver Creek,Cherry Creek, and Cedar Creek. Headwaters of Big Creek originate inthe Outer Marquette Moraine. Headwaters of Silver, Cherry and CedarCreeks originate from springs in the outwash plain associated withthe Inner Marquette Moraine (Grannemann, 1984). The Carp Riverdirectly feeds Deer Lake Reservoir and is utilized for hydroelectricpower near Harvey. Tributaries of the East Branch Escanaba Riverare Warner Creek, Goose Lake Outlet, Powell Lake Outlet, andSchweitzer Creek. Goose Lake Outlet and its tributaries drain thearea surrounding Gribben tailings basin.
Since precipitation is the main source of recharge for theSands Plain area, seasonal variations in rate and volume ofprecipitation due to climatic changes have marked effects on streamdischarge. Grannemann (1984) indicates that the Chocolay River aswell as Cherry and Cedar Creek have only slight stage fluctuationsover one seasonal cycle. Small stage fluctuations in these streamsare attributed to highly permeable stream beds, and a relatively
16
n»w. H24W,
T.46N.
1 2 3 4 5 6 JOLOMETEBS
Figure 6 Drainage Divides
16
&26W. IL24W. R.23W.
T.4TH
T.46R
T.4SR
R23W.
KILOMETERS
Figure 6 Drainage Divides
17constant baseflow influx. This can be observed in the flow durationcurve Figure 7, where the recurrence interval of equal streamdischarge in one annual cycle is plotted versus discharge in cubicfeet for Cherry Creek. The relatively flat slope of this curveindicates a steady baseflow component for Cherry Creek. This figurecan also be used as an indicator of Cherry Creek's low-flow capacityat approximately the 90 percentile level. High flow capacityvolumes indicated at the 10 and 25 percentiles for the Cherry Creekduration curve are only slightly greater than low flow capacityvolumes at the 90 percentile, suggesting that runoff during floodintervals is rapidly dissipated.
Big Creek and Silver Creek have greater seasonal variability inflow than Cedar and Cherry Creeks. Variability in flow for Big Creekand Silver Creek is due to higher amounts of silt and clay in outwashor till material making up the streambed which reduce the rate ofinfiltration by precipitation and promotes greater stream dischargeduring storm events. This can be observed in the flow duration curvefor Big Creek (Figure 8). The relatively steep slope of this curveindicates the creek has a higher degree of variability due to thelarger component of runoff. High flow capacity volumes indicated atthe 10 and 25 percentiles for the Big Creek duration curve are onlyslightly greater then low flow capacity volumes at the 90percentile, suggesting once again that runoff during flood intervalsis rapidly dissipated. Goose Lake Outlet has a similar seasonalvariability as observed for Big and Silver Creek, but has a muchhigher rate of discharge (see Figure 9).
As a result of the nearly constant baseflow influx into BigCreek, Cherry Creek and Cedar Creek, discharge rates increase downstream (Grannemann, 1984). Not all streams in the Sands Plainregion are effluent Grannemann (1984) indicates that Silver Creekloses water to groundwater from gaging stations 18 to 23downstream at rates that ranging from 0.1 to 1.2 cubic feet/second.Goose Lake Outlet loses approximately 10.3 cubic feet/second in athree-mile reach from Goose Lake to gaging site 33 (see Figure 6)(Wiitala, 1967); (Grannemann, 1984).
OUchAff • In CFS 30ftOf&ftlf)
IW
90
80
70
60
50
40
30
20
10
A
(
\
«dtr
Chtrry
•
A
t
tattMD)(yean
JrttkCi
(9
A
ftJnaffI98<M
teftiaagtiMI* I960
trt«9.04981)
rta4<1981
iq.ml
13 »q.
