. .iaaslkmna...
.a‘mmhm, : .. Nb.

It.
fififitmn
.c Kira—4%..

j .5: nu.
qufiflnuflwfl.
I“:

1.5.5. .
3.3-.

J

.911.

. . at.
. ‘ . 1.17;”:

."a' 'lSITJ’s‘h an
annaithflo

‘ . ‘ .-

, ,1 “Li.

Hflv‘.tit(:. .I Ann.
I.

:14.) c}...
51...! ‘ x.
r

.I 2:90.53
...§vl...,.:.r!:.,u.x 3!. .
. .342. :1...

 

This is to certify that the

thesis entitled

Characterization of Aquifer Heterogeneity and Tracer Test
Simulations by Incorporating Geologic Information in the
Form of Outcrop Analog and Well Core at a Distal and
Medial Outwash Aquifer in Schoolcraft, Michigan

presented by

Susanne Evelyn Biteman

has been accepted towards fulfillment
of the requirements for

' a
Master 5 degree In Geology

 

 

18 Major professor

Date August 22, 2002

 

0-7639 MS U i: an Affirmative Action/Equal Opportunity Institution

 

 

LIBRARY
Michigan State
University

 

 

 

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

 

OCT 0 1 2006

 

W i ii at n

U

 

“ V V v "

 

 

 

 

 

 

 

 

 

 

 

 

6/01 chlRC/Datoouopes-ms

CHARACTERIZATION OF AQUIFER HETEROGENEITY AND TRACER TEST
SIIVIULATIONS BY INCORPORATING GEOLOGIC INFORMATION IN THE
FORM OF OUTCROP ANALOG AND WELL CORE AT A DISTAL AND MEDIAL
OUTWASH AQUIFER IN SCHOOLCRAFT, MICHIGAN

By

Susanne Evelyn Biteman

A THESIS

Submitted to
Michigan State University
In partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Department of Geological Sciences

2002

r y ‘N

ZL'

l‘.

MU.

J,
Li

it.

-f'
iii

3»

ABSTRACT

CHARATERIZATION OF AQUIFER HETEROGENEITY AND TRACER TEST
SIMULATIONS BY INCORPORATIN G GEOLOGIC INFORMATION IN THE
FORM OF OUTCROP ANALOG AND WELL CORE AT DISTAL AND MEDIAL
OUTWASH AQUIFER IN SCHOOLCRAFT, MICHIGAN

by

Susanne Evelyn Biteman

The integration of geostatistics and geologic data in the form of outcrop analogs
and well core produced detailed models of aquifer heterogeneity using two methods,
zonal kriging and transition probability geostatistics (Carle and Fogg, 1996). This study
was conducted at the Schoolcrafl Plume A site, located in Kalamazoo County, Michigan.
Sediments at this site were deposited as outwash medial and distal to the Kalamazoo
Moraine. Using core data from the site and outcrop analog data from nearby sand and
gravel pits, the aquifer was separated into major sedimentologic units to preserve abrupt
changes in hydraulic conductivity and to better honor the assumption of stationarity. For
the first geostatistical method, zonal kriging was used to estimate hydraulic conductivity
distribution. After geostatistic estimation all the zones were combined to produce a
complete aquifer realization. A non-zonal krige was also generated using the same
conditioning data but without the geologically based zones. Both the zonal krige and the
non-zonal krige hydraulic conductivity values were used as inputs for three dimensional
ground water flow models and transport simulations. The zonal krige more accurately

simulates the migration of an injected tracer pulse.

(m

(I 3'

ACKNOWLEDGEMENTS

I extend my thanks to my advisor Dr. Gary Weissmann who has helped me over
the past two years by providing stimulating discussion which have lead me to have a
greater knowledge and understanding of the many aspects of hydrogeology. His
insightful reviews of this manuscript were greatly appreciated. Committee members Dr.
David Hyndman and Dr. Grahame Larson provided much knowledge and experience ~
throughout the different stages of this project. M.S. Phanikumar’s help with the
groundwater modeling and transport simulations was irreplaceable. I thank Michelle Vit
for her help conducting the field work portion of this study. This research was funded in
part by Michigan Department of Environmental Quality (Contract # Y403 86) and in part
by Michigan State University through Dr. Gary Weissmann’s set up fund.

I would thank my parents and my brother who have supported me throughout my
educational endeavors. I would also like to thank my friends who have experienced
graduate school with me, most of all Jennifer Wade, who has been a wonderful friend to

me. I am especially grateful to my husband, Adam for his absolute love and support.

iii

TABLE OF CONTENTS

LIST OF TABLES ................................................................................ vii
LIST OF FIGURES ................................................................................ x
I. INDRODUCTION AND SCOPE OF WORK ............................................. 1
INTRODUCTION ........................................................................... 1
PURPOSE OF THE STUDY .............................................................. 3
THESIS OUTLINE .......................................................................... 5

II. PREVIOUS WORK ............................................................................. 7
GEOLOGY .................................................................................. 7
Regional Bedrock Geology ....................................................... 7
Glacial History of Southwestern Michigan ..................................... 9

General Stratigraphy and Sedimentology of Glaciofluvial Deposit ....... 12
HYDROGEOLOGY ....................................................................... 20
Regional Hydrogeology ......................................................... 20

History of Contamination at Plume A ......................................... 23
Bioremediation of Plume A ..................................................... 24

III. INTEGRATION OF SEDIMENTOLOGIC AND HYDROGEOLOGIC
PROPERTIES FOR INPROVED TRANSPORT SIMULATIONS: DETAILED
CHARACTERIZATION OF A GLACIAL OUTWASH AQUIFER AT THE

SCHOOLCRAFI‘ BIOREMEDIATION SITE, MICHIGAN ............................ 25
INTRODUCTION ........................................................................ 26
SITE DESCRIPTION ..................................................................... 28
CORE DESCRIPTIONS .................................................................. 32

iv

Methods ............................................................................ 32

Core facies interpretation .............................................. 34
Schoolcrafi Stratigraphic Evaluation ........................................... 37
OUTCROP ANALOG .................................................................... 41
GEOSTATISTICS ........................................................................ 42
GROUNDWATER MODELING AND TRACER TEST SIMULATIONS ...... 44
RESULTS AND DISCUSSION ......................................................... 49
CONCLUSTIONS ........................................................................ 5 1
IV. TRANSITION PROBABILITY GROSTATISTICS ................................. 52
INTRODUCTION ........................................................................ 52
STRATIFICATION OF ZONES ....................................................... 54
Explanation of Hydrofacies from Categories ................................. 54
Top Zone .................................................................. 56
Bottom Zone ............................................................. 61
MARKOV CHAIN MODELS ........................................................... 64
Top Zone Markov Chain Models ............................................... 64
Bottom Zone Markov Chain Models .......................................... 69
CONDITIONAL SEQUENTIAL INDICATOR SMULATION .................. 77
COMBINING THE ZONES ............................................................. 77
CONCLUSIONS .......................................................................... 81
V. CONCLUSIONS .............................................................................. 82
DEPOSTIONAL EVIRONMENT OF THE AQUIFER BELOW
SCHOOLCRAFT ........................................................................... 82
AQUIFER CHARACTERIZATIONS .................................................. 83

Kriging Method ................................................................... 84
Transition Probability Method ................................................. 84

APPENDIX A. STRATIGRAPHY OF THE SCHOOLCRAFT PLUME A SITE ..86

CORE DATA .............................................................................. 86

INTERPRETATIONS OF DEPOSITIONAL ENVIRONMENT OF THE

AQUIFER BENEATH SCHOOLCRAFT ............................................. 86
APPENDIX B. OUTCROP ANALOGS .................................................... 123
REFERENCES ................................................................................... 134

vi

TABLE 3.1

TABLE 3.2

TABLE 3.3

TABLE 3.4

TABLE 4.1

TABLE 4.2

TABLE 4.3

TABLE 4.4

TABLE 4.5

TABLE 4.6

LIST OF TABLES

Anisotropy values for vertical K and horizontal K, determined by
comparing repacked to non-repacked core .................................... 35

Vertical K (cm/s) of facies vs. sorting from well P18 ........................ 35

Horizontal K (cm/s) statistics for the four stratigraphic zones and the non-
zonal model ........................................................................ 4O

Variogram parameters for zonal krige and non—zonal krige. A11
variograms fit using exponential model ........................................ 45

Mean grain size (GS) in Phi unit, corresponding Wentworth size class
(Boggs, 2001), mean standard deviation (SD), sorting based on standard
deviation (Boggs, 2001), and number of data points in each zone for the
transition probability (T .P.) zones and the four zonal kriging zones. Zone
one is the same for the transition probability and zonal krige zones ......... 55

Proportions of hydrofacies in the top and bottom zone ...................... 55

Hydrofacies and corresponding assigned category for the top zone with
high K value, low K value, average K value, and number of points ....... 57

Hydrofacies and corresponding assigned category for the bottom zone
with high K value, low K value, average K value, and number of
points ................................................................................ 57

Embedded transition probability matrices for the top zone. These matrices
are read as transition probabilities from the row hydrofacies to the column
hydrofacies. (hydrofacies labels: PMS, poorly sorted medium sand; PCS,
poorly sorted coarse sand; MWFS, moderately to well sorted fine sand;

PF S, poorly sorted fine sand; VFS, very fine sand. Other labels: L, mean
length. Bold numbers indicate background category with computed values
listed in the table.) ................................................................. 66

Embedded transition probability matrices for the bottom zone. These
matrices are read as transition probabilities from the row hydrofacies to
the column hydrofacies. (hydrofacies labels: PGVCS, pebbles, gravel and
very coarse sand; WMS, well sorted medium sand; PCS, poorly sorted
coarse sand; PMS, poorly sorted medium sand; MWFS, moderately to
well sorted fine sand. Other labels: L, mean length. Bold numbers

vii

TABLE 4.7

TABLE A.l

TABLE A.2

TABLE A.3

TABLE A.4

TABLE A.5

TABLE A.6

TABLE A.7

TABLE A.8

TABLE B.l

indicate background category with computed values listed in the table).
........................................................................................ 73

W.r.t independent transition frequencies matrices for the bottom zone.
These matrices are read as independent transition probabilities from the
row hydrofacies to the column hydrofacies. (hydrofacies labels: PGVCS,
pebbles, gravel and very coarse sand; WMS, well sorted medium sand;
PCS, poorly sorted coarse sand; PMS, poorly sorted medium sand;
MWFS, moderately to well sorted fine sand. Other labels: L, mean length.
Bold numbers indicate background category with computed values listed
in the table.) ........................................................................ 75

Key for the core descriptions .................................................... 92

Well D2. X, Y, and Z location, Log K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain
size (Boggs, 2001)) .............................................................. 116

Well D4. X, Y, and Z location, Log K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain
size (Boggs, 2001)) .............................................................. 117

Well D6. X, Y, and Z location, Log K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain
size (Boggs, 2001 )) .............................................................. 1 18

Well D8. X, Y, and Z location, Log K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain
size (Boggs, 2001)) .............................................................. 119

Well D10. X, Y, and Z location, Log K value, mean grain size
(determined by sieve analysis), and sorting (determined by standard
deviation of grain .size (Boggs, 2001)) ....................................... 120

Well D12. X, Y, and Z location, Log K value, mean grain size
(determined by sieve analysis), and sorting (determined by standard
deviation of grain size (Boggs, 2001)) ........................................ 121
Well D14. X, Y, and Z location, Log K value, mean grain size
(determined by sieve analysis), and sorting (determined by standard
deviation of grain size (Boggs, 2001)) ........................................ 122

Key for photomosaics ........................................................... 123

viii

TABLE B.2 Description of outcrop analog shown in Figure B.5. Bed number, mean
grain size, grain size range, sorting, facies, and apparent trends are
given ............................................................................... 129

TABLE 33 Description of outcrop analog shown in Figure 85. Bed number, mean
grain size, grain size range, sorting, facies, and apparent trends are
given ............................................................................... 13 1

TABLE B.4 Description of outcrop analog shown in Figure B.5. Bed number, mean

grain size, grain size range, sorting, facies, and apparent trends are
given ............................................................................... 133

ix

FIGURE 2.1

FIGURE 2.2

FIGURE 2.3

FIGURE 2.4

FIGURE 2.5

FIGURE 2.6

FIGURE 2.7

FIGURE 2.8

FIGURE 3.1

FIGURE 3.2

FIGURE 3.3

LIST OF FIGURES

Bedrock map of the Lower Peninsula of Michigan. Kalamazoo County is
outlined in black. Modified from Milstein, 1987. Image is presented in
color .................................................................................. 8

Location of Moraines in southwestern Michigan. Notice the interlobate
boundary marking the extent of both the Saginaw and Lake Michigan
lobes. Kalamazoo County is marked by a bold outline. Modified from
Monaghan et al., 1986 ............................................................ 10

Till sheets underlying moraines in southwestern Michigan. Modified from
Monaghan et al.,1986 ............................................................ 11

Figure 2.4. Study areas of Steinmann (1994) and Baresse (1991).
Boundary of Prarie Rhonde Fan as described by Steinmann, maked by
Steinmann’s study area. Modified from Baresse, 1991 and Steinmann,
1994 .................................................................................. 13

Principal bar types and where they form in a stream. Modified from
Miall, 1977 .......................................................................... 16

Vertical facies models of Scott-type, Donjeck-type and Bijou Creek type.
Modified from Miall, 1977; 1996 ................................................ 17

Block diagrams showing the braided stream patterns associated with
Scott-type streams (A), Donjeck-type streams (B), and Platte type Streams
(C). Modified from Miall, 1985 ................................................. 18

Study area of Lipinski (2002) regional model. Lipinski’s smaller scale
transport model is outlined by the bold rectangle near Schoolcraft.
Modified from Lipinski, 2002 ................................................... 22

Kalamazoo County Surficial Geology, location of Kalamazoo Moraine
and outcrop site used for this study. Modified from Monaghan and
Larson, (1982) ...................................................................... 29

Schoolcraft Plume A site. Location of biocurtian is marked by arrow.
Modified from Mayotte (1991) ................................................... 31

Location of delivery wells (D) and monitoring wells (MW) at the Plume A
Schoolcrafi site ..................................................................... 33

Fl

FIGL'

FIGL’P

FIGURE 3.4

FIGURE 3.5

FIGURE 3.6

FIGURE 3.7

FIGURE 3.8

FIGURE 3.9

FIGURE 4.1

FIGURE 4.2

FIGURE 4.3

FIGURE 4.4

FIGURE 4.5

Core description of P18 showing zonal boundaries (marked by arrows)
and measured vertical K .......................................................... 38

Cross-section of log K data points used for interpolation. Figure 3.5a is
transect through the delivery wells, Figure 3.5b is a transect through the
down gradient wells (Figure 3.3 for location). Zonal boundaries picked at
16 m bgs, 20 m bgs, and 21 m bgs. Scalebar is in log K (cm/s) ............... 39

A portion of the outcrop analog showing the stratigraphic character of the
outwash material with a shovel for scale. Image is presented in color. . ..43

Zonal (top) and non-zonal (bottom) kriged results. Arrows mark
boundaries of zones in zonal krige. Note that abrupt changes in K are
preserved in zonal krige approach. Scalebar is in log K (cm/s) ............ 46

Discretization of grid in plan view, delivery well gallery is located near
the center of the grid, marked by light colored squares where discretization
is the finest ......................................................................... 48

Measured tracer test from site (x) with simulated zonal krige (black line)
and simulated non-zonal krige (gray line) ..................................... 50

Top zone of Schoolcraft site, 1-5 are categories of hydrofacies,
corresponds to Table 4.3. Top diamond is high K value, bottom diamond
is low K value and middle diamond is mean K value.