«)
aife
A
>
•
r
A
•
. J
00
0.13 0.5 t 2 tO 20 30 40 50 60 70 80 90 95 98 99 99,5 99.87Percent or Time Indicated Vkli* wu Equakd or Ewteded
Figure 7 Stream Flow Duration Curves for Cherry and Cedar Creeks
in CFS
IUV
90
80
70
60
40
20
in
•
A
*
A
A
*
*
,~— «-4S
AA
*•
*
Crttk
AAA
litvtrC(Years
*
Drainageet» iftfiii
AA
reek (Oral1980-19
Arta
>age>It)
17.00
rea8
Klft
$89
U
I Ml)
*
A
0.13 0.5 I 2 10 20 30 40 50 60 70 80 90Percent or Time Indicated WIuc WAS Equated or Eiceeded
98 99 99.5 99.87
Figure 8 Stream Flow Duration Curves for Big and Silver Creeks
IWW
100
10Dfecltftfft in CPS
ILofScmk)
1.0
0,1
0.01
1a a
oo o
DOa
<
OD
OOMUfc<nfartlS
oo o
OutU56-1
i o
tlDrm)82J
CJ 1
Dtf<
(
Ar«
a
(37.$
o
sai
o
ID
O 1oo
c '
0.13 04 1 2 5 10 20 30 40 50 60 70 80 90 9$ 98 99 90,5 99,87Percent or Tiro* Indicated Wlue WM Equated or Exeetdtd
Figure 9 Stream Flow Duration Curves for Goose Lake Outlet
too
21Groundwater
The movement and storage of groundwater in the lowerPrecambrian igneous and metasedimentary units are considered to beminimal compared to those of Pleistocene glacial deposits in theSands Plain area, Hydroiogic characteristics of igneous andmetasedimentary units depend almost entirely on the fractures andjoints in the rock (Grannemann, 1979). Upper PrecambrianJacobsville Sandstone, which covers the eastern portion of SandsPlain area, is also considered to be a poor aquifer; hydraulicconductivity values are on the order of 1 foot/day (Doonan and VanAlstine, 1982). The Munising Sandstone and the TremaeleauFormation have higher hydraulic conductivities (on the order of 8 to10 feet/day), but are not areally extensive (Grannemann, 1979).Bedrock units composed of low permeability materials bound theSands Plain area to the north and west creating the basin-likestructure which has an outlet towards Lake Superior.
Glacial deposits lying in this basin-like structure make up theprimary aquifer material These deposits have hydraulicconductivity (K) values ranging from 30 to 180 feet/day(Grannemann, 1984). Hydraulic conductivity values were determinedby various pumping tests throughout the Sands Plain area by Wiitala(1967). Storage coefficients (S) determined by pumping testsranged from 0.01 to 0.0001. This range in storage coefficientsindicates that the Sands Plain aquifer is mainly unconfined, but maybe locally confined or semiconfined in morainal areas with abundantamounts of clay and/or silt
Groundwater contained in glacial deposits receives minimalinterbasin flow from the Carp River Basin and is bounded to thesouth by the Silver Lead Creek and Chocolay River drainage divide.Groundwater is recharged directly by precipitation, and by a threemile stretch along Goose Lake Outlet. Groundwater is discharged tocreeks and lakes contained in the Sands Plain area and to LakeSuperior. Thus the hydrologic conditions of the Sands Plain areaindicates a fairly isolated system which is well constrained, andthus ideal for this study.
Hydrologic Simulation of the Sands Plain Region%
Sands Plain ModelA numerical model simulating steady-state hydrologic
conditions in the Sands Plain area was developed to quantify waterinflows and outflows, to test hypotheses involving stream andaquifer interactions, and to generate flow path direction and traveltime information to help interpret concentration distribution fordissolved constituents. The Sands Plain numerical modelsintegrates broad conceptual ideas about the aquifer's hydraulicswith geologic and hydrologic data established by field measurementand observations. Aquifer characteristics and concepts are thendiscretized in time and space and entered into governing flowequations to which computer solution techniques are applied.
The numerical model used for the Sands Plain area is based ona previous model developed and calibrated by Grannemann (1984).The Sands Plain model presented by Grannemann (1984) was used toinvestigate hydrologic and geochemical effects of the GribbenTailings Basin and hypothetical tailings basins to the Sands Plainaquifer. Modifications made to this model were primarily inupdating its capability to graphically display input parameters, thepotentiometric surface, and cell-by-cell flow data generated by themodel. Only minor changes were made to initial input data. Post-processing techniques such as particle tracking and zone budgetanalysis were also linked to the Sands Plain model.