Cross-section of categories in top zone. Category 1, poorly sorted
medium sand (PMS); category 2, poorly sorted coarse sand (PCS);
category 3, moderately to well sorted fine sand (MWFS); category 4,
poorly sorted fine sand (PF S); category 5, very fine sand (VF S). X axis is
depth (m). Image is presented in color .......................................... 60

Bottom zone of Schoolcrafi site, 1-5 categories of hydrofacies,
corresponds to Table 4.4. Top diamond is high K value, bottom diamond
is low K value and middle diamond is mean K value ........................ 62

Matrix of vertical (z)-direction transition probabilities showing core data
as measurements (open circles) and the Markov chain model (solid line)
from the core measurements. The diagonal elements represent auto-
transition probabilities within each category, and the off diagonal elements
represent transition probabilities between categories ........................ 63

Matrix of perpendicular to flow (y)-direction transition probabilities

showing core data as measurements (open circles) and the Markov chain
model (solid line) from the core measurements. The diagonal elements

xi

represent auto-transition probabilities within each category, and the off
diagonal elements represent transition probabilities between categories..68

FIGURE 4.6 Cross-section of category in bottom zone. Category 1, pebbles, granules
and very coarse sand (PGVCS); category 2, poorly sorted medium sand
(PMS); category 3 poorly sorted coarse sand (PCS); category 4, poorly
sorted medium sand (PMS); category 5, moderately and moderately to
well sorted fine sand (MWFS). Image is presented in color ................ 70

FIGURE 4.7 Outcrop analog with 0.25 m grid overlay. Image is presented in color. . .71

FIGURE 4.8 Matrix of vertical (z)—direction transition probabilities showing core data
as measurements (open circles) and the Markov chain model (solid line)
from the core measurements. The diagonal elements represent auto-
transition probabilities within each category, and the off diagonal elements
represent transition probabilities between categories ........................ 72

FIGURE 4.9 Matrix of perpendicular to flow (y)-direction transition probabilities
showing core data as measurements (open circles) and the Markov chain
model (solid line) from the core measurements. The diagonal elements
represent auto-transition probabilities within each category, and the off
diagonal elements represent transition probabilities between categories..76

FIGURE 4.10 A realization of the Schoolcrafi plume A site based on transition
probability/ Markov chain geostatistics in two zones. This image is
presented in color .................................................................. 79

FIGURE 4.11 A realization of the Schoolcrafi plume A site based on transition
probability/ Markov chain geostatistics in two zones. This image is
presented in color .................................................................. 80

FIGURE A.1 Location of delivery wells (D), monitoring wells (MW), peizometers (P)
and other wells at the Schoolcraft plume A site.

FIGURE A.2a Core descriptions of well P18. Vertical K measurements shown in second
column .............................................................................. 93

FIGURE A.2b Core descriptions of well P18. Vertical K measurements shown in second
column .............................................................................. 94

FIGURE A.2c Core descriptions of well P18. Vertical K measurements shown in second
column .............................................................................. 95

FIGURE A.2d Core descriptions of well P18. Vertical K measurements shown in second
column .............................................................................. 96

xii

FIGURE A.2c Core descriptions of well P18. Vertical K measurements shown in second

column .............................................................................. 97
FIGURE A.3a Core descriptions of well P6 ..................................................... 98
FIGURE A.3b Core descriptions of well P6 ..................................................... 99
FIGURE A.4a Core descriptions of well P7 ................................................... 100
FIGURE A.4b Core descriptions of well P7 ................................................... 101
FIGURE A.5a Core descriptions of well P8 ................................................... 102
FIGURE A.5b Core descriptions of well P8 ................................................... 103
FIGURE A.50 Core descriptions of well P8 ................................................... 104
FIGURE A.5d Core descriptions of well P8 ................................................... 105
FIGURE A.5e Core descriptions of well P8 ................................................... 106
FIGURE A.5f Core descriptions of well P8 ................................................... 107
FIGURE A.6a Core descriptions of well P9 ................................................... 108
FIGURE A.6b Core descriptions of well P9 ................................................... 109
FIGURE A.7a Core descriptions of well P11 .................................................. 110
FIGURE A.7b Core descriptions of well P11 .................................................. 111
FIGURE A.8a Core descriptions of well P12 .................................................. 112
FIGURE A.8b Core descriptions of well P12 .................................................. 113
FIGURE A.8c Core descriptions of well P12 .................................................. 114
FIGURE A.8d Core descriptions of well P12 .................................................. 115

FIGURE 3.1 Photomosaic of outcrop analog at Azon pit. Yellow lines represent
bounding surfaces. Thinner lines represent lower order bounding surfaces.
Shovel for scale. Image is presented in color ............................... 124

xiii

FIGURE 82

FIGURE 3.3

FIGURE B.4

FIGURE 3.5

FIGURE B.6

FIGURE B.7

Photomosaic of outcrop analog at Azon pit. Yellow lines represent
bounding surfaces. Thinner lines represent lower order bounding surfaces.
Shovel for scale. Image is presented in color ............................... 125

Photomosaic of outcrop analog at Azon pit. Yellow lines represent
bounding surfaces. Thinner lines represent lower order bounding surfaces.
All measurements are in centimeters and are shown by pink arrows.
Image is presented in color ..................................................... 126

Photomosaic of outcrop analog at Stadler sand and gravel pit. Yellow
lines represent bounding surfaces. Thinner lines represent lower order
bounding surfaces. Height of outcrop is approximately 12 meters. Image
is presented in color ............................................................. 127

Photomosaic of outcrop analog at Stadler sand and gravel pit. Yellow
lines represent bounding surfaces. Thinner lines represent lower order
bounding surfaces. Measurements are in centimeters and are shown by
pink arrows. Bed numbers correspond to descriptions given in Table B.2.
Image is presented in color .................................................... 128

Photomosaic of outcrop analog at Stadler sand and gravel pit. Yellow
lines represent bounding surfaces. Thinner lines represent lower order
bounding surfaces. Measurements are in centimeters and are shown by
pink arrows. Bed numbers correspond to descriptions given in Table B.3.
Image is presented in color .................................................... 130

Photomosaic of outcrop analog at Stadler sand and gravel pit. Yellow
lines represent bounding surfaces. Thinner lines represent lower order
bounding surfaces. Measurements are in centimeters and are shown by
pink arrows. Bed numbers correspond to descriptions given in Table B.4.
Image is presented in color .................................................... 132

xiv

Chapter One

Introduction and Scope of Work

Introduction

Inaccurate characterizations of aquifer heterogeneity commonly result in
simulations of groundwater flow and contaminant transport that are inadequate to fully
understand the migration of contaminant plumes. The need to more accurately simulate
the migration of fluids has led to better heterogeneity characterization approaches in
aquifers through incorporation of more geologic information. Koltermann and Gorelick
(1996) stated that in order to accurately model and predict containment transport, we
must be able to predict the spatial variability of hydraulic properties. Many studies e. g
Borden (Sudicky, 1986); Cape Cod (Hess et al., 1991); Twin Lake (Moltyaner and
Killey, 1998; MADE site (Adams and Gelhar, 1992; Boggs er al., 1992; Renfeldt et al.,
1992) have shown that the resolution of acquired data necessary for predicting solute
transport cannot be achieved by standard subsurface investigation techniques such as
pumping tests, flow meter measurements or core analysis. Klingbeil et al. (1999) stated
that aquifer stress tests by pumping yield parameters at a scale much larger than the
typical length of structures in a heterogeneous aquifer. When characterizing an aquifer,
focus should be placed on fine scale geologic properties rather than on larger scale
hydraulic responses of the wells to accurately predict fluid flow and solute transport
(Klingbeil et al., 1999). Weissmann and Fogg (1999) used the geology and stratigraphy

as a framework for geostatical realizations in order to preserve stationarity.

Most geologic information comes from boreholes. Core from boreholes can be
described for geologic features, or core can be described using standard engineering
geology techniques. Geophysical logs may also be used in boreholes to test resistivity,
conductivity, porosity, density and other parameters. Although measurements that come
from a borehole may yield enough detailed information to characterize heterogeneity in
the vertical direction, the spacing between boreholes is generally too great to yield
sufficient information on the heterogeneity in the horizontal direction.

One approach to increasing lateral geologic information is studying an outcrop
composed of similar stratigraphy and similar lithologies as the aquifer of interest. This
outcrop may be viewed as an analog to the aquifer (“outcrop analog”). Miall, (1985),
used outcrop analogs to interpret facies architecture in fluvial settings, however using
them as a characterization tool of the subsurface to create a geologic based
characterization of an aquifer or reservoir is a relatively new idea.

Many recent researchers have used outcrop analogs as an aid to subsurface
characterization of aquifers and reservoirs (Davis et al., 1993; Davis et al., 1997;
Robinson and McCabe, 1997 ; Anderson et al., 1999; Bersezio et al., 1999; Klingbeil et
al., 1999; and Hornung and Aigner, 1999). Anderson et a1. (1999) quantified the spatial
distribution of hydrofacies in braided stream deposits using two approaches 1) a small
scale outcrop analog and 2) a large scale model of a medial braided stream deposit. In
both approaches, three-dimensional images showing interconnectedness of high K units
were used as inputs to a groundwater flow model and flow paths were analyzed by
following the transport of particles. They found that particles that were uniformly

distributed at the up gradient end of the model were clustered along preferential flow

paths during transport, showing that connection among high permeability facies is a
critical factor in hydrogeological investigations involving solute transport. Bersezio et al.
(1999) found strong anisotropic behavior in two-dimensional groundwater models based
on an outcrop analog exposure in a delta environment, which was attributed to the
layering of sedimentologic units. Davis et al. (1993) suggested that architectural
elements may provide a role in evaluating aquifer heterogeneity at the scale of meters.
Davis et al. (1997) used three different meter scale outcrops to demonstrate that fluvial
bounding surfaces provide a good geological basis for conceptualizing and modeling
heterogeneity in alluvial deposits. Klingbeil et al. (1999) defined 5 hydrofacies from 23
lithofacies at an outcrop analog composed primarily of gravel. These hydrofacies were to
be used in subsequent hydrogeological analysis. Homung and Aigner (1999)
characterized an aquifer using fluvial architectural elements (Miall, 1985, 1996) on a
large alluvial plain outcrop analog. Nine architectural elements were recognized in their
study, and information collected in this study provided a database for subsequent fluid
flow simulations. Robinson and McCabe (1997) measured geometries of sand and shale
bodies in a braided fluvial environment, generated a three-dimensional stratigraphic
model of the reservoir, and simulated flow through the model. They found that flood
deposits affected flow in the simulation by having sufficient thickness and lateral extent

to compartmentalize reservoir flow units and increase vertical flow tortuosity.

Purpose of the study
The purpose of this study was to apply stratigraphic methods of aquifer

characterization of Weissmann and Fogg (1999) to a smaller scale in a glacial outwash

setting. This method uses stratigraphic breaks identified in core to define zones of
similar deposits and model heterogeneity within the stratigraphic framework. This is
important because it provides a framework to better honor the stationary assumption.
The mean and standard deviation of the zones is more spatially consistent within the
zones than in the entire aquifer. This method also better preserves abrupt changes in K
that would commonly be smoothed with traditional geostatistics.

This method was implemented at the Schoolcraft Plume A site, located in the
village of Schoolcrafl southwest of Kalamazoo in Kalamazoo County, Michigan. The
Schoolcrafi Plume A site is located in a 27 m thick sand and gravel aquifer that overlies a
regional till layer that acts as an aquitard, with a water table at 4.5 m below ground
surface. Seven contaminant plumes (A-G) have been identified in the Schoolcrafi area.
The focus of this project is a transect across the down gradient end of plume A, at the site
of the Schoolcraft bioaugrnentation project (Hyndman et al. , 2000; Dybas et al., 2002).
Plume A is roughly 1.2 km long and 90 m wide, and extends from 8 to 26.5 m below
ground surface (Hyndman et al., 2000). This study site was selected for this project for
the following reasons: 1) much closely spaced core data is available, 2) hydrologic and
injected tracer test measurements exist for the site, 3) it is a site of ongoing groundwater
investigation, and 4) detailed characterization approaches are needed to improve previous
groundwater flow models and transport simulations.

The main approach used to characterize aquifer heterogeneity at the Schoolcraft
site was zonal kriging, which is a method where the aquifer parameters are kriged within
zones of similar geology. The mean and standard deviation of K is more spatially

consistent within each zone, thereby better honoring the stationarity assumption. To test

the effects of zonal kriging, a non-zonal krige distribution of K was estimated using the
same conditioning data but without dividing the data into stratigraphically based zones.
Transition probability geostatistics was used as another method to characterize the
heterogeneity. Both the zonal krige and the transition probability geostatistics methods
were used within a stratigraphic framework provided by core descriptions, with lateral
correlation lengths supplemented by measured data from outcrop analogs. Stratigraphic
breaks in core identifying major bounding surfaces were used to define stratigraphic
zones used in zonal kriging. To test the hypothesis that a better aquifer characterization
would lead to a better tracer test comparison, both the zonal and non-zonal kriged aquifer
characterizations were used as inputs to groundwater flow models and tracer test
simulations. A closer match of the simulated tracer to the measured tracer breakthrough
fiom field tracer tests indicates a positive test of the hypothesis. Running tracer
simulations through the transition probability geostatistics results was beyond the scope

of this thesis.

Thesis outline

This document is divided into four main sections with subsequent appendices:

I The first section, covered by chapter two, describes the bedrock and glacial
geology of the study region. A brief review is given of previous hydrological
studies in Kalamazoo County, specifically in the Schoolcrafi Aquifer, and the
history of contamination and bioremediation of Plume A.

I The second section, covered by chapter three, presents an approach to
incorporate geostatistical estimation into a stratigraphic framework utilizing

zonal kriging to better represent aquifer characterization. Using a hi gh-

resolution three-dimensional flow model and transport simulation, simulation
of an injected tracer pulse was improved. This chapter forms a manuscript
that has been submitted for publication (Biteman, et al. submitted).

The third section, covered by chapter four, applies a transition probability /
Markov chain approach in a stratigraphic framework to simulate the
distribution of K. The stratigraphy provides a framework in which the K
distribution can be simulated while better honoring the stationarity
assumption.

The fourth section, covered by chapter five, reviews the overall conclusions of
the work conducted for this thesis.

Subsequent appendices are attached and list core descriptions, outcrop analog
photomosaics and descriptions, and data tables that were used / collected in

this study.

Chapter 2

Previous Work

Geology
Regional Bedrock Geology

Kalamazoo County overlies the southwestern rim of the Michigan basin. The
Michigan Basin is nearly 5 km deep with a radius of approximately 250 km (Howell and
van der Pluijm, 1990). The Michigan Basin reveals a number of subsidence reactivations
and cessations (Sleep and $1033, 1978). Howell and van der Pluijm (1999) cite evidence
to support their theory that episodes of different types of subsidence occurred from the
Cambrian to Pennsylvanian to form the Michigan Basin. The Mississippian Marshall
Formation subcrops in the most northeastern comer of Kalamazoo County and the
Mississippian Coldwater Shale Formation subcrops over the remainder of Kalamazoo
County (figure 2.1) (Forstat and Sorensen, 1982). In the early Mississippian, the
Coldwater Shale was deposited as a fine grained mud in an off shore environment and at
the close of the early Mississippian, a reduction of the seas caused much of southern
Michigan to become a beach environment (Dorr and Eschman, 1970). The Marshall
Formation was deposited at this time and is composed of the Marshall Sandstone and
Napoleon Sandstone. The Coldwater Shale is blue-gray in color and contains small
amounts of limestone, dolomite, siltstone and sandstone (Martin, 1957). The bedrock
which subcrops under Schoolcraft consists of a limestone facies of the Coldwater Shale.

The surface of the limestone slopes gently towards the northwest (Mayotte, 1991 ).

..28 c. 8.88... m. 8%. NS. 5.92.5. 59. 858.2
.xomfi :. 85.50 m. 2560 83.5.3. €3.55. .0 385:9". 826.. on. .0 one xoofiom .FN 659".

9.9.0 3.3m I

9.90 9.6.2 wmmm I
games". 28.2 c6990 I
5.89:0“. ocmE m.om I
c.0065 85.8.2 I
ocofiocmw m.cm>.>m I
9.2.0 82m .650 I
ocofioE... mousse U
6.2m :3. I

9.90 6299... I

aficw E...E< I

295 ..toBflm I

6.95 2058 I

...w 2058 new mm mobm I
99% 23:5 I

2.2m 5.96.00 I
8.885“. ..m52m2 I
5:950“. 595.... a
6:288... togmm I
cozmctou 393mm I
coszBH. 52$ 9.96 !
moon com. I

232.com $25.. .0 .6280 sootuom

   

Glacial History of Southwestern Michigan

With the exception of recent alluvial deposits, the sediments that overlie bedrock
in Kalamazoo County were deposited in the last glacial episode, the late Wisconsinan, as
the Laurentide Ice Sheet retreated. Researchers believe that both the Saginaw and
Michigan Lobes of the Laurentide Ice Sheet contributed to surfical deposits in
Kalamazoo County (Monahan and Larson, 1982; Monahan et al., 1986; Monahan and
Larson, 1986; F inkbeiner, 1994). Finkbeiner (1994) stated that the county is located
within a reentrant between the Lake Michigan Lobe and the Saginaw Lobe. Figure 2.2
shows the location of moraines in southwest Michigan and the extent of both the Saginaw
and Lake Michigan Lobes.

The prominent features of the Kalamazoo Moraine are the main reason that so
many authors believe a significant retreat and readvance occurred before its formation
(F inkbeiner, 1994). The presence of lacustrine sediments between the Ganges and
Saugatuck Tills implies that a significant retreat of the Lake Michigan Lobe occurred
between the formation of the Tekonsha Moraine and the Sturgis-Kalamazoo Morainic
System (Monaghan et al., 1986). Monaghan et al. (1986) showed contrasting mineralogy
of clays within different till units and concluded that the Saugatuck, Ganges and Glenn
Shores tills are unlikely to represent till facies of the same advance. An east-west cross-
section through southwestern Michigan summarizes the correlations of Moraine and
underlying till unit as proposed by Monaghan et al. (1986) (Figure 2.3).

Outwash deposited as meltwater from the glacier compose approximately two
thirds of the surfical deposits in Kalamazoo County. Studies by Straw (1991) and

Steinman (1994) presented descriptions of the outwash fans in Kalamazoo County. The

 

 

 

\‘l \\ \‘8M
1

4//l\-.\ \ChM

 

 

 

 

 

 

 

 

 

I I ‘
I -.
\ \
\
, ; +41 — County border
‘3‘ ,’ I ./ — Moraines
a g“ $ ,1 I _/ —--- Intenobate boundary
0 ‘ I / 4 0
¢/ / ' I b N g 1' 210
\+ +9 / o m' 15
as a!“ “7?"

Lake Michigan Lobe Moraines Saginaw Lobe Moraines
LBM- Lake Boarder Moraine LaM- Lansing Moraine
lVM- Inner Valpraiso Moraine ChM- Charlotte Moraine
OVM- Outer Valpraiso Moraine KM- Kalamazoo Moraine
IKM- Inner Kalamazoo Moraine TkM- Tekonska Moraine

OKM- Outer Kalamazoo Moraine
StM- Sturgis Moraine
TkM- Tekonsha Moraine

 

 

 

Figure 2.2. Location of Moraines in southwestern Michigan. Notice the interlobate
boundary marking the extent of both the Saginaw and Lake Michigan lobes. Kalamazoo
County is marked by a bold outline. Modified from Monaghan et al., 1986.

10

 

 

 

Figure 2.3. Till sheets underlying moraines in southwestern Michigan. Modified from
Monaghan et al., 1986.

11

city of Schoolcraft lies within the Prairie Rhonde Fan, described by Steinman (1994).
The Prairie Rhone Fan was deposited when meltwater and outwash broke through a
narrow breach in the Kalamazoo Moraine near Paw Paw Lake and deposited the material
from braided streams throughout Prairie Rhonde and Schoolcrafi Townships Steinman

(1994) (Figure 2.4).

General Stratigraphy and Sedimentology of Glaciofluvial Deposits

Streams that are fed from glacial meltwater carry enormous quantities of detrital
sediment, most of which is deposited in broad alluvial plains called sandar or sandurs
(Icelandic: “sand plains”, singular sandur) (Smith, 1985). Two different kinds of sandur
were recognized by Kingstrom (1962) — valley sandar and plain sandur. Valley sandur
occur within well defined valley walls, are usually formed by one main channel and are
commonly associated with individual cirque or valley glaciers. In North America, valley
sandur are called valley trains. Plain sandar are broad and unconfined when compared to
valley sandar; they are formed by the coalescence of many braided streams that deposit
sediment to form extensive alluvial plains. They are usually associated with large ice
sheets and are common features of the Pleistocene landscape of North America (Smith,
1985). Plain sandur are commonly called outwash plains.

Krigstrom (1962) recognized three rather distinct zones of sandar related to
distance from the ice margin, 1) proximal, 2) intermediate and 3) distal. Meltwater flow
in the proximal zone is confined to only a few main channels that are relatively deep and

narrow. Incision of proximal channels into alluvium is common and is likely a result of

slope adjustment to a normal hydraulic and sediment-load regimen. The intermediate

12

 

odi"""'—.
c‘.’ .....
's
g N
e
‘. U
\ ‘\
\ e
\y ‘.
. \ ‘.
o u
I \ ‘\
D \ \
n
‘ e
: \ \ ,
| ‘s .‘
.0 “Q\
! ‘5 \
I ‘~
3
’\

\ ~.‘\ \
, \ “x‘ \' \
a: \ s‘~~. \ \
\“ \ ‘4“ I \ i
‘3‘ \ ‘x I ' ‘
l 11}, J 1
\ ‘.
\ Schoolcraft ‘y

 

 
     

 
 

Kalamazoo County E“
St. Joesph County i

 

 

 

 

Stream/river/wetland/lake
------ Boundary of Baresse's study area
----------------- Boundary of Steimann’s study area

 

 

 

Figure 2.4. Study areas of Steinmann (1994) and Baresse (1991). Boundary of Prarie
Rhonde Fan as described by Steinmann, maked by Steinmann’s study area. Modified
from Baresse, 1991 and Steinmann, 1994.

13

zone is characterized by networks of wide, shallow distinct channels that shift frequently.
Abandoned channels are prominent during normal discharge. In the distal zone, channels
become shallower and poorly defined. At periods of high discharge, flow may merge
into a single sheet.

The streams responsible for sandur are typically braided (Benn and Evans 1998).
Braided streams are characterized by high width to depth ratios, steep slopes, abundant
bedload, cohesionless banks, fluctuating discharge and generally low sinuosities (Benn
and Evans, 1998; Miall, 1977). At least nine variables interact to determine the nature of
the resulting braided stream channel: 1) discharge amount, 2) discharge variability, 3)
sediment load, 4) grain size of sediment load, 5) width, 6) depth, 7) velocity, 8) slope and
9) bed roughness (Leopold and Wohnan 1957). Braiding is developed by sorting action
as a stream leaves behind particle sizes that it is unable to transport. Deposition of the
coarser bed load causes mid-channel bars to form (Boggs, 1987). Repeated bar formation
and channel branching generate a braided network of bars and channels over the stream
bed (Leopold and Wohnan 1957).

There are three main bar types found in braided stream systems 1) longitudinal,
comprising crudely bedded gravel sheets, 2) transverse to linguoid, consisting of sand or
gravel and formed by downstream avalanche-face progradation and 3) point or side bars,
formed by bedform coalescence and chute and swale development in areas of low energy
(Miall, 1977). Longitudinal bars are mid-channel bars that form when the coarsest part of
the stream load is deposited as stream flow wanes (Boggs, 1987). They are diamond
shaped in plan view with the long axis in the parallel to flow direction. They are bounded

on either side by active channels and may therefore have partially eroded margins. Bars

14

formed in gravel are most common, although some sand bars show similar morphology
(Miall, 1977). The internal structure in longitudinal bars is weak horizontal bedding
commonly with imbrication (Miall, 1996). Common in sandy streams transverse and
linguiod bars are also called 2-D and 3-D dunes, respectively. 2-D dunes are formed
under lower flow velocities than 3-D dunes for any grain size (Ashley, 1990). The
internal structure in 2-D dunes is characterized by planar tabular cross-bedding while the
internal structure of 3-D dunes is characterized by trough cross-bedding (Ashley, 1990).
Point and side bars are genetically the same as lateral bars, they form in areas of low
energy, such as the inside of a meander (Miall, 1977). Although point bars are usually
thought to occur in meandering streams, they may also occur in braided streams (Miall,
1977). The internal structure in lateral, point and side bars is complex, and may include
planar-tabular cross-bedding, trough cross-bedding, ripple cross-lamination, coarse-
grained lag deposits, and fine-grained drape and fill deposits (Miall, 1977). Figure 2.5
shows the location and geometries of bars within braided streams.

Miall (1977) proposed four vertical models that can develop under varying
conditions of bedload and discharge, these are 1) Scott-type, 2) Donjeck-type, 3) Platte
type and 4) Bijou Creek-type (Figure 2.6 and 2.7). The Scott-type is characteristic of
proximal, gravelly rivers and consists of mainly longitudinal bars gravels with sand
lenses formed by infill of channels and scours during low water (Miall, 1977). The
Donj eck-type, characteristic of medial streams, may be dominated by either sand or
gravel, and consists of fining upward cycles of variable scale. Longitudinal-bar deposits,
linguoid-bar deposits, channel-floor dune deposits, bar-top and overbank deposits all may

be important (Miall, 1977). The Platte-type model, representative of distal streams, is

15

 

   

 

0 50M 0 500M o 1000"
i | L_.___..J L.___l
Bartypes

Lo longitudinal bars
D longitudinal with diaginal flow
R eroded bar remnant

Li linguoid

M modified linguiod
P point

S side

 

 

 

Figure 2.5. Principal bar types and where they form in a stream. Modified from Miall,
1977.

Scott-type Donjeck-type Platte-type Bijou Creek-type

 

 

 

 

 

 

. .. :43?»
. f, raga
' fir "-
' l 0 c.
9 I'- was!
raw,
1’?! 2'1 ‘ r "3)
I

 

 

Figure 2.6. Vertical facies models of Scott-type, Donjeck—type, Platte-type and Bijou
Creek-type. Modified from Miall, 1977; 1996.

 

Figure 2.7. Block diagrams showing the braided stream patterns associated with Scott-
type streams (A), Donjeck-type streams (B), and Platte type Streams (C). Modified from
Miall, 1985.

18

characterized by linguoid and transverse bars that generate largely cross-bedded, sandy
deposits (Miall, 1977). There is no well developed cyclicity, although some fining
upward sequences may be identified (Boggs, 1987). Bijou Creek-type deposits consist of
horizontally laminated sand and a lesser amount of sand showing planar cross-bedding
and ripple cross-lamination (Miall, 197 7). Deposits are formed during flash flood, and
each flood event is represented by a fining upward sequence (Miall, 1977).

There is no absolute indicator that can be used to estimate proximity to source in
braided stream environments (Miall, 1977). However, grain size is one of the best
parameters that shows downstream variation. Thus, grain size is useful in measuring
proximity (Miall, 1977). Mean and maximum grain size tend to decrease down stream at
a rate that varies depending on stream power (Miall, 1977). Smith (1985) stated that
grain size tends to decrease down-sandar in an approximately exponential way, but, the
rate of diminution varies significantly between sandar. Smith (1985) suggested that this
is probably a reflection of different aggradation rates. The coarsest gravel in the system
is moved infrequently and only for short distances, during high energy flows, whereas
finer grained material can be transported progressively farther during any time or flow
event. If a stream is degrading, little or no downstream fining occurs because sediments
that enter the system must leave it (Smith, 1985). If a stream is aggrading, the coarse
sediments will be buried before they are transported far. Thus the faster they are buried
(i.e. the higher the rate of aggradation), the shorter their final distance can be, resulting in

a downstream-fining gradient proportional to the aggradation rate (Norman, 1985).

19

Hydrogeology
Regional Hydrogeology

Groundwater studies have been conducted within the Kalamazoo outwash plain
by Barrese (1991), Steinmann (1994), with subsequent results presented by Kehew et al.
(1996), Hyndman et al. (2000) and Lipinski (2002). Barrese (1991) divided the
Schoolcraft Aquifer, into two flow systems based on geochemical results. A northern
flow system located north of the Spring Creek-Government Marsh and a southern flow
system located south of the Spring Creek-Government Marsh (figure 2.4). Barrese
(1991) defined the Schoolcraft aquifer as ranging in saturated thickness fi'om 230 feet to
200 feet, the base of the aquifer is defined as the top of the Coldwater Shale. He stated
that the aquifer is composed of sand, gravel and diarnicton. Three aquifer stress tests by
pumping conducted near the village of Schoolcraft yielded calculated hydraulic
conductivity (K) values ranging fi'om 3.6 x10'2 cm/s to 5.2 x 10'2 cm/s (Barrese, 1991).
Maps of groundwater elevation near the village of Schoolcraft show groundwater flowing
generally southeast.

Steinmann (1994) used newly installed monitoring wells, county well logs,
gamma ray logs, head measurements, flow nets, slug tests, and aquifer stress tests by
pumping to characterize the geologic and hydrologic conditions of the aquifer below
Schoolcraft. Steinmann’s study area was similar to Barrese’s (1991) (figure 2.4).
Steinmann (1994) used tritium data and flow nets to support the conclusion of two
separate ground water flow systems, where horizontal flow dominates the region with the
exception of limited recharge and discharge areas. K values were calculated from three

aquifer stress tests by pumping conducted near the village of Schoolcraft. Calculated K

20

values are 3.8 x 10'2 cm/s, 3.9 x 10'2 curls and 5.0 x 10'2 cnr/s based on estimated
saturated thicknesses of 152.5 feet, 168 feet, and 130 feet respectively. Steinmann (1994)
stated that the regional gradient ranges from 0.0022 to 0.0012, depending on location
within the Prairie Rhonde fan. Groundwater flow direction is down fan or southeast.
Kehew et al. (1996) published much of Barrese (1991) and Steinmann (1994) results,
focusing on the area within the Prairie Rhonde fan.