22
23Conceptual Model
A conceptual model of the Sands Plain aquifer system is basedon the geologic setting, and hydrologic parameters discussedpreviously* Basic assumptions of the conceptual flow model includeboundary conditions, hydrologic properties, and stresses to thesystem which are summarized below:1. Precambrian deposits composed of igneous and metamorphic rocktypes are considered to have poor conductivities for flow and littleto no storage capacity, although various degrees of fracturing mayoccur in them. Precambrian and Paleozoic sandstones are consideredto have low conductivities and/or are areally insignificant to themodeled study area.2. Rock types ascribing the hydrologic characteristics stated aboveare assigned a Type 2 impermeable no-flow boundary condition. AType 2 boundary (Neumann type) assumes that flux across theimpermeable boundary is negligible (McDonald and Harbaugh, 1988;Kinzelbach, 1986). This assumption is justified by hydraulicconductivities values that are two orders of magnitude lower forthese rock units than for that of the modeled system (Anderson andWoessner, 1991). Type 2 boundaries are assigned to bedrock unitsunderlying the Sands Plain glacial package and to the Marquette IronRange lying to the northwest.3. A Type 1 prescribed head boundary is assigned to Lake Superior inthe northeast section of the study area, to the West Branch of theChocolay River, and to the Chocolay River in the south, south easternsection of the modeled area. A Type 1 boundary (Dirichlet type)specifies a constant head (equipotential line) and is necessary toguarantee the uniqueness of the solution (McDonald and Harbaugh,1988; Kinzelbach, 1986; Schwartz and Domenico, 1990).4. Interbasin flow from areas outside of the defined boundaries isnegligible.5. Proterozoic glacial deposits make up the principalhydrostratigraphic unit for the Sands Plain area.6. Hydraulic conductivity values assigned to glacial deposits areisotropic, vertically averaged over depth, and are based on thematerial type and its thickness. Hydraulic conductivity values are
24distributed spatially in Sands Plain area according to the glacialdepositional environments described by Hughes (1978).7. The Sands Plain aquifer is unconfined. Although this assumptionseems valid, perched water tables and leaky confining layers may bepresent in some localities.8. The head values can be calculated from Dupuit's flowassumptions, which states that flow lines are horizontal andequipotential lines are vertical, and that the hydraulic gradient isequal to the slope of the free surface and is invariant with depth(Freeze and Cherry, 1979; Kinzelbach, 1986; Anderson and Woessner,1991).9. Water levels in all streams are considered to be constant. Thisassumption is considered valid because the greatest measureddifference in gage height over a year period at designated gagingstations on tributaries of the Chocolay River and for Goose LakeOutlet are approximately 2 feet and 4.5 feet, respectively,(Grannemann, 1984). These differences are insignificant comparedto a total head difference of about 600 feet between Goose LakeOutlet and Lake Superior (Grannemann, 1984).10. The Sands Plain area is recharged directly by precipitation,runoff from Marquette Iron Range, .and by losing reaches of GooseLake Outlet.11. Groundwater is discharged from the Sands Plain area by leakageto Lake Superior and to tributary streams of the Chocolay River.12. The water consumption in the Sands Plain area by domestic andindustrial use is minimal.
25Mathematical Basis
A two-dimensional groundwater flow equation describing the1"conservation of mass is given by,
Pvx + pvy = pvx - a/ax (PVx) + pvy- a/ay(pvy) (i)
were Vx and Vy are the specific discharge in the x and y direction,respectively, and p is the density (Kinzelbach, 1986; Freeze andCherry, 1979). The conservation of mass requires that the rate offluid mass flow into an elemental control area is equal to the rateof the fluid mass flow out. For steady state flow with a constantfluid density the equation for flow translates into,
a(Vx)/ax + a(vy)/ay « o (2)
Substitution of Darcy's Law (V=K3h/3x) for Vx and Vy into equation(2) gives
a/ax(Kx an/ax) + a/ax(Ky ah/ay)=o (3)
for steady state flow through an anisotropic saturated porous media.The flow equation for an isotropic steady state unconfined aquifercan be described by
(4).
where b is the saturated thickness of the aquifer and Kx = Ky = Cwhere is a C=constant. The equation for a isotropic steady stateunconfined aquifer is used for the Sands Plain modeled area.
26Numerical Model Design
The Sands Plain model is defined by a one layer grid havingdimensions of 74 columns by 64 rows (see Figure Grid). Each grid-cell face has a length of 1000 feet, with the total area of the modelto covering approximately 160 sq. miles. The grid was oriented 40°counter clockwise from the lower left hand origin and roughly alignsthe principal directions of the hydraulic conductivity with the x andy coordinate axes. Prescribed head boundary cells were positionedsuch that their nodal centers lay along the boundaries, andimpermeable no-flow boundaries cells were positioned such thattheir cell face(s) lay along the boundary (see Figure 10). Thedistribution of hydraulic conductivity values for grid-cells range invalues from 0 to 120 feet/day.
The position of the top and the bottom of the glacialhydrostratigraphic unit is referenced to a sea level datum andentered into model grid-cell locations by use of computer programGRID.F77 (Swain, 1974). This program utilizes the spline fit methodof data point interpolation to discretize topographic and bedrock topsurfaces into 74x64 gridded area. Thickness data of glacial depositsfor each grid-cell could then be derived by subtracting griddedglacial top data from gridded glacial bottom data.
Groundwater flow between the Sands Plain aquifer andoverlying streams and lakes was simulated by the use of the River(RIV) package (McDonald and Harbaugh, 1988) (see leaky layers inFigure Grid). Streambeds and lake bottoms are assigned conductancevalues (CRIV). Conductance values are calculated by multiplying thevertical conductivity values (Kr) for stream sediments by the length(L) and width (W) of the stream channel and dividing by the thickness(M) of the streambed sediments, where
CRIV=KrLW/M (5)
These values generally range from 0.36 to 1.68 square feet/day inthe Sands Plain area. The rate of water leakage (QRIV) is calculatedby multiplying the streambed conductance by the difference betweenthe stream head (HRIV) and the aquifer head (h), where
01
Ktt 0
imo.