Hyndman et al. (2000) studied a small area within the Schoolcraft Aquifer where
they developed a bioremediation system that degrades carbon tetrachloride, discussed in
more depth below. Small scale groundwater and transport models were developed to
predict the migration of tracer. This aquifer is composed of 27 meters of glaciofluvial
sediment overlying a regional clay layer with a water table at approximately 4.5 m below
ground surface (bgs). 220 core segments fiom 7 wells were analyzed for K. The average
K value was approximately 2.7 x 10'2 cm/s and a log (K) variance of 0.12, highest K
values are at the base of the aquifer. The gradient is 0.0011 and flow is to the southeast.

Lipinski (2002) built a regional scale model with boundaries based primarily on
surface water bodies in the vicinity of Schoolcraft, Michigan (figure 2.8). In this model,
Lipinski (2002) used regional geological studies to build a three layer model 60 to 90 m
thick. Two layers representing outwash were separated by a third layer that represented a
discontinuous till. K values used in this model were taken from previous studies.
Lipinski (2002) also constructed a smaller scale groundwater model and reactive
transport simulation around plume G based on head results from the regional model
(figure 2.8). This model was approximately 27 meters thick and represented the upper

unit of outwash. A clay layer at the base of the aquifer marks the bottom of the model.

21

 

 

 

‘J

 

 

 

 

 

‘I—QJ

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. t/

Schoolcraft '! ;

 

 

 

 

 

i /\d
C
l

 

 

U}: "

 

/\/ Roads N Rivers/Streams Moraine

, "x I Drain % Wetland lllllllllllllllll Lake
0 4
5.25555
km

 

 

 

Figure 2.8. Study area of Lipinski (2002) regional model. Lipinski’s smaller scale
transport model is outlined by the bold rectangle near Schoolcraft. Modified from
Lipinski, 2002.

22

The gradient of the water table is approximately 0.0012 near Schoolcraft and flow is to

the southeast.

History of Contamination at Plume A

A history of the contamination source at plume A, located in Schoolcraft,
Michigan, is given by Mayotte (1991) and is briefly described here. In 1986, significant
concentrations of carbon tetrachloride and trichloroethene were detected in ground water
quality samples obtained from residential wells just outside the village of Schoolcraft.
Water quality data reveled that the contamination most likely originated at the
Countyka Incorporated facility (Mayotte, 1991). The data also indicate that
chloroform and 1,1,1-trichloroethane are minor constituents of the pltune. At the date of
Mayotte’s report, the plume was known to extend 1128 m down gradient of the
Countrymark property and was present throughout most of the saturated thickness of the
aquifer.

Mayotte (1991) reported that evidence obtained by the Michigan Department of
Natural Resources indicates that carbon tetrachloride was used as a fire retardant, as a
transport agent for carbon disulfide, and as an insecticide applied to grain silos on the
Countrymark property. It is believed that carbon tetrachloride and other solvents were
stored in bulk in above ground storage tanks located near the silos. Throughout the
history of the facility, contaminants leaked from the silos and/or storage tanks and

infiltrated the vadose zone and the ground water (Mayotte, 1991).

23

Bioremediation of Plume A

Bioremediation studies at Plume A focused on the design and implementation of a
delivery well system for bioaugrnentation (Hyndman et al., 2000; Dybas et al., 2002).
Bioaugrnentation is a process that involves the addition of microbes and the necessary
nutrients to sustain them in an aquifer. The researchers designed a system where water
was allowed to flow passively through a treatment zone, or biocurtain, where injected
bacteria attach to the sediments degrade carbon tetrachloride. A row of delivery wells
spaced one-meter apart and screened across the vertical extent of contamination were
installed. Ground water is periodically extracted from alternating delivery wells,
amended with the chemicals needed to support bioremediation, and then the amended
water is injected into adjacent wells. An intermittent feeding schedule maintains
adequate biomass to efficiently remove contaminants without significant reduction to
aquifer conductivity. This biocurtain has proven to be 97% effective (Hyndman, et al.,
2000).

To better understand the heterogeneity at the Schoolcraft plume A site, core from
11 wells were collected and either sieved, measured for K value on a constant head
perrnearneter, described for sedimentologic features, or a combination of these. A firll
scale field tracer test was conducted in November 1997, using Flourescein and Bromide
as relatively conservative tracers to verify to the researchers that their selected pumping
strategy would provide roughly the expected extraction well concentrations and to
evaluate the fate of fluid pumped through the delivery well gallery (Hyndman et al.,

2000).

24

Chapter 3
Integration of Sedimentologic and Hydrologic Properties for Improved Transport
Simulations: Detailed Characterization of a Glacial Outwash Aquifer at the

Schoolcraft Bioremediation Site, Michigan

(Note: This chapter forms a manuscript that has been submitted to SEPM for publication
and was authored by BI T EMAN, S.E., H YNDMAN, D. W., PHANIKUMAR, MS, and
WEISSMAMV, G.S., Dept. of Geological Sciences, Michigan State University, East

Lansing, MI, 48824)

Abstract: Traditional geostatistical approaches for estimating hydraulic conductivity
values fail to reflect sharp contrasts that occur at boundaries between different
stratigraphic units, thus limiting the accuracy of contaminant transport models. We
present a novel approach to incorporate geostatistical simulation into a stratigraphic
framework to better represent aquifer heterogeneity. The approach was developed and
tested at the Schoolcraft Bioremediation Site in Southwestern Michigan, where detailed
aquifer property estimates were needed to accurately simulate multicomponent reactive
transport and to design an effective bioremediation strategy. The sediments at the site
were deposited as outwash medial and distal to the Kalamazoo Moraine, and consist of
primarily fine to medium sands with interbedded gravels and silts. A series of l8-meter
long continuous cores were collected in the vicinity of the 210 square meter biocurtain
injection zone. These cores were assessed for sedimentologic character, grain size
distribution, porosity, and hydraulic conductivity values. Sedirnentologic character and

correlation lengths from outcrop analogs in the same outwash unit supplemented the

25

site’s core data. Based on the core data, the aquifer was separated into major
sedimentologic units to preserve abrupt natural changes in conductivity, and the
conductivity values were then geostatistically interpolated within each stratigraphic unit
to better preserve stationarity. The stratigraphically-based conductivity and porosity
estimates were used as inputs to a high-resolution three-dimensional flow and transport
model of the region. The model with stratigraphic interpolation provided improved

transport predictions for an injected tracer pulse.

Introduction

In recent years, the need to more accurately predict contaminant migration has
focused efforts to develop novel aquifer characterization approaches, such as those that
incorporate geological information. In the Northeastern and upper Midwestern United
States, many aquifers for domestic and agricultural uses are composed of unconsolidated
Quaternary glacial sediments. These aquifers have a high risk of point and non-point
source contamination due to their generally unconfined nature. Glacial deposits typically
have moderate to high levels of heterogeneity, which often complicates transport
predictions within these environments. Contaminants follow subsurface preferential
flow paths, such as paleo-river channels, flowing faster and in slightly different directions
than the regional groundwater flow. Without an understanding of the depositional setting
and the geometries of heterogeneities, an accurate characterization of the aquifer is
unlikely.

Since the distribution of sedimentologic heterogeneities is directly related to the
distribution of hydraulic conductivity (K) and other aquifer properties, stratigraphic

complexities should be incorporated into heterogeneity models. This concept has lead to

26

sedimentologic research for estimating the heterogeneity of hydrologic properties (e.g.,
Webb and Anderson, 1996; Carle et al., 1997; Davis et al., 1997). Traditional
geostatistical methods that interpolate between hydrologic property estimates without
considering stratigraphy commonly misrepresent a site’s geology by not honoring the
stationarity assumption and smoothing the data rather than preserving abrupt conductivity
changes (Weissmann and Fogg, 1999).

The need to improve simulation of fluid and solute migration has led to
characterization approaches in aquifers and petroleum reservoirs that incorporate more
geologic information. Eschard et al. (1998) and Weissmann and Fogg (1999) used the
geology and stratigraphy as a framework for geostatistical realizations in order to better
preserve stationarity. While outcrop analogs have been used to interpret facies
architecture (Miall, 1985), there is a renewed interest in such subsurface characterization
tools to estimate aquifer hydrologic properties (Davis et al., 1997; Anderson et al., 1999;
Bersezio et al., 1999; Homung and Aiger, 1999; Klingbeil et al., 1999). Facies
distributions can be measured on outcrop analogs and can be applied to anisotropic
variograms, as has been applied at a small scale by Davis et al. (1997) to interpret
bounding surfaces.

In this paper, we present a method to better characterize aquifer heterogeneity
through separate geostatistical interpolation within zones of similar sedimentologic
character. Within each zone, geologic information distinct to that zone was used to
estimate the lateral correlation lengths. Each hydrostratigraphic zone was then kriged
separately to estimate that zone’s K field. The different zonal conductivity estimates

were then merged to create a single K field. By dividing the data into geologically based

27

zones of similar sedimentologic character, where the mean and standard deviation of K
are distinct from other zones, stationarity is much more likely to be honored.
Additionally, K values are estimated (kriged) separately by zone, thereby preserving the
natural abrupt changes in K across bounding surfaces. Core from the site yielded
stratigraphic and sedimentologic data that was used to identify zones. In addition, core
material was used to measure horizontal and vertical K values using constant head
permeameters, with dense vertical information but relatively sparse horizontal
information. To supplement the common lack of information to accurately estimate the
horizontal correlation parameters, measurements from outcrop analogs were used to
evaluate K distributions, facies distributions, and width to thickness ratios. Our results
show that the transport simulations based on the stratigraphically-based model more

closely match measured tracer concentrations at the site.

Site Description

The Schoolcraft site is composed of glaciofluvial sands and gravels that were
deposited as water drained away from a retreating ice margin. The ice margin later
stagnated long enough for sediments to accumulate and form the Kalamazoo Moraine.
The moraine, located west of Kalamazoo, was deposited in the last glacial episode (late
Wisconsinan), and trends northeast-southwest (Figure 1). The glaciofluvial sediments are
coarsest near the glacier where braided rivers had the highest energy as they flowed
southeastward away from the ice margin. The energy of the rivers decreased with
distance from the ice margin. Thus the further the glacier retreated, the finer the

sediments that were deposited at any location. Subsurface sediments at the Schoolcraft

28

Kalamazoo County Surfical Geology
1:] Kalamazoo River

Climax-Scott

 

      

Kalamazoo
Moraine

  
   
 

. Schoolcraft

 

Figure 3.1. Kalamazoo County Surficial Geology, location of Kalamazoo Moraine and
outcrop site used for this study. Modified from Monaghan and Larson, (1982).

29

site are composed of gravels at the base of the aquifer, fining upward to medium and fine
sands with some interbedded very fine silty sands. This is the typical fining outward and
upward succession present in glacial outwash systems, as described by Miall (1977).

The Schoolcraft Plume A study site is located within an unconfined aquifer
composed of a 27.5 m thick sequence of glaciofluvial sediments that lie directly over a
regional clay-rich till that acts as an aquitard in Schoolcraft, Michigan. Plume A is the
location of a carbon tetrachloride (CT) bioaugrnentation experiment, where microbes and
substrate were injected to degrade aqueous and sorbed phase contaminants (Figure 2)
(Hyndman et al., 2000; Dybas et al., 2002). The contaminant plume is about 1.2 km
long and 90 m wide, extending from roughly 8 to 26.5 m below ground surface (bgs),
with CT concentrations from 5 to 150 parts per billion (ppb). The water table at this site
lies at roughly 4.5 m bgs. The Schoolcraft site was chosen for this study because a large
existing database is available that includes core descriptions, K measurements from many
wells, and many measured concentration histories collected during tracer tests.

The bioaugrnentation system at this site was designed as a series of closely spaced
recirculation wells for delivery of microbes and nutrients. The spacing and flow rates
were chosen by minimizing the total design and operation cost using a flow and transport
model coupled to an optimization routine (Hyndman, et al. 2000). A portion of the
contaminant plume flows passively into the treatment zone or “biocurtain,” where
microbes degrade the CT. Groundwater is periodically extracted from alternating
delivery wells, amended with the chemicals needed to support bioremediation, and then
injected into adjacent wells. An intermittent feeding schedule maintains adequate

biomass to efficiently remove contaminants without significantly reducing the aquifer

30

Location of biocurtain

the q“.

WL
'“Vfi'IWI'

F‘Jrr —. L-w’te___-_
m
Mt‘r ' ,
nil

    
      
 

l" l

:-
,fi-

0 1 000

r l r
SCALE IN FEET
1" =1000'

Figure 3.2. Schoolcraft Plume A site. Location of biocurtian is marked by arrow.
Modified from Mayotte (1991).

31

conductivity. This biocurtain has proven to be greater than 97 % effective at removing
aqueous phase contaminants (Hyndman, et al., 2000), and effective at removing sorbed
phase CT in the biocurtain region over greater than 4 years of operation (Dybas et al.,
2002)
Core Descriptions
Methods

We evaluated core from seven of the 15 delivery wells in the biocurtain (D2, D4,
D6, D8, D10, D12 and D14), spaced 2m apart in a transect perpendicular to flow, and
from wells P6, P7 P8, and P18, located approximately 2-3 m downgradient of the
biocurtain (see Figure 3). These cores were collected in 1.5 m sections using the
Waterloo continuous core sampler that advances a core barrel ahead of an auger string. A
vacuum sealed core liner minimized sediment loss. Sediment from each core except well
P18 was sieved for grain size distribution, repacked to its original volume, flushed with
carbon dioxide, and tested for K values on a constant head perrneameter. For this study
we assumed that these repacked K values were representative of local horizontal K. Core
segments from P18 were cut into 15.2 cm sections, flushed with carbon dioxide, saturated
with water from below over a 4-8 hour period, and then tested for vertical hydraulic
conductivity values using a constant head permeameter. Each core segment was weighed
while saturated, photographed and described for sedimentologic characteristics, dried and
reweighed to estimate porosity, then sieved for the grain size distribution. Additionally,
core from wells P6, P7, P8, and P18 were described for geologic properties including
grain size, sorting and stratigraphic bedding character and grain size trends.

Vertical K values were measured from P18. To obtain horizontal K, vertical K to

32

 

ground water
“Erection
N ms .
' 014
D13 I MW11
, e
012 .
I .
. 011
' o1o '
[)9 I MW13 . P18
' . e
08 0 P8 ' .
I .
, D7 MW1o.
. - 0
06 0 P7
D5 I MW12 . p5
. e
D4 ' .
. 03 I .
O o 0
D2
01 I
. ' “1m IKdata
O K and strat data
e monitoring well
0
ootherwells
-:|
0m 2m

 

 

 

Figure 3.3. Location of delivery wells (D) and monitoring wells (MW) at the Plume A
Schoolcraft site.

33

horizontal K anisotropy factors were applied to the vertical K values. The average K
anisotropy factor of each zone was calculated as the ratio of the average vertical K values
from undisturbed vertical core samples (well P18) to average horizontal K (assumed
equivalent to the repacked K from D2 to D14 samples). The values were 0.72, 0.81, 0.73,

0.75, for zones 1 to 4 respectively, the average across all the zones was 0.74 (Table 3.1).

Corefacies interpretation

Six facies were interpreted from the core: 1) massively bedded to finely laminated
very fine sand, 2) massively bedded medium to coarse sand, 3) massively bedded fine
sand, 4) cross-stratified fine sand, 5) cross-stratified medium to coarse sand and 6) cross-
stratified gravel.

The massively bedded to finely laminated very fine sand facies is moderately well
sorted with a high abundance (16%) of silt and clay. Grain sizes for this facies ranged
from medium sand to clay. Vertical K was only measured on one sample from this facies
due to lack of available material, with a measured value of 1.45 x10"3 cm/s (Table 3.2).

The massively bedded medium to coarse sand facies is well, moderately-well or
poorly sorted. Vertical K values range from 1.14x10'2 to 1.82x10'2 crn/s in the poorly
sorted samples. Vertical K was only measured on one well sorted sample and one
moderately to well sorted sample; the values were 1.38 x10'2 cm/s and 2.40x10'2 cm/s

The massively bedded fine sand facies is well to moderately-well sorted, with
coarse sand to silt grain sizes. Vertical K values in the well sorted sand range fi'om 8.62
x10’3 to 1.61x10'2 cm/s, while vertical K values in the moderately to well sorted fine sand

respectively (Table 3.2). Grain size for this facies ranged from pebbles to very fine

34

vertical to

 

 

 

 

 

 

horizontal
zone four 1.25
zone three 1.27
zone two 1.19
zone one 1.28
average 1.26

 

 

 

Table 3.1. Anisotropy values for vertical K and horizontal K, determined by comparing

repacked to non-repacked core.

O>Tl

I11

Table 3.2. Vertical K (cm/s) of facies vs. sorting from well P18.

 

 

 

 

 

 

 

SORTI NG
well "Highly moderately poorly very poorly
low 1.14E-02
lMassl I"? 'y everege 1385-02 2405-02 1575-02
medium and high 1.825-02
coarse sand no. or maples 1 1 3
low 8.62503 4.35503
363::ny average 1.23502 1.01502
fine sand high 1.61502 1.62502
no. of samples 5 1 1
low
Exam average 1455-03
very fine sand Nah
no. 01 san'oles 1
low 2.625-02
Cross stratified avenge 1.785-02 3.85E-02 3.50503
medium and high 5.085-02
coarse sand no. or We: 1 2 1
low 2.84503 3.28503 1.50503
Cross stratified average 6.67503 4.46503 7.20503 3.62503
fine sand nigh 1.25502 5.00503 5.36503
no. 0! samples 7 3 1 4
low
Cross stratified average 2.40E-02
gravel high
no. 01 sanples 1

 

 

 

 

 

 

 

35

 

sands. range from 4.35x10'3 to 1.62x10'2 cm/s. Generally, the better sorted the massively
bedded fine sand, the greater the K value (Table 3.2).

The cross—stratifiedfine sand facies is well to poorly sorted, and vertical K values
generally increased with better sorting (Table 3.2). Grain sizes for this facies ranged
from pebbles to silt. Vertical K values of the well sorted fine sand ranged from 2.84x10'3
to 1.25x10’2 cm/s. Vertical K values of the moderately well sorted fine sand ranged from
3.28x10'3 to 5.00x10'3 cm/s. The moderately sorted sand had a vertical K value of
7.20x10'3 cm/s and the poorly sorted fine sand had a vertical K value of 3.50x10 ’3 cm/s.

Cross-stratified medium to coarse sand facies is moderately, poorly or very
poorly sorted. Grain size for this facies ranged from pebbles to very fine sand.
Moderately sorted medium to coarse sand had a vertical K value of 1.78x10'2 cm/s.
Cross-stratified medium to coarse sand with poor sorting had vertical K values of
2.62x10’2 and 5.08x10'2 crn/s. The very poorly sorted medium to coarse sand had a
vertical K value of 3.50x10'3 m/s (Table 3.2).

The cross-stratified gravel facies is very poorly sorted with a vertical K value of
2.40x 10'2 cm/s (Table 3.2). Grain size for this facies ranged from pebbles to very fine
sand.

The massively bedded to finely laminated veryfine sand facies was interpreted as
abandoned channel fill, commonly found in distal outwash facies (Donj eck and Platte
type, as described by Miall, 1977). All other massively bedded sand facies were
interpreted as having been deposited during high discharge when flow becomes less
channelized and forms a large sheet. It is also possible that the stratification is too subtle

to be identified at the core scale, or core collection disturbed the subtle cross-

36

stratification. All cross-stratified facies were interpreted as channel bar deposits in a
braided stream environment. In all facies, a larger grain sizes indicate either greater

stream energy and/or a more proximal location of the glacier.

Schoolcraft Stratigraphic Evaluation

Three stratigraphic breaks were identified in the cores, which provided a
framework of four zones for K interpolation (Figure 3.4). Stratigraphic breaks were
identified in cores as one or more of the following: 1) an erosional surface, 2) abrupt
changes in mean grain size and 3) abrupt changes in sedimentary structures.
Additionally, the same four zones were apparent in the measured K distribution: zone 1 —
a high K zone at the base of the section (Kmmge= 5.44 x 10'2 cm/s), zone 2 - a small low
K unit (14.,ng 1.56 x 10'2 cm/s), zone 3 — a medium high K unit (lemme: 3.2 x 10'2
cm/s) and zone 4 — a medium low K unit (Kevmge: 1.38 x 10'2 cm/s) at the top of the
section (Figure 5). Table 3.3 lists the statistics of K values for each zone.

Zone 1 (21-27.4 m bgs) is located at the base of the aquifer and is generally
composed of the cross-stratified gravel, cross-stratified coarse sand and cross-stratified
medium to coarse sand facies. Zone 2 (21-20 m bgs) is composed of massively bedded
fine sand facies and finely laminated very fine sand facies. The very fine sand facies
comprises a low K area that was identified within this zone in three delivery wells (D8,
D10 and D12) and in three down-gradient wells (PS, P7 and P18). Zone 3 (16-20 m bgs)
is composed of massively bedded fine sand facies and the massively bedded medium to
coarse sand facies. Zone 4 (16-8 m bgs) consists of the cross-stratifiedfine sand facies

with cross-stratified medium to coarse sand facies at the erosional base of the zone. Pilot

37

Grain Sizel K cmlsec Grain Size K cmlsec

 

 

 

 

 

 

 

 

 

3 NN 13 30.10101
5 3 Ehaaggi a 5 Elwfi'fi
macaw-new: (DC/JQN'v-‘éo'v'
ii . S
14‘13 ‘ :3 3: 0
ml
Sm o
15"”

 

 

 

 

 

 

   

 

 

- '5
° 0 g
N
17- , °
g o Sm °
:4 22 - ° last core
‘5.
a , e
D - a /
184 “

 

 

 

 

 

 

 

Figure 3.4. Core description of P1 8 showing zonal boundaries (marked by arrows) and
measured vertical K.

38

 

A 8— - 4-
DZ D4 D6 D10 D12 D14 1

10*

12*

a U

3. m

5;; a
“51'

3 G33?) m

E” E5952).
3 ”Or-3:1?) O“

16

 

 

18*

 

 

 

.lir @ J

20

C
-=O
J

 

9

22 * 21..

.7-
1n-

‘5'

24*

 

 

 

21 23 25 27 29 31 33 35

1o~
12»
14-

18

*1

a

18* 11%
11

 

P7

 

 

 

 

 

20

 

‘
Edam ' 3 NW!- ” 3
5.

22- 8

24*

26*

 

L L 1

28 1 1 1
41 41 .5 42 42.5 43 43.5 44

 

Figure 3.5a and 3.5b. Cross-section of log K data points used for interpolation. Figure
3.53 is a transect through the delivery wells, Figure 3.5b is a transect through the down
gradient wells (Figure 3.3 for location). Zonal boundaries picked at 16 m bgs, 20 m bgs
and 21 m bgs, identifying 4 zones. Scalebar is in log K (cm/s).

39

mean

median

standard deviation
no. of data points

Table 3.3. Horizontal K (cm/s) statistics for the the four stratigraphic zones and the non-

zonal model.

 

 

 

 

 

zone 1 zone 2 zone 3 zone 4 all zones
5.44 E-02 1 .56E-02 3.20E-02 1 .38E-02 3.07E-02
5285-02 1.74E-02 3.125—02 1.27E-02 2.59E-02
2905-02 1 .1 1 E-02 9.40E-03 5.40E-03 2.43E-02
106 25 82 133 346

 

 

 

 

 

4O

 

studies indicated that sediments above 8 m b gs did not have CT contamination, and were,

therefore, not included in this study.

Outcrop Analog

Cores at this site provided an excellent sampling of vertical distributions for K
analysis, but lateral data was limited even with the closely spaced cores from this site. To
supplement the limited lateral core data, outcrop analogs were used to evaluate lateral
variability of hydraulic conductivity at the site and to aid in the interpretation of bounding
surface geometries. The excavation of sand and gravel pits in the Kalamazoo moraine
outwash deposits exposed unconsolidated outcrops that could be analyzed for facies
distribution, channel size, bed thickness, and lateral bed variability.

Outcrop analogs to the Schoolcraft Plume A site (Figure 3.1) were selected based
on: 1) location within the same outwash complex, 2) comparable distance from the
Kalamazoo Moraine, 3) grain size, 4) bed thickness and 5) sedimentary structures. Using
these criteria, the analog sites and the Schoolcraft site appear to have been deposited
under a similar energy conditions in the same type of environment. Once the analog sites
were chosen, beds and lithofacies were identified, measured and then drawn and labeled
on photo mosaics (Appendix B). Samples for horizontal K and grain size distributions
were collected by inserting 15.2 cm long brass sleeves in lateral and vertical rows within
chosen facies of the outcrop. These samples were tested for hydraulic and
sedimentologic properties using similar methods as core from well P18. Width to
thickness ratios of facies units for this study were estimated using the sand and gravel pit

measurements that compared most closely to the grain size, bed thickness and bedding

41

structures of the core (Figure 3.6). The sand and gravel pit that was most similar to
sediments at the Schoolcraft site is composed of cross-stratified gravel facies, cross-
stratified medium to coarse sand facies and some cross-stratifiedfine sand facies. Width
to thickness ratios were estimated for bed sets from the outcrop analog to be an average
of 7.7: 1. This ratio corresponds well to published data by Robinson and McCabe (1997),
who reported trough cross bed sets have an average width to thickness ratio of 8.5:1 to
10.4: 1. It was not possible to measure the bed lengths, since the outcrop is two
dimensional and bed lengths at the site are longer than the scale of the outcrop. No
published data were available on the bed width, depth and length ratios, likely for the

reason stated above.

Geostatistics

We use the stratigraphic zones identified at the Schoolcraft site as geologic
regions for geostatistical interpolation. For each zone, experimental variograms were
developed to estimate the correlation lengths in the vertical, parallel to paleoflow, and
perpendicular to paleoflow directions. Variogram parameters were estimated using
horizontal log K values measured from core samples. Log K data were used instead of K
in this and other geostatistical analysis because hydraulic conductivity values for this site
are more log-normally distributed than normally distributed (Hoeksema and Kitanidis,
1985). The parameters for the vertical correlation length were estimated to fit the
experimental variogram. Because horizontal data are sparse, the horizontal,
perpendicular to flow correlation length structures were determined using the 7.7:] width

to thickness ratio measured from outcrop analog. The horizontal, parallel to flow

42

 

Figure 3.6. A portion of the outcrop analog showing the stratigraphic character of the
outwash material with a shovel for scale. Image is presented in color.

43

correlation length structures were extended as far as Groundwater Modeling System
(GMS3.1, BYU, 2000) would allow, approximately 18 m. Yet, we believe this is shorter
than the true parallel-to-flow correlation length based on the outcrop analogs and the
nature of these fluvial deposits. Table 3.4 summarizes the sill, nugget, parallel to flow
range, perpendicular to flow range, and vertical range for each stratigraphic unit. Using
the estimated variogram models and conditioning data, we generated a three-dimensional
K field for each hydrostrati graphic zone using ordinary kriging. The four zones were
then merged to create a final K field surrounding the biocurtain area. This zonal kriged
field preserved abrupt K changes, especially noticeable at approximately 20-21 m depth
in zone 2 (Figure 3.5).

In addition to the zonal kriged results, a K field was estimated by kriging without
the zonal boundaries to evaluate the influence of the additional geologic information. For
this model, the variogram range was set at 18.0 (to be consistent with the zonal kriging).
The nugget, sill, and range in all three orthogonal directions were fit to the experimental
variograms based on the core data (Table 3.4). In a visual comparison, the zonal kriged
field had significantly more horizontal bedding-like features than the traditional kriged
field, which had smoother features with less horizontal continuity. Additionally, the low
K area, (20—21 m bgs), interpreted as an abandoned channel fill, is much larger and

diffuse in the non-zonal kriged case than suggested by the core data (Figure 3.5).

Groundwater Modeling and Tracer Test Simulation

We used the three-dimensional groundwater flow model MODFLOW-96

(McDonald and Harbaugh, 1988; Harbaugh and McDonald, 1996) to compute the flow

Range

 

 

 

 

 

 

nu et Sill parallel to flow perpendicular to flow vertical
non-zonal model 0.011 I 0.065 I 18.0 T 6.84 i 2.700 ]
zonal model
zone 1 0.012 0.026 18.3 2.61 0.353
zone 2 0.027 0.21 17.7 4.09 0.531
zone 3 0.0017 0.020 18.0 8.33 1.081
zone 4 0.0053 0.024 18.4 12.51 1.619

 

 

 

 

 

 

Table 3.4. Variogram parameters for zonational krige and non-zonational krige. All

variograms fit using exponential model.

45

 

/ Non-zonal Kriged Results

 

Figure 3.7. Zonal (top) and non-zonal (bottom) kriged results. Arrows mark boundaries
of zones in zonal krige. Note that abrupt changes in K are preserved in zonal krige
approach. Scalebar is in log K (cm/s).

46

and heads for the region. Constant head boundaries were used in the flow direction to
provide a gradient of 0.0011 based on regional head measurements (Figure 3.2). The
model domain is a rectangular region 101.5 m wide (Y) by 57.2 m long (X) by 27.4 m in
depth (Z). We discretized this region using a computational grid with 136 (Y) x 86 (X) x
44 (Z) cells (Figure 3.8). The delivery well gallery is located at the center of the
computational domain such that fine cells (20 cm x 20 cm) approximately equal to the
size of the well boreholes surround the delivery wells. Cell size increased in a geometric
progression away from the well gallery. The resolution in the vertical direction varies
and is discretized most finely around the low conductivity zone located at approximately
20-21 m bgs (Table 5).

During the tracer test, a conservative tracer (bromide) was injected into the seven
even-numbered wells and extracted from the eight odd-numbered wells for the first five
hours. During the next hour the same pumping rate was used in a flow reversal phase
where tracer was extracted fiom the even wells and injected into the odd wells. We used
the reactive transport code RT3D (Clement and Jones, 1998; Clemet, 1997) to simulate
three-dimensional tracer transport through the site. Since the injected bromide
concentration changed during the tracer test, we divided the five-hour interval in the
transport model into three stress periods with different concentrations, 18, 14, and 17
ppm respectively. The one hour flow reversal phase was simulated with a single stress
period where injected concentrations were 23.5 ppm. The last stress period in this
simulation represents a 20-day natural gradient period with no pumping. The differences
in hydraulic conductivity across the vertical extent of the aquifer, cause proportional

differences in the flux through different layers of the well bore. To represent this

47

 

Figure 3.8. Discretization of grid in plan view, delivery well gallery is located near the
center of the grid, marked by light colored squares, where dicretization is the finest.

layers thickness of layer (m)

 

Table 3.5. Vertical discretization of layers.

48

behavior, we used a conductivity-weighted average to compute the fluxes for the

injection and extraction wells.