01
waia
28
QRIV=CRIV(HRlV-h) (6)
Losing reaches of streams are simulated when QRIV is a positivenumber and gaining reaches are simulated when QRIV is negativenumber.
The recharge (RCH) package is used to simulate precipitationpercolating into the groundwater system (McDonald and Harbaugh,1988). The recharge rate (QRij) is specified as the rate of flux (ly)applied to the top cell face DELRj- DELCi, where
QRy - ly -DELRj-DELCj (7)
Recharge is distributed areally over the whole grid and isapproximately 15.0 inches/year (Grannemann, 1984).
The well package is used to simulate pumpage from threewells having withdraw rates of approximately 1 cubic foot/sec fromthe southwestern region of the study area. The pumping or injectionrates (Qjj.k) for each layer can be approximately calculated by
Qy,k * Ty,k(Qwr/S Ty.k) (8)
where Qwr is the total pumping or injection rates for the well, Ty^is the transmissivity of a layer and S Tytk is the sum of thetransmissivities of all layers penetrated by the well (McDonald andHarbaugh, 1988).
The evapotranspiration package was not included in the flowmodel because no confirmed estimates for the rates ofevapotranspiration could be obtained.
29Model Calibration
The ability of the Sands Plain model to simulate fieldmeasured head and flow values is demonstrated through thecalibration process. The calibration process is accomplishedindirectly by a series of trial and error adjustments of the modelparameters* In each simulation, modeled head and flow values werecompared to field measurements. In areas where poor matchesbetween modeled and measured head existed, hydraulic conductivityvalues, river conductance and/or recharge values where changeduntil the difference between modeled and measured head valuesaveraged no more that 1.4 feet
The model shows no apparent bias for head values (Holtschlag,oral communication). Table 2 indicates the differences in modeledand measured head values for each well location. Differences inhead values ranged from 6.0 feet in Well 21 to -5.3 feet in Well 5.The standard deviation between measured and modeled head valuesis 3.2 feet; the mean difference between measured and modeled headvalues is 1.35 feet
The model shows no apparent bias for fl<5& values (Holtschlag,oral communication). Measured baseflow values from gagingstations were compared to modeled stream flow values (see Table3). Simulated flow velocities (leakage) were calculated by summinggrid-cell flow values along stream reaches. Negative leakage valuesindicate that the stream is effluent (gaining water from baseflow),and positive leakage values indicate the stream is influent (losingwater to the groundwater system). In areas where poor matchesbetween observed and measured stream flow existed, conductance inthe River Package was changed until the average difference was *1.05 cubic feet/second. The standard deviation between measuredand modeled flow was 1.91 cubic feet/second; the mean differencebetween measured and modeled flow was 1.05 cubic feet/second.
30
Table 2
Measured and Modeled Head Values
Head in feetWell #34333029262321191514111087651
P4P1
I43565242524452605554252936373439424851
J32231617212230263642212424343937563537
Modeled1024.41176.91194.01198.01182.51141.51101.41185.31058.4998.9
1148.31130.01125.11022.8939.5963.1634.8
1047.51035.8
Measured1045.61179.61198.41199.01182.71144.11107.41188.91061.6994.2
1147.71132.91127.01025.2953.3957.8638.2
1049.21038.0
Difference3.20.34.41.00.22.66.03.63.2
-4.7-0.62.91.92.4
-2.3-5.3 '3.41.72.2
* I and J values define grid-cell locations with respect to thelower left comer of the model grid
Table 3
Measured and Simulated Stream Flow Rates
Leakage in cubic feet per secondLocation Gaging Station Simulated Observed Difference
Silver Creek 2£ -6.851 -9.2 -2.3Cherry Creek 16 -18.481 -19.2 -0.7Cedar Creek 11 -10.742 -12.8 -2.1Upper Reach of
Goose Lake Outlet 10.841 i
Lower Reach of £2Goose Lake Outlet 0.040
Strawberry Lake -9.708Big Creek 7 -30.285 -29.3 0.9
32Sensitivity Analysis
Sensitivity analysis is performed to understand how changesin parameter values effects the model's solution. Grannemann(1984) analyzed the sensitivity of the Sands Plain model to changesin three sets of interrelated parameters: hydraulic conductivity,recharge, and stream conductance. Grannemann (1984) indicatesthat estimates to recharge are consistent with baseflowmeasurements Given this recharge information, variation inhydraulic conductivity caused greater variation in head and flowvalues than similar variations in stream conductance.