Results and Discussion

The tracer test comparison shows that the addition of geologic data can significantly
improve simulations in some localities while in others the zonal and non-zonal simulated
tracers are very similar (Figure 9). In most cases, the hydrostratigraphic interpretation
improves the match between simulated and observed tracer concentrations. The most
significant improvement was at 19.8 m depth, especially at wells MW10 and MW 13.
The massively bedded to finely laminated very fine sand facies, interpreted as an
abandoned channel fill, is present at this depth over about half of the simulation area.
The core descriptions and K data place this channel fill through delivery wells 8, 10, and
12. MW 13 and MW10 are located downgradient from delivery wells 10 and 8
respectively (figure 3). MW10 and MW13 are the most affected by the addition of
geologic data because this is where the low K channel fill is located. At the same depth
with no evidence for the low K abandoned channel fill, tracer simulations MW11 and
MW12 respond similarly for both types of kriging. No core was recovered from well D2
at this depth, however, the measured tracer breakthrough at MW9 is very similar to the
measured data from MW13, suggesting another possible abandoned channel fill at this
location.

At most other depths, the comparison between tracer tests simulated in the zonal
and traditional kriging simulations are very similar. Slight improvements are made in

MW10, and MW13 at depth 22.9 m, and MW10 and MW11 at depth 16.8 m, and MW11

49

10-7 13.7

 

 

 

 

 

 

 

 

 

1P 7 b r
_ Zonal
0.5er 1‘ 1' "MN
1 ’ ‘r
a
0.5' " ’ W10
“-4 ii a
a
or 4 3 ‘ 2
1 r 1-
a
C10. 0.15» .. . W11
11
1" I a n
o W ‘ ..
1 V v v 4}
0.5' 1r 1i MW12
1i 0
0.5: 1’ MW13
0* W‘s—q

 

 

 

 

10

°
d ’
O
O

Time (Days)

Figure 3.9. Measured tracer test from site (x) with simulated zonal krige (black line) and
simulated non-zonal krige (gray line).

50

at 13.7 m and 10.7 m, where the breakthrough occurs a bit earlier in the zonal kriging
tracer simulation.

There are a few instances where the two kriging approaches provide similar
transport results and a few sites where the traditional kriging breakthrough better matches
the measured tracer concentrations. The nonzonal krige transport simulation better
matches the measured tracer at MW12 at 17.8 m depth. This is because the nonzonal

krige interpolated a higher K value than the zonal krige interpolated at 17.8 m depth.

Conclusions

The addition of stratigraphically significant boundaries into geostatistical
interpolation methods helps preserve abrupt changes in K as they occur in the
sedimentologic record, thus better honoring the stationarity assumption implicit in the
kriging method. The zonal kriging approach requires an understanding of the
depositional environment, which can be accomplished by visual analysis of the core
material to evaluate stratigraphic units. In locations where core data do not provide
enough information, outcrop analogs were used to provide approximate correlation
lengths for different facies type.

Kriging K values using the stratigraphic fi'amework identified by the core yields a
more realistic aquifer characterization that better matches the true geology of a site. The
zonal model of K improved the tracer simulations, most evident where the heterogeneity
is the highest or in regions where there is a contrast in K of an order of magnitude or

more.

51

Chapter 4

Transition Probability Geostatistics

Introduction

Transition Probability Geostatistics is an indicator geostatistical method that can
be used to estimate the spatial distribution of facies based on geologic information input
by the modeler. This is done by first categorizing the data set into three to five
geologically distinct facies or hydrofacies. Second, a three-dimentional Markov chain
model is fit or developed for the data. Third, this model is used in sequential indicator
simulation to simulate the three dimensional facies distributions. Important geological
attributes such as 1) volume fraction (proportions), 2) mean lengths (e. g. thickness and
lateral extent) and 3) juxtapositional tendencies (how one category tends to locate in
space relative to another) are considered in this approach (Carle et al., 1998). Markov
chains offer a stochastic model for categorical variables that incorporate all of these
geological attributes (Carle and Fogg, 1998). The Markov chain is a model used to
mathematically describe the transition probabilities seen in geologic medium. The
probability of one facies transitioning directly to another is used to help define the
Markov chain model. After the Markov chain models are developed in three principle
directions (vertical parallel to flow and perpendicular to flow), conditional simulation
followed by simulated annealing was run. Conditional simulation is a process that
creates multiple, equally probable spatial distributions of random variables or
“realizations” that honor hard data conditioning points at specified locations (Carle et al.,

1998). This is a much different method of estimating distribution of K than kriging,

52

which will only produce one distrubution of K (Deutsch and J ounel, 1992). Simulated
images do not exhibit the characteristic smoothing effect of kriging (Deutsch and J ounel,
1992). Therefore, simulated images better preserve abrupt changes in K.

Existing variogram based methods do not provide 1) consideration for asymmetric
juxtaposition relationships such as fining upward tendencies; 2) a framework for
incorporating geologic interpretation of proportions, lengths, and juxtaposition tendencies
into models of spatial variability; or 3) consideration for locally variable anisotropy
directions (Carle et al. 1998). Incorporating geologic interpretation of proportions,
lengths, and juxtaposition tendencies into models of spatial variability, as well as
consideration for asymmetric juxtaposition relationships are very important to creating an
accurate geologic model at the Schoolcraft site. By enforcing juxtaposition relationships,
proportions, and lengths, based on geologic data from the site, geologically plausible
simulations of the distribution of facies will are produced even where the data are sparse.

Although the transition probability geostatistical approach produces geologically _
plausible distributions of heterogeneity, this is only accurate where stationarity can be
assumed, or where the mean and standard deviation do not vary spatially (Weissmann
and F ogg, 1999). The Schoolcraft site as a whole is not stationary, as described
previously in chapter 3. For the transition probability/ Markov chain based simulations,
the Schoolcraft site was divided into 2 zones, to better honor stationarity --a coarse
grained bottom zone (medial deposits) and a finer grained upper zone (distal deposits).
The upper and bottom zones were simulated separately and later combined to create a full

aquifer characterization.

53

Stratigraphic Zones

Two zones were identified at the Schoolcraft Plume A site —a gravely zone at the
bottom of the aquifer (22-27 m bgs) that represents intermediate (Donj eck-type) outwash
deposits and —a sandy zone at the top of the aquifer (8-22 m bgs) that represents distal
(Platte-type) outwash deposits (see Appendix A). The bottom zone was the same as zone
1 used in the zonal kriging (see chapter 3). The top zone was a compilation of zones 2-4
used in the kriging chapter (3). These upper zones were combined since similar
hydrofacies are present.

The mean and standard deviation of grain size are significantly different between
the two zones (Table 4.1). To show that mean grain size and standard deviation do not
depend on location in the top zone of the transition probability geostatistics simulations,
the data were divided into the same three zones as the kriging zones from chapter 3 and
assessed for mean grain size and standard deviation. Division of the region into two
stratigraphic zones for transition probability modeling appears justified since mean and
standard deviation are approximately the same for kriging zones 2-4 (Table 4.2).
Additionally, the bottom zone consists of predominately medium (46.6%) and coarse

sand (23.6%) while the top zone consists of predominately fine sand (74.2%).

Explanation of Hydrofacies from Categories
Within each zone, five distinct hydrofacies were identified by comparing grain
size, sorting and K. Mean grain size and standard deviation of grain size were measured
from sections of sieved cores (P18, D2, D4, D6, D8, D10, D12 and D14). Core P18 was

also described for sedimentologic properties (Appendix A, Figure A.2). Many

54

T. P. zones mean GS mean GS mean SD mean sorting no. data points

 

1 0.11 very coarse sand 1.73 poorly 29

2 2.29 upper fine sand 0.77 moderately 84
Kriging zones

1 0.11 very coarse sand 1.73 poorly 29

2 2.28 upper fine sand 0.62 moderately well 5

3 2.08 upper fine sand 0.78 moderately 36

4 2.48 fine sand 0.79 moderately 42

Table 4.1. Mean grain size (GS) in Phi unit, corresponding Wentworth size class (Boggs,
2001), mean standard deviation (SD), sorting based on standard deviation (Boggs, 2001),
and number of data points in each zone for the two transition probability (T. P.) zones
and the four zonal kriging zones. Zone one is the same for the transition probability and
zonal krige zones.

 

 

Ifldrofacies of the Top Zone Proportions
Poorly sorted medium sand (PMS) 0.065
Poorly sorted coarse sand (PCS) 0.032
Moderately to well sorted fine sand (MW FS) 0.742
Poorly sorted fine sand (PFS) 0.130
Very fine sand (VFS) 0.032
Hydrofacies of the Bottom Zone Proportions
Pebbles, gravel and very coarse sand (PGVCS) 0.135
Moderately well to well sorted medium sand (W MS) 0.115
Poorly sorted coarse sand (PCS) 0.236
Poorly sorted medium sand (PMS) 0.466
Moderately to well sorted fine sand (MWFS) 0.048

Table 4.2. Proportions of hydrofacies in the top and bottom zone.

55

things were considered when choosing categories, including: 1) relationship between
grain size and sorting, 2) the interpreted process under which each grain size / sorting
combination was deposited and 3) the distribution of K for each grain size / sorting
combination. Generally, a relationship exists where, for any grain size, the better a sand
is sorted the higher the K value, assuming a normal distribution of grain size (Beard and
Weyl, 1973). Measured data indicate from the Schoolcraft site indicate that this
assumption is true, even though the sands and gravels from the Schoolcraft site lack
normal distribution (Table 4.3 and Table 4.4). Thus, it is believed that this is still

reasonable to assume this relationship exists for sands and gravels at the Schoolcraft site.

Top Zone

For the top zone, the five identified categories are: l) moderately to poorly sorted
medium sand (PMS), 2) poorly to very poorly sorted coarse sand with pebbles (PCS), 3)
well to moderately sorted fine sand (M WFS), 4) poorly sorted fine sand (PFS) and 5)
moderately to poorly sorted very fine sand ( VF S). Mean K decreases fiom category one
to category five. Mean, high and low K values are given in Table 4.3, and plotted in
Figure 4.1.

PMS (category 1) is the hydrofacies with highest K in the top zone, and is poorly
to moderately sorted. In core, this hydrofacies displayed either faint cross-stratified
bedding or massive bedding and grain shape is rounded to subrounded. The sediment
color is light olive brown (Munsell 2.5Y 5/4) when wet and pale yellow (Munsell 2.5Y
7/3) when dry. This hydrofacies tends to occur as lenses within the MWF S hydrofacies

and is interpreted to have been deposited on channel bars.

56

Category Hydro facies

1

01-wa

TOP ZONE

 

no. of

averafi K high K low K points
oorly sorted medium sand (PMS) 0.025 0.04 0.005 16
Poorly sorted coarse sand (PCS) 0.019 0.044 0.004 4
oderately to well sorted fine sand (MWFS) 0.013 0.029 0.007 51
oorly sorted fine sand (PFS) 0.010 0.032 0.002 6
ery fine sand (VFS) 0.008 0.012 0.002 7

Table 4.3. Hydrofacies and corresponding assigned category for top zone with high K
value, low K value, average K value, and number of points.

 

BOTTOM ZONE
no. of
Category Hydro facies avenge K hiLhK low K points
1 ebbles, gravel and very coarse sand (PGVCS) 0.057 0.103 0.024 11
2 oderately well to well sorted medium sand (WMS) 0.043 0.066 0.018 4
3 oorly sorted coarse sand (PCS) 0.041 0.062 0.018 7
4 oorly sorted medium sand (PMS) 0.032 0.046 0.017 5
5 oderately to well sorted fine sand (MWFS) 0.017 0.019 0.016 3

Table 4.4. Hydrofacies and corresponding assigned category for bottom zone with high K
value, low K value, average K value, and number of points.

57

 

 

 

 

 

 

 

 

 

 

 

00500«-—~r—~—1+——’T’*T*-j~+jl
I . ’ .
a E 004001 _ - f - - __-1___-._W ”.5, l
l E 00300 '—- — --—# ----- “w“? “55—4-

1 '4 00200I i . i i J

l 5 i i fir 1» l 1
g x 0.0100 —-—-4— 1’ 1i "“1 i
1 . ‘ l

, 0.0000 ,_ d4 1 21

‘ o 1 2 3 4 5 6

l Category

 

 

Figure 4.1. Top zone of Schoolcraft site, 1-5 are categories of hydrofacies, corresponds
to Table 4.3. Top diamond is high K value, bottom diamond is low K value and middle
diamond is mean K value.

58

The PCS hydrofacies (category 2) is poorly to very poorly sorted and contains
both pebbly coarse sand and coarse sand without pebbles. These two lithologies were
combined because they both display similar K values and represent the coarsest grained
sediments in the upper zone. The pebbly sand is matrix supported with rounded pebbles
and subrounded grains. The matrix consists of coarse and medium sand. The color is
olive brown (Munsell 2.5Y 4/4) when wet, and pale yellow (Munsell 2.5Y 7/3) when dry.
The samples of coarse sand without pebbles could not be described for sedimentologic
features since only sieve data exists for these samples. This hydrofacies is interpreted to
have been deposited on channel bars.

The M WF S (category 3) is the most abundant hydrofacies in the top zone (Table
4.2). The moderately to well sorted sand appears to have thicker and more continuous
beds than any of the other hydrofacies in this zone (Figure 4.2). In core, MWF S displays
massive bedding from the base of the zone to about 16 m bgs, above which it becomes
cross-stratified (see Appendix A Figure A2). The color of the sand is generally olive
brown (Munsell 2.5Y 4/4) when wet and pale yellow (Munsell 2.5Y 7/3) when dry, and
the grain shape is generally subrounded. This hydrofacies is interpreted to have been
deposited on channel bars.

The PF S (category 4) has a lower K than the moderately to well sorted fine sand
due to its poorer sorting. It is generally cross-stratified with either coarser grains along
cross-stratified beds or with seemingly randomly placed pebbles composed of sand
cemented with calcium carbonate. This hydrofacies appears to be thinner and laterally
discontinuous when compared to M WFS. This hydrofacies is interpreted to have been

deposited on channel bars.

59

Cross-section of top zone

 

 

 

 

8' D4 06 D10 D12 5
DZ . . .

101 ‘ P18 01.4 l -4.5
e O ' ' e

.2. D8 __ . . .,
O . ' D .

14. . . . 35
O

16* . . ‘ . 3
e I .1.

18— 7 1 2.5
e . . . e . 5 9

20— . . ' . . 2

22* O . . 1.5

24 L i 4 r 1 1 1 d1
22 24 26 23 30 32 34 36

Y direction (m)

Figure 4.2. Cross-section of categories in top zone. Category 1, poorly sorted medium
sand (PMS); category 2, poorly sorted coarse sand (PCS); category 3, moderately to well
sorted fine sand (MWFS); category 4, poorly sorted fine sand (PFS); category 5, very fine
sand (VFS). X axis is depth (m). Image is presented in color.

60

The VF S (category 5) is the lowest K hydrofacies and it is this facies that forms
the “low K zone”, discussed in earlier chapters. In core, the very fine sand is finely
laminated, had a high amount of silt and clay (16%), and was well to moderately sorted.
The sediment color is olive brown (Munsell 2.5Y 4/4) when wet and pale yellow
(Munsell 2.5Y 7/3) when dry. This hydrofacies was interpreted as abandoned channel
fill, commonly found in medial and distal outwash facies (Donjeck and Platte type, as
described by Miall, 1977 ). Abandoned channel fills are common in medial and distal
outwash facies during flooding where what is commonly referred to as overbank fines

infill abandoned channels (Miall 1985; Benn and Evans 1998.)

Bottom Zone

For the bottom zone, the five identified categories are: 1) pebbles, granules, and
very coarse sand (PG V CS), 2) moderately well to well sorted medium sand (WS), 3)
poorly to very poorly sorted coarse sand (PCS), 4) poorly to very poorly sorted medium
sand (PMS) and 5) moderately and moderately to well sorted fine sand (MWF S). Mean K
decreases from category one to category five. Mean, high and low K values are given in
Table 4.4, and plotted in Figure 4.3.

The PG VCS (category 1) is the hydrofacies with the highest K in the bottom zone
(Figure 4.3, Table 4.4) and is composed of three different mean grain sizes pebbles,
gravels and very coarse sand, all of which are poorly or very poorly sorted. These
different sub-categories were grouped together because within each there was a high

abundance of the other two grain sizes (e. g. the very coarse sand had pebbles and

61

 

Bottom Zone

 

 

 

 

i 0.1000 1 , T—H I -—-»| , - -j
7,; 0.0800 I ,- + i
I 3 0.0600 --- f” A“ -- ;

1 x 0.0400 —-——~ "—1 -__,_
0.0200 -.——_ 4 __

I 4
4
0.0000 1 -- l i , -—- —T i

I ‘.
i o 1 2 3 4 5 6
l
l

 

0.1200 ll -——-
_

 

 

 

 

 

 

 

 

Figure 4.3. Bottom zone of Schoolcraft site, 1-5 categories of hydrofacies, corresponds
to Table 4.4. Top diamond is high K value, bottom diamond is low K value and middle
diamond is mean K value.

62

granules present) and all three subcategories had similar K values. In core, the pebbles
and granules are rounded and grain supported and display cross-stratified bedding with
the coarsest grains tending to occur along cross-stratified beds. The samples of very
coarse sand were not described for sedimentologic features since only sieve data exists
for these samples. This hydrofacies is present along the base of the aquifer and is
interpreted to represent channel lag deposits.

The WMS (category 2) is well to moderately well sorted, has rounded grains, and
displays cross-stratified bedding. Coarser grains exist along bedding planes. The
sediment color is olive brown when wet and light olive brown to pale yellow when dry.
This hydrofacies is interpreted to have been deposited on channel bars.

The PCS (category 3) is poorly to very poorly sorted. The coarse sand is cross-
stratified with subrounded to rounded grains. Coarser grains exist along bedding planes.
Grain size ranges from silt to granules. The coarse sand is light olive brown (Munsell
2.5Y 5/4) when wet and pale yellow (Munsell 2.5Y 7/3) when dry. This hydrofacies is
interpreted to have been deposited on channel bars. This hydrofacies is very similar to
the PCS hydrofacies in the upper zone and consequently has the same name.

The PMS (category 4) is distinct from category 2 because of differences in
average K (Table 4), and sorting. Grain size ranges from silt to granule within this facies.
There is a fining upward relationship from PMS (category 4) to WMSM (category 2). In
core, the PMS hydrofacies is cross-stratified and displays coarser grains along bedding
planes. This hydrofacies is interpreted to have been deposited on channel bars. This
hydrofacies is very similar to the PMS hydrofacies in the upper zone and consequently

has the same name.

63

The M WF S (Category 5) is the lowest K hydrofacies in the bottom zone. This
hydrofacies is distributed in very small proportions in the bottom zone. Because of the
finer grain size than the rest of the bottom zone, this hydrofacies is interpreted to have
been deposited on channel bars under a lower energy than the rest of the sediments in the
bottom zone. This hydrofacies is very similar to the MWF S hydrofacies in the upper zone

and consequently has the same name.

Markov Chain Models
Top Zone Markov Chain Models
The vertical (z)-direction transition probability values for the top zone were
measured from core based on a 0.25 m-spacing. Proportions of categories were estimated
from the core data (Figure 4.2). In this zone, the MWFS hydrofacies has the highest
proportions (74.2%) and seems to fill in around all the other categories. Thus MWF S was
used as the background category (Table 4.2). Unfortunately, the data are very sparse and
the addition of more data would have significantly improved the measured and modeled
Markov chain models. Moreover, an outcrop analog that closely matched the distribution
of grain size and beds to this upper zone was not found, so a Markov chain model could
not be developed fi'om analog data. Using the available data, however, a Markov chain
model was fit to the measured data (Figure 4.4). The vertical embedded transition
probabilities and the mean lengths of the Top zone are given in Table 4.5. Because the
M WFS hydrofacies is the background category, the vertical (z)-direction embedded

transition probabilities are the highest where M WF S occurs. The highest embedded

64

PFS MWFS PCS PMS

VFS

Transition Probabili

Top Zone

PCS

MWFS

PFS

Vertical Transition Probability

VFS

 

 

 

 

f

I l A L A

 

 

 

 

 

 

v

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.5
0.0 4 r3' . D
o
I-aa (m)
Measured Modeled
Figure 4.4.

Matrix of vertical (z)-direction transition probabilities showing core data as
measurements (open circles) and the Markov chain model (solid line) from

the core measurements. The diagonal elements represent auto-transition
probabities within a category, and the off diagonal elements represent transition
probabilites between categories.

65

PMS
PCS
MWFS
PFS
VFS

 

PMS
PCS
MWFS
PFS

 

VFS_

PMS?

PCS
MWFS
PFS
VF S

 

Table 4.5. Embedded transition probability matrices for the top zone. These matrices are
read as transition probabilities fiom the row hydrofacies to the column hydrofacies.
(hydrofacies labels: PMS, poorly sorted medium sand; PCS, poorly sorted coarse sand;
MWFS, moderately to well sorted fine sand; PFS, poorly sorted fine sand; VFS, very fine
sand. Other labels: L, mean length. Bold numbers indicate background category with

PMS

L=.750

0.100
0.208
0.001
0.050

PMS

L=30.0

0.100
0.178
0.001
0.050

PMS

L=3.00
0.100
0.182

0.001

0.050

Vertical (z)-direction

PCS MWFS PFS
0.100 0.889 0.010
L=.400 0.898 0.001
0.203 L=2.089 0.363
0.001 0.997 L=1 .000
0.001 0.948 0.001

Parallel to flow (x)-direction

PCS MWFS PFS
0.100 0.799 0.100
L=10.0 0.898 0.001
0.303 L=74.74 0.414
0.001 0.997 L=30.0
0.001 0.948 0.001

Perpendicular to flow (y)-direction

PCS MWFS PFS
0.200 0.699 0.100
L=1.0 0.898 0.001
0.288 L=7.64 0.424
0.001 0.997 L=3.0
0.001 0.948 0.001

computed values listed in the table.)

66

VFS

0.001
0.001
0.227
0.001
L=.4OO

VFS

0.001
0.001
0.104
0.001
L=31.0

VFS

0.001
0.001
0.106
0.001
L=3.1

 

 

 

transition probability is fi'om PFS to M WFS (0.997). This is likely due to the fact PFS
hydrofacies is often found as an erosional base beneath the MWF S hydrofacies and
because the proportion of the M WFS hydrofacies. The mean lengths were estimated from
the core description and the cross-section of categories in the top zone (Figure 4.2).

The parameters used to estimate the perpendicular to flow (y)-direction Markov
chain model (Figure 4.5) were estimated from geologic reasoning, juxtaposition
relationships and the vertical (z)-direction embedded transition probabilities in an
application of Walther’s law (Carle et al., 1998). Walters law states that only those facies
that can be found forming side by side in nature can occur in contact with one another in
a conformable vertical succession (Miall, 1996). The mean lengths were estimated from
the (z)-direction mean lengths. Robinson and McCabe (1997) reported that trough cross
bed sets have an average width to depth ratio of 8.5:1 to 10.4:1, depending on unit. A
width to thickness ratio of 7.7 :1, measured in an outcrop analog which most closely
matched the grain size distribution and vertical thickness of beds in this zone, was
applied to the mean lengths in the (z)-direction to estimate (y)-direction mean lengths.
Even though this outcrop analog was not a close enough analog to add additional
information on juxtapositions, based on the similarity of the width to thickness ratio to
published data and the fact that it is the best data available, I applied this width to
thickness ratio in this zone. However, there is a high amount of uncertainty in the
embedded transition probabilities and the mean lengths due to lack of sufficient
continuous data.

The parallel to flow (x)-direction Markov chain model were estimated fi'om the

perpendicular (y)-direction embedded transition probabilities in an application of

67

PFS MWFS PCS PMS

VFS

Top Zone
Horizontal Transition Probability

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4 V 4 4
I 4
0 4 .
o J .
'1 O -I 0 — .-
1 I 1
0 ° ° 0 ‘
I: 1 I i 1
'- 4 4 1 4
.n . . J l _
fl 4 4 4 4 4
.fl ‘ I I I I
o 4 4 4 4
h T.fi~r.:. 4 :Ir.= 4% 4pm
I. . . .
I I\ . ,
3 : i e :
o _ - - ° - -
'3 O 4 4 m 4 4
I- O O : I 0 j 0 m
U, ‘ o ‘ o
: 4m1¢m rsT . .4, -e . ..fi.
:- - i o
4 4 4 1
P 4 J 4 4 4
‘. . ‘ ‘. 1
4 4 4
4 o 4 4
4 4 4 4
4M4m Ice fies, .: v...
1.0, ‘ x -
4 4 4 4
4 4 4 4 4
1 4 d 4 J
0-5: 1 1 1 1
4 4 4 4 4
4 4 4 4
4 4 4 4 4
Ono4p-r-III-IIIIU-T-Pvll-Tflsr:| 4W . .Te 1 .e. :3. 4m . . r. 1 .3. cc.
0 2 5
Lag (m)
Measured Modeled
O O 0 O O O 0
Figure 4.5.

Matrix of perpendicular to flow (y)-direction transition probabilities showing core
data as measurements (open circles) and the Markov chain model (solid line) from
the core measurements. The diagonal elements represent auto-transition probabities
within a category, and the off diagonal elements represent transition probabilites
between categories.

68

Walther’s law (Carle et al., 1998). The mean lengths were estimated from the (y)-
direction mean lengths. The embedded transition probabilities are ahnost the same as the
perpendicular to flow (y)-direction embedded transition probabilities (Table 4.5). The
mean lengths in the (x)-direction were estimated using the perpendicular to flow mean
lengths and multiplying them by 10, an estimate of what the width to length may be off
outcrop analogs that were most similar to the characteristics of the top zone.
Unfortunately the outcrops were not large enough to measure the length of these units so

there is significant uncertainty in the mean lengths in the (x)-direction.

Bottom Zone Markov Chain Models

Vertical (z)-direction transition probability values for the bottom zone were
measured from the data, based on a 0.25 m spacing (Figure 4.6). The core data are
sparse, and, for that reason, the Markov chain model fit to these measurements was not
very good. In an effort to incorporate more data for model development, a 0.25 m grid
was overlain on 3 described outcrop analog which seemed to best fit the facies
relationships, proportions, and the thickness of beds in the bottom zone (Figure 4.7).
Proportions of categories were estimated based on core data and the outcrop analog
(Table 4.2). The additional information on the spatial distribution of facies provided by
the outcrop analog helped the Markov chain model, although the addition of more data
would further improve the model if the data were available. Figure 4.8 shows the vertical
(z)-direction Markov chain model that was fit to the measured data. The embedded

transition probabilities and mean lengths used in this model are given in Table 4.6.

69

Cross-section of the bottom zone

 

 

 

 

D10
22 02 “I P18 H 5
e 9 '
22 5 » DG 45
. .
23» .
3‘ D12 - ‘4
23.5
1’ D14
24 _ . r - 3.5
O
24.5 C 3
O
25 0
. 2.5
O O
255-.
O 2
25. S n9
0 e e
O 1.5
265 : .
. .
27 A A A J A . l 1
22 24 26 28 30 32 34 36

 

Y direction (m)

Figure 4.6. Cross-section of category in bottom zone. Category 1, pebbles, granules and
very coarse sand (PGVCS); category 2, poorly sorted medium sand (PMS); category 3
poorly sorted coarse sand (PCS); category 4, poorly sorted medium sand (PMS); category
5, moderately and moderately to well sorted fine sand (MWFS). Image is presented in
color.

70

 

Figure 4.7. Outcrop analog with 0.25 m grid overlay. Image is presented in color.

71

WMS PGVCS
Probabili

PCS

 

Trans

Bottom Zone
Vertical Transition Probability

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 1 o 1
4 4 4
‘ 4 ‘1 0
fl 1 A
- -i 0 -1 0
4 o 4 4
4 0 4 0 4 0 O
. 4 o 1
‘ 1 A j °o° o o
r V v v I v v V W. 'vY v v v ' 'v'vY W "V
. . °
o°°
4 o 4 °
4 ‘ 0
0

II! -I A %

4 o o 4

4 4

I v m V v V ' l m
o o 0 O

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o o 0000 c o o 2 o
O 4
° - . . , . - M 4m
1 o c
. : . 1 00
e: i o .: :4 00°
1 d d 1
'1 4 o 0 o 0 0°
1 » ‘ .
O " 0
1m ' :V c l I I: I‘fT Iif
, I . .f ..
J o 1 0°
4 1
4 . 0
o - ° -
4 1 V
o 1
O 0 V 4{ 0
0.
T ' T ' v ' T ' ' ' V ' r T ' ' ' '
2
Lag (m)
Measured Modeled

0000000

Figure 4.8.

 

Matrix of vertical (z)-direction transition probabilities showing core data as
measurements (open circles) andthe Markov chain model (solid line) from the
core and outcrop analog measurements. The diagonal elements represent
auto-transition probabities within a category, and the off diagonal elements
represent transition probabilites between categories.

72

PGVCS
WMS
PCS
PMS

 

MWFS_

PGVCS
WMS
PCS
PMS

 

MWFS_

PGVCS
WMS
PCS
PMS

 

MWFSL

_

Vertical (z)-direction

PGVCS WMS PCS PMS MWFS

L=.420 0.500 0.240 0.250 0.010
0.050 L=.190 0.250 0.690 0.010
0.010 0.005 L=.200 0.885 0.100
0.162 0.257 0.542 L=.273 0.038
0.010 0.010 0.120 0.860 L=.25

 

Parallel to flow (x)-direction

PGVCS WMS PCS PMS MWFS

 

L=32.0 0.250 0.450 0.290 0.010
s L=15.0 0.250 0.603 0.010
s s L=15.4 0.735 3
0.065 0.248 0.602 L=25.0 0.084
s 3 0.120 0.818 L=25.0 #

Perpendicular to flow (y)-direction

PGVCS WMS PCS PMS MWFS

 

L=3.20 0.250 0.450 0.290 0.010
s L=1.50 0.250 0.603 0.010
s s L=1 .54 0.735 3
0.065 0.248 0.602 L=2.50 0.084
s 3 0.120 0.818 L=2.50 _

Table 4.6. Embedded transition probability matrices for the bottom zone. These matrices
are read as transition probabilities from the row hydrofacies to the column hydrofacies.
(hydrofacies labels: PGVCS, pebbles, gravel and very coarse sand; WMS, well sorted
medium sand; PCS, poorly sorted coarse sand; PMS, poorly sorted medium sand;
MWFS, moderately to well sorted fine sand. Other labels: L, mean length. Bold
numbers indicate background category with computed values listed in the table.)