Potentiometric Surface
Simulated potentiometric surface for the Sands Plain aquiferis displayed graphically in Figure 11. The potentiometric surfacewas contour at 20 feet interval. Hydraulic-heads range from 1200feet in the west portion of the modeled area to 610 feet near LakeSuperior. The potentiometric surface in Figure 11 indicates thatflow is towards Lake Superior (northeast), and that hydraulicgradients are generally highest in the central portion of the modeledarea. Figure 11 also indicates discharge to groundwater from GooseLake Outlet near Goose Lake, and that tributaries of the ChocolayRiver receive recharge from groundwater.
ou;amopua^od n
€8
33
R24W.
JUJW
RJ3W.
Figure 11 Potentiometric Surface
34MODPATH Simulation
Mathematical BasisHead output from the calibrated Sands Plain model are linked
to the post-processing particle tracking package MODPATH andMODPATH-PLOT (Pollock, 1 989) to determine path directions andresidence times for solutes in the system. MODPATH utilizesadvective flow transport to simulate the movement of dissolvedconstituents in groundwater.
Although modeling solely with advective transport does notaccount for effects of dispersion and chemical reactions along flowpath, it provides a good first order approximation because of theuncertainties in assigning parameters for these variables.Estimates of particle-travel time in the Sands Plain region thusrepresent the fastest rate of particle movement along a flow line.These estimates assume that groundwater volumetric flow rates donot fluctuate rapidly over time.
MODPATH solves the advective component of transportequation by calculating average linear velocity. The average linearvelocity across a face in a grid-cell is obtained by dividing thevolumetric flow rate (Qx) across the face by the cross sectional areaof the face (Dx Dy) and the effective porosity (n) of the porous mediaas given by,
Qxi/(n.Dx Dy) (9)
in the x-direction. Effective porosity was estimated from tables inFreeze and Cherry (1979) and assigned a value of 0.20 correspondingto medium size sands.
MODPATH calculates groundwater velocity vectors by simplelinear interpolation. Simple linear interpolation assumes that acomponent of velocity varies linearly along a principle axisdirection, and that these variations are independent of the othervelocity components which vary linearly along their associatedprinciple axis. Linear interpolation for the x-direction is describedby
(10)
35where Ax is a constant that corresponds to the component of thevelocity gradient inside the grid-cell (Pollock, 1988). Equation (10)can be rearranged into
Ax=(VX2-Vxi)/Ax or Ax=dVx/dx (11)
The rate of change in the particles velocity through the grid-cell isgiven by,
(dVx/dt)p = (dVx/dx)(dx/dt)p (12)
where the term (dx/dt)p»Vxp is the change in the particle's x-location with respect to time. Substituting into equation (12) gives,
«
(dVx/dt)p=AxVxp or (1/Vxp)dVxp=Axdt (13)
MODPATH uses a semianalytical method to solve the integration ofequation (13) directly to give,
ln[Vxp(t2)/Vxp(ti )]=AxAt where At=tz-tt (14)
Equation (14) finally rearranges into,
Xp(t2)=xi + (1 /Ax)[Vxp(t1 )exp(AxAt)-Vxi ] (15)
which determines the particle's new coordinate location in the x-direction at a specified time interval.
MODPATH has the ability to track particles backward in timeby multiplying the velocity component by -1. This feature is usefulin delineating recharge sources in the Sands Plain area by trackingparticles backward from discharge sinks.
The initial starting location for particles in the grid-cell att|-0 can also be varied by MODPATH. Starting locations of particlescan be generated inside a grid-cell (internally), or can be generatedon a grid-cell face(s). The Sands Plain model described here utilizesthis feature to reflect differences between particles whichoriginated at streams sites and are then tracked backwards to a
36groundwater source, and particles which originated at a welllocation and are tracked forwards to a stream sink. Particlesgenerated at a stream or surface water site are initiated on the topface of the grid cell, while particles generated at well locations areinitiated internally.
Flow System AnalysisParticle tracking techniques were first performed from select
well locations to delineate their sources of recharge. Figure 12 wasconstructed by the backward^ tracking of four particles at each welllocation over an 80 year period. Four particles were used in trackingsimulations because this number of particles could accuratelydepict individual flow paths while retaining the clarity of thesimulation. Flow paths in Figure 12 are represented by dashes amican be traced backwards towards their recharge source. Time seriesdata is represented as colored circles lying along a flow path, andhave a constant time step of 10 years. Figure 12 indicates that flowpaths traced from wells 21, 19, 15, B. McDonald, P4, and PI in thenorthern region of the modeled area derive a proportion of theirrecharge directly from the upper reaches of Goose Lake Outlet. Timeseries data show that it takes a period of 10 to 30 years for aparticle of water to reach these wells from their respectiverecharge sources. Flow paths traced from wells 34, 30, 29, 26, and23 located in the west central part of the modeled area derive aproportion of their recharge indirectly from Goose Lake Outlet Timeseries data indicates that it takes a period of 10 to 30 years for aparticle of water to reach these wells from their respectiverecharge sources. Flow paths for wells 7, 6, and 5 located in thecentral region of the modeled area derive their recharge primarilyfrom precipitation, and require a time period of 40 to 50 years.Flow paths traced from wells 8, 10, D. McDonald and 11 derive theirrecharge primarily from precipitation, but only in a period of 10years or less. Well 1 located near Lake Superior derives a proportionof recharge from both the upper reaches of Goose Lake Outlet in thenorth and from a groundwater source in the western central region
37
Directions and imttt of fnoadtat*
t«t*ntcd b? tbe taekwtb tnctia<
GrxHwB 4oMMiM* tat
0 t
Figure 12
Wl 0001*0001 «»
38of the modeled area. Time series data indicate that it takes a periodof 50 and 80 years for a particle of water to reach these wells fromthese recharge sources, respectively.
Particle tracking techniques were then performed from welllocations to delineate their points of discharge in the modeled area.Figure 13 was constructed by the forward tracking of four particlesat each well location over a 80-year period. Figure 13 indicatesthat flow paths traced from wells 33, 19, 15, 14, P4, PI andHeidtman discharge into the upper and tower reaches of Silver Creekand Cherry Creek. Time series data for these wells indicate thatwater is discharged to Silver Creek and Cherry Creek in a period of20 to 50 years. Flow paths can be traced from wells 34, 30, 29, 26,and 23 in Figure 13 to the headwaters of Cedar Creek in a period of10 to 30 years. Flow paths observed for wells 8, 7, 6, and 5 can betraced into the lower reaches of Cherry Creek and Cedar Creek in aperiod of 30 to 80 years. Flow paths traced from wells 11, 10, 6,and D. McDonald discharge into Big Creek in a period of 30 to 70years.
Particle tracking techniques were performed from streamsample site locations to delineate their sources of recharge in themodeled area. Figure 14 was constructed by the backward trackingof four particles at each well location over a 40-year period. Figure14 indicates that flow path traced from Silver Creek headwatersderive a proportion of their recharge indirectly from Goose LakeOutlet in a period of 20 years. Flow path for Cedar Creek trace itssource of recharge to a groundwater source in the western centralregion of the modeled area. Time series data indicate that it takes a40-year period for a particle of water to recharge Cedar Creek fromits source area. Flow path traced backwards from Big Creek indicatea southern groundwater source which recharge Big Creek over a 40-year period. Other stream site locations in Figure 14 are strongsinks and have no single source area in which they are recharged.
39
Dim!*** aad mt*» of ftwH
fftttrtlad by tht fotwfe tn
MILES
KKHCETERS
Figure 13
39
•caaitta* by tbt fomnb tmfcfef of partkaw ftoai vd
Grid-«tO<)taMosioM at* 1000*1000 ft*t
5 6
Figure 13
40
D Rtoraod*
O Stmrtinf be*tkmfor«iaroktioo
Figure14
H
*»J 0001*0001
««1 WRUWI jo i«K3tn tp
41Volumetric Water Budget
The volumetric groundwater budget for the Sands Plain area isthe summary of all inflows and outflows of water to the modeledarea contributed by each flow component package. The differencebetween these inflows and outflows should equal zero for a balancedvolumetric water budget. The volumetric budget for the Sands Plainarea indicates that the total inflow rate of water to the modeledarea is 171.77 cubic feet/second, and the total outflow rate ofwater from the modeled area is 171.78 cubic feet/second, giving adifference of approximately -0.007 cubic feet/second.
Inflows of water to the Sands Plain araa are composed ofrecharge from precipitation, specified head boundaries, and leakagefrom streams. The volumetric inflow rates are 10.2 cubicfeet/second from specified head boundaries, 142.31 cubicfeet/second from precipitation, and 19.3 cubic feet/second fromriver leakage. Although their is some amount inflow into specifiedhead, it occurs primarily near the southern boundary of the modeledarea and does not effect simulated head results. Recharge is fairlyuniformly distributed throughout the modeled area and is thegreatest contributor of water to the Sands Plain area. Outflows ofwater from the Sands Plain area are composed of pumping wells,specified head boundaries, and leakage to streams. The volumetricoutflow rates are 48.9 cubic feet/second to constant head, 3.3 cubicfeet/second to pumping wells, and 119.5 cubic feet/second to riverleakage. Leakage of water to streams are the greatest means ofwater discharge from the modeled area.
Zone BudgetFigure 15 displays a series of ten zones which divide the Sands
Plain area into distinct hydrologic areas. These zones are used toisolate individual groundwater-streamflow systems and to evaluatethe interflow rates of groundwater from one zone of the modeledarea to another. The boundaries of these zones are constructed alonggroundwater drainage divides by a series of trial and erroradjustments in which head distribution, stream orientation, and
42
Figure 15
42
T.4SK.
Figure 15
43groundwater flow paths generated by MODPATH (Pollock, 1989) wereobserved.
Cell by cell interflow datum is used by the post processingpackage Zonebudget (Harbaugh, 1990) to calculate groundwaterinterflow rates between zones* Table 4 summaries interflow ratesbetween one particular zone, and those zones and/or to specifiedhead boundaries which surrounding it.
Zone Goose Lake in Figure 15 represents a groundwaterrecharge area in which Goose Lake and the upper portion of GooseLake Outlet discharge water into the system. Surrounding zoneGoose Lake are zone Silver/Cherry Creek and zone Goose Lake OutletThe boundaries of zone Silver/Cherry Creek were constructed toisolate a subregion of groundwater which is thought to be the sourceof recharge to the upper reaches of Silver and Cherry Creeks. Theboundaries surrounding zone Goose Lake Outlet isolate an area whichcontains the lower reaches of Goose Lake Outlet and a portion of theMiddle Branch of the Escanaba River. Table 4 indicates that their isno interflow from zone Silver/Cherry Creek to zone Goose Lake, andthat only a small rate of interflow (1.1 cubic feet/second) from zoneGoose Lake Outlet enters zone Goose Lake. Table 4 also indicatesthat a large rate of interflow (13.9 cubic feet/second) occurs fromzone Goose Lake to zone Silver/Cherry Creek, and that only a smallrate of interflow (0.17 cubic feet/second) occurs from zone GooseLake to zone Goose Lake Outlet. This data suggests that groundwaterin zone Goose Lake which is recharged from leakage by Goose Lakeand Goose Lake Outlet will eventually act as a source of recharge forgroundwater in zone Silver/Cherry Creek.
The boundaries surrounding zone Cedar Creek were constructedto isolate an area containing Qthe headwaters of Cedar Creek. Theboundaries surrounding zone Big Creek were constructed to isolatean area containing the headwaters of Big Creek and its tributaries.Only a small rate of interflow occurs between zone Goose LakeOutlet and zones Silver/Cherry, Cedar, and Big Creeks as indicatedby Table 4.
A significant rate of interflow occurs between zoneSilver/Cherry Creek and zone Cedar Creek. Attempts to redraw
44
Table 4
Zone BudgetInterflow rates in cubic feet per second
zone 1 = zone Chocolay River zone 6= zone Goose Lake Outletzone 2= zone Silver/Cherry Creek zone 7 = zone Escanabazone 3- zone Cedar Creek zone 8 = zone Goose Lakezone 4 = zone Big Creek zone 9 = zone Pleisser Lakezone 5 = zone West Branch of the zone 10 = Intermediate zone
of Chocolay zone 12 = Intermediate zone
Flow Budget for zone 1In: Out
Constant Head =4.96 Constant Head = 17.20Zone 2 to 1 = 7.67 Zone 1 to 2 = 6.07Zone 4 to 1 = 0.86 Zone 1 to 4 = 0.99Zone 5 to 1 = 0.65 Zone 1 to 5 = 0.45Zone 9 to 1 = 2.98 Zone 1 to 9 = 0.00Zone 10to 1 = 1.98 Zone 1 to 10 .= 0.03Zone 12 to 1 - 1.84 Zone 1 to 12 - 1.07
Total in =20.94 Total out = 25.81In - Out - -4.87
Flow Budget for Zone 2In: Out
Constant Head =0.00 Constant Head = 0.00Zone 1 to 2 = 6.07 Zone 2 to 1 = 7.67Zone 3 to 2 » 5.97 Zone 2 to 3 =10.20Zone 6 to 2 =1.15 Zone 2 to 6 = 0.00Zone 8 to 2 =13.98 Zone 2 to 8 = 0.00Zone 9 to 2 = 7.73 Zone 2 to 9 = 6.73Zone 10 to 2 = 0.10 Zone 2 to 10 = 0.00Total in = 35.00 Total out = 24.60
In-Out =10.40
45
Table 4 (cont'd)
Flow Budaet for Zone 3In: OutConstant Head =0.00 Constant Head = 0.00Zone 2 to 3 =10.21 Zone 3 to 2 = 5.97Zone 4 to 3 = 0.68 Zone 3 to 4 = 0.89Zone 6 to 3 = 0.33 Zone 3 to 6 » 0.00Zone 10 to 3 = 1.88 Zone 3 to 10 = 3.22Total in =13.10 Total out = 10.08
In-Out =3.02
Flow Budget for Zone 4In: Out-
Constant Head =0.00 Constant Head = 0.00Zone 1 to 4 = 0.99 Zone 4 to 1 » 0.86Zone 3 to 4 = 0.88 Zone 4 to 3 = 0.68Zone 5 to 4 = 1.61 Zone 4 to 5 = 0.93Zone 6 to 4 = 0.72 Zone 4 to 6 = 0.08Zone 7 to 4 = 1.66 Zone 4 to 7 =1.49Zone 10 to 4 = 0.08 Zone 4 to 10 = 1.22Zone 12 to 4 =2.10 Zone 4 to 12 = 1.14Total in = 7.98 Total out - 6.40
In-Out =1.49
Flow Budget for Zone 5In: Out
Constant Head =0.99 Constant Head = 9.81Zone 1 to 5 = 0.45 Zone 5 to 1 = 0.64Zone 4 to 5 = 0.92 Zone 5 to 4 = 1.61Zone 7 to 5 = 0.66 Zone 5 to 7 = 0.65Zone 12 to 5= 0.53 Zone 5 to 12 = 0.61Total in =3 .55 Total out = 13.32
In-Out =-9.77
46
Table 4 (cont'd)
Flow Budget for Zone 6In:Constant HeadZone 2 to 6Zone 3 to 6Zone 4 to 6Zone 7 to 6Zone 8 to 6Total in
0.00= 0.00= 0.00= 0.00= 0.20= 0.17= 0.37
In-Out
OutConstant Head = 0.00Zone 6 to 2 = 1.10Zone 6 to 3 « 0.33Zone 6 to 4 = 0.72Zone 6 to 7 = 0.74Zone 6 to 8 = 1.10Total out = 3.99
-3.62
Flow Budget for Zone 7In:
Constant Head =0.25Zone 4 to 7Zone 5 to 7Zone 6 to 7Total in
1.490.650.743.13
In-Out
OutConstant Head = 18.24Zone 7 to 3 = 1.66Zone 7 to 5 = 0.66Zone 7 to 6 = 0.20Total out = 20.76
-17.63
Flow Budget for Zone 8In: Out
Constant Head =0.00 Constant Head = 0.00Zone 2 to 8 = 0.00 Zone 8 to 2 =13.94Zone 6 to 8 = 1.10 Zone 8 to 6 = 0.17Total in =1.10 Total out =14.11
In-Out =-13.01
47
Table 4 (cont'd)
Flow Budget for Zone 9In: Out
Constant Head =0.00 Constant Head = 0.00Zone 1 to 9 - 0.00 Zone 9 to 1 = 2.98Zone 2 to 9 =6.73 Zone 9 to 2 - 7.74Total in =6.73 Total out =10.72
In-Out —3.99
Flow Budget for Zone 10In: OutConstant Head =0.00 Constant Head = 0.00Zone 1 to 10 » 0.04 Zone 10 to 1 = 1.98Zone 2 to 10 - 0.00 Zone 10 to 2 =0.10Zone 3 to 10 = 3.22 Zone 10 to 3 = 1.88Zone 4 to 10 = 1.22 Zone 10 to 4 = 0.08Total in = 4.48 Total out - 4.04
In-Out = 0.44
Flow Budget for Zone 12In: Out
Constant Head =0.00 Constant Head = 0.00Zone 1 to 12 = 1.07 Zone 12 to 1 = 1.84Zone 4 to 12 = 1.14 Zone 12 to 4 = 2.10Zone 5 to 12 = 0.61 Zone 12 to 5 = 0.53Total in - 2.82 Total out = 4.47
In-Out =-1.65
48these boundary lines to reduce the rate of interflow between thesetwo zones failed, and suggest that these two zones could not behydrologically separated. Small rates of interflow exist betweenzone Cedar Creek and zone Big Creek, allowing zone Silver/CherryCreek and zone Cedar Creek to be partially hydrologically separatedfrom zone Big Creek (see Figure 15). This suggests that groundwaterrecharged from Goose Lake lying in the northern region of themodeled area in zones Silver/Cherry and Cedar Creeks ishydrologically distinct from groundwater in the southern region ofthe modeled area in zone Big Creek.
The boundaries surrounding zone Chocolay were constructed toisolate an area of groundwater discharge, and area where streamwater tributaries converge on the Chocolay River. The boundariessurrounding zone Pleisser were constructed to isolate the PleisserLake area. Zones Chocolay, Silver/Cherry Creek and Pleisser displayhigh rates of interflow between each other, suggesting thatgroundwater in this area may be mixing.