73

Both fining upward and coarsening upward tendencies are present in the bottom
zone and preserved in the Markov chain model evident in the W.r.t independent transition
frequencies. If the W.r.t values are greater than 1.0 then the transition is more likely to
occur and if the W.r.t. values are less than 1.0 than the transition is less likely to occur
(Table 4.7). There is a slight fining upward tendency, from PG VCS up to WMS, from
PG VCS up to PCS and from PCS up to PMS. However, it is not uncommon to coarsen
upward as well. This occurs from PCS up to PG VS, from WMS up to PCS and from
MWF S up to PMS.

The perpendicular to flow (y)-direction Markov chain model and mean lengths
(Figure 4.9) were estimated from the vertical (z)-direction embedded transition
probabilities and vertical (z)-direction mean lengths in an application of Walther’s law
(Carle et al., 1998). To estimate (y)-direction mean lengths, a 7.721 width to thiclmess
ratio, measured in the outcrop analog, was applied to the mean lengths in the (z)-
direction. Fining outward tendencies are present in the bottom zone and preserved in the
Markov chain model, evident in the W.r.t independent transition fiequencies (Table 4.7).
The PG VCS transitions outward to WMS and PCS, WMS transitions outward to PMS, and
PMS transitions outward to M WFS. For the (y)-direction Markov chain model symmetry
was forced, noted by an s in Table 4.6.

The parallel to flow (x)-direction Markov chain model (Figure 4.5) and mean
lengths were estimated from the perpendicular to flow (y)-direction embedded transition
probabilities and the (y)-direction mean lengths in an application of Walther’s law (Carle

et al., 1998). The embedded transition probabilities are the same as for the perpendicular

74

PGVCS
WMS
PCS
PMS
MWFS

 

PGVCS
WMS
PCS
PMS

 

MWFs_

PGVCS
WMS
PCS
PMS

 

MWFS_

Table 4.7. W.r.t independent transition fi'equencies matrices for the bottom zone. These
matrices are read as independent transition probabilities from the row hydrofacies to the
column hydrofacies. (hydrofacies labels: PGVCS, pebbles, gravel and very coarse sand;
WMS, well sorted medium sand; PCS, poorly sorted coarse sand; PMS, poorly sorted
medium sand; MWFS, moderately to well sorted fine sand. Other labels: L, mean length.
Bold numbers indicate background category with computed values listed in the table.)

PGVCS

L=.420
1.000
0.177
1.288
0.215

PGVCS

L=32.0
s
3
0.534
s

Perpendicular to flow (y)-directlon

PGVCS

L=3.20
s
3
0.534
3

Vertical (z)-direction

WMS PCS PMS
5.290 1.154 0.373
L=.190 1.146 0.982
0.044 L=.200 1.110
1.030 0.986 L=.273
0.108 0.588 1.310

Parallel to flow (x)-direction

WMS PCS PMS
2.019 1.475 0.534
L=15.0 0.770 1.044
s L=15.4 1.028

1.044 1.028 L=25.0
8 0.407 1.561

WMS PCS PMS
2.019 1.475 0.534
L=1.50 0.770 1.044
s L=1 .54 1.028
1.044 1.028 L=2.50
8 0.407 1.561

75

MWFS

0.358
0.341
3.001
0.517
L=.25

MWFS

0.356
0.334

1.561
L=25.0

MW FS

0.356
0.334
8
1.561
L=2.50

 

 

 

PMS PCS WMS PGVCS

MWFS

 

Bottom Zone
Hortizontal Transition Probability

 

 

 

 

 

 

 

 

 

 

 

 

 

j o
4
4 oo o -
4 o 4
4 ow ‘
O .
I . .23: 4W
1 .
4 4
1 1 000
.1 00% a: O u
a l0 4
O” 4 .
4 I j T I a «W
< o
4
‘ o
d d o q
o 4 4
4
. r I r T- «W
4
4
1
4
. o
d
4
1
1

 

 

 

 

 

 

  

 

 

 

 

 

 

 

Lag (m)

Measured Modeled

0000000

Figure 4.9.

Matrix of perpendicular to flow (y)-direction transition probabilities showing
core data as measurements (open circles) andthe Markov chain model (solid
line) from the core and outcrop analog measurements. The diagonal elements
represent auto-transition probabities within a category, and the off diagonal
elements represent transition probabilites between categories.

76

to flow (y)-direction (Table 4.6). The mean lengths were estimated using the
perpendicular to flow mean lengths and multiplying them by 10, an estimate of what the
width to length may be based on the outcrop analogs. Unfortunately the outcrops were
not large enough to measure the length of these units so there is much uncertainty in the
mean lengths in the (x)-direction. For the (x)-direction Markov chain model symmetry

was forced, noted by an s in Table 4.6.

Conditional Sequential Indicator Simulation
Conditional sequential indicator simulation, followed by simulated annealing
(Carle et al., 1998), was run on the top and bottom zones separately using the appropriate
Markov Chain model for each zone. The cell size for the realization of the Schoolcraft
site was 0.2 m in the lateral directions (x and y) and 0.25 m in the vertical direction (2).
This small grid spacing was used so that the realizations simulated here could be input
into an existing GMS grid with similar dimensions in the future. Information from core

were used as hard conditioning data in the simulations.

Combining the Zones
Once the realizations were complete for the top and bottom zones individually,
the results were combined to create a final whole aquifer characterization. There are
some categories that occur in both the top and bottom zone. After simulation, these
categories were combined to simplify the final product and show similarities between the
two zones. Category 5 from the bottom zone (M WF S) was combined with category 3

from the top zone (M WF S) and on the final plot is called moderately-well sorted fine

77

sand. Category 4 from the bottom zone (PMS) and category 1 from the top zone (PMS)
were combined and is called poorly sorted medium sand in the final realization.

Category 2 from the top zone (PCS) was combined with category 3 from the bottom zone
and is called poorly sorted coarse sand in the final realization. The final realizations of
the Schoolcrafl site are Figure 4.10 and Figure 4.11. Figure 4.11 is separated into five
volurnetrically equal blocks allowing the inside to be more readily viewed.

The final geostatistical realization reflects juxtaposition tendencies, proportions
and the channel orientation used as inputs to the Markov chain model. Fining upward
tendencies are preserved in the top zone where F SP fines upward to MWF S. In the
bottom zone, both the fining and coarsening upward tendencies are preserved from
PGVCS and PCS up to WMSM and from WMSM up to PCS and from MWFS up to
WMSM. Channels are orientated perpendicular to Y axis along the maximum correlation
direction. In the top zone, the high proportion of MWFS is preserved and other
hydrofacies are distributed in channel like geometries. The most striking channel like
geometry is the VFS hydrofacies simulated channel located at about (5, 10, 18) m in
Figure 4.10 and Figure 4.11. This is interesting because no conditioning data indicate
that a channel is present at this location. Other realizations would likely not simulate a
channel at this location. The realization presented here simulated a channel there based
on the Markov chain model. By assigning average hydraulic conductivity values (Table
4.3 and Table 4.4) to each hydrofacies, this realization could be used in groundwater flow
modeling and contaminant transport simulations. Assigning an average K value to each
hydrofacies would preserve abrupt changes in K where other traditional geostatistical

tools such as kriging would smooth K values.

78

Schoolcraft Site

2 Vertical (m)
N

 

4 3 8 1
0
x Paralleldiz)“

Pebbles, granules and very coarse sand (PGVCS)
7 Moderately well to well sorted (WMS)

1 Poorly sorted coarse sand (PCS)

i Poorly sorted medlum sand (PMS)

1 moderately-well sorted fine sand (MWFS)

i Poorly sorted flne sand (PFS)

I Very fine sand (VFS)

  
 
 
 
  

Figure 4.10

A realization of the Schoolcraft plume A site based on transition
probability/ Markov chain geostatistics in two zones. This image is
presented in color.

79

Schoolcraft Site

 

Pebbles, granules and very coarse sand (PGVCS)
Moderately well to well sorted (WMS)

fl Poorly sorted coarse sand (PCS)

E Poorly sorted medlum sand (PMS)

1 moderately-well sorted fine sand (MWFS)

! Poorly sorted flne sand (PFS)

I Very fine sand (VFS)

  
 
 
 
  

Figure 4.11
A realization of the Schoolcraft plume A site based on transition probability/ Markov chain

geostatitics in two zones. This image is presented in color.

80

Conclusions

Using the transition probability geostatistics and the Markov chain models, a
geologically reasonable aquifer characterization was simulated that better honors the
stationarity assumption and better preserves abrupt changes in K. Although this is a very
reasonable approach to characterizing heterogeneity in the subsurface, it is unclear how
much this realization will improve transport studies at the Schoolcraft Plume A site. This
is due primarily to lack of data. More conditioning data would greatly improve the
simulated hydrofacies because the final characterization would honor more real data
points. A few wells with continuous core data, described for sedimentary features could
then be classified into the same hydrofacies used in this study. The continuous data
collected would greatly improve the vertical Markov chain fit thereby improving the
estimated horizontal models. Selected samples could be tested for K as a quality control
measure and to increase the data base where only a few samples were tested previously.
At a small scale such as this, it is the fine details that are of concern to produce accurate
groundwater models and transport simulations. Without closely spaced detailed
continuous core descriptions, only general concepts can be obtained fiom this type of

modeling, and at this site accurate detailed aquifer characterizations are the goal.

81

Chapter 5

Conclusions

Depositional Environment of the Aquifer Below Schoolcraft

Based on grain size data, bedding structures present, location of aquifer and
depositional history of Kalamazoo County it was concluded that the base of the aquifer is
composed of braided stream deposits located at a position medial (intermediate) the
glacier front (Donjeck-type) and the upper portion of the aquifer is composed of braided
stream deposits located at a position distal to the glacier front (Platte-type). The grain
size of the bottom zone ranges from gravel to medium sand and is primarily gravel to
coarse sand. Benn and Evans (1998) stated that a Scott-type stream is composed of
greater than 90 percent gravel and a Donj eck-type stream is composed of 10 to 90 percent
gravel. Thus, it can be reasoned that a Platte-type stream is composed of less than 10
percent gravel. Benn and Evan’s (1998) grain size based divisions of proximal, medial
and distal seem realistic upon comparison to Miall’s (1977) braided stream facies models.
The bottom zone is not classified as proximal to the glacier front (Scott-type braided
stream) because it is not composed of more than 90 percent gravel and because the
aquifer is located to far away from the Kalamazoo Moraine. Proximal (Scott-type)
deposits can be found 0.5 miles from the Kalamazoo Moraine (figure 1). The proximal
deposits at this gravel pit are composed almost entirely of gravel. Medial deposits are
found 2 miles from the Kalamazoo Moraine at a sand and gravel pit where outcrop
analogs were constructed to aid in the interpretation of subsurface geology in this study.

An outcrop analog that was interpreted to have been deposited distal to the glacier was

82

not found. This is likely due to the high water table at distal locations and the fact that
the fine sand is not worth as much as gravel and boulders, so it is not as cost effective to
have a sand and gravel pit in most distal locations. The top of the Schoolcraft aquifer is
interpreted as being deposited distal to the glacier, and is located approximately 6 miles
away from the Kalamazoo Moraine. The fining upward sequence from medial to distal
braided stream deposits at the Schoolcraft site indicate that a glacial retreat was recorded

in this sediments.

Aquifer Characterizations

For all of the geostatistical methods used, lack of data was the biggest problem.
With more data, better variograrns and Markov chain models could be developed.
Increased conditioning data would significantly improve both the zonal krige and the
transition probability simulation because the final characterizations would honor more
real data points. A few wells with continuous core described for sedimentologic features,
sieved to measure grain size distribution and standard deviation, and tested for K value
would accomplish this. Another way to increase continuous core data is to do borehole
geophysics (e. g., resistivity and conductivity logs) in boreholes where core was not fully
recovered. Geophysical logs could then be compared to core descriptions and
interpretations could be made to what type of sediment exist where there was no
recovery.

Closely spaced continuous core is critical to further improvement of the small-
scale characterization of this aquifer. When focusing on such a small area, fine details in

the heterogeneity must be considered, this can not be achieved without sufficient data.

83

Without close-spaced core data, only general concepts of heterogeneity and facies
distribution can be obtained from geostatistical interpolation and simulation methods. At
this site, accurate fine scale aquifer characterizations are the goal. Hence, quality close-

spaced core data are needed.

Kriging Method

The addition of geologically based zones significantly improved the aquifer
characterization as shown by the comparison of simulated to field tracer tests.
Stationarity was better honored in the zonal krige characterization because zones were
based on K distribution, such that each zone had less of a range of K than all the zones
together did. Abrupt changes in K were preserved, especially around the low K zone,
with the addition of stratigraphic zones. By understanding the heterogeneity present in
the model and how that relates to the simulated verse field tracer tests, predictions can be
made of what the geology is where tracer test data are available but no core was

recovered.

Transition Probability Method
Stratigraphically based zones used in the transition probability simulation better
honors the stationary assumption because the zones were delineated by mean grain size.
The top and bottom zone have a mean grain size and a standard deviation of grain size
that are more spatially consistant within that zone than across the boundary and into the
other zone. Transition probability simulations preserve abrupt changes in K by not

smoothing the data, and instead simulating a hydrofacies to every cell. Each hydrofacies

84

has a distinct K value that can be put into each cell. This is an advantage over kriging
which smoothes data and forces the mean of the geostatistical variable at the edge of the
model. As with the other geostatistical methods, lack of quality conditioning data limited
the use of transition probability modeling. It is unclear how much this characterization
method will improve transport studies at the Schoolcraft site, because not enough data

was available to do a thorough characterization.

85

Appendix A

Stratigraphy of the Schoolcraft Plume A Site

Core Data
Core was collected at several wells in the Schoolcraft Plume A site. This core

was described for visual grain size, measured grain size by use of sieves, and/or tested for
K on a constant head permeameter. For this study, all recovered core from well P18 (see
Figure A.1 for location) was described in full for sedimentologic features (Table A.l and
Figure A.2). Portions of core P18 were sieved and measured for vertical K value.
Additionally, segments of core from wells P6, P7 and P8 were described for
sedimentologic features (Figures A.3-A.8). Core descriptions of wells P18, P6, P7, and
P8 and vertical K measurements of P18 were conducted as part of this study. Previous
work by Hoard (2002) assessed the grain size distribution and the horizontal K of
portions of seven of the 15 delivery wells in the biocurtain (D2, D4, D6, D8, D10, D12
and D14) and wells P6, P7 and P8, located down-gradient of the biocurtain (Table A.2-
A.8) Core was collected from approximately 5 m bgs (below ground surface) to 26.5 m
bgs. Previous studies indicated that sediment above 8 m bgs did not show evidence of

carbon tetrachloride contamination thus, data were not collected above 5 m bgs.

Interpretations of Depositional Environment of the Aquifer Below Schoolcraft
Gravel and sand pits located in Kalamazoo County aid in the interpretation of
proximal, medial, and distal location from the glacier at the time the sediments that
compose the Schoolcraft aquifer were deposited. Benn and Evans (1998) state that

proximal (Scott-type) braided stream deposits are composed of greater than 90 percent

86

 

Z

 

round water
“Section
5

D1 .
O 0
' D14
D13 I MW11
, 0
D12 .
I .
. D11
0
' D10 '
09 I MW13 0 P18
° . 0
DB 0 P8' .
I
, 9" MW10. °
. - 0
D6 0 P7
05 I MW12 . p6
. 0
D4 ‘ ,
. 03 I .
I Q 0
02
D1 I
. ° W9 IKdata
0 K and strat data
0 monitoring well
0
ootherwells
-:1
0m 2m

 

 

Figure A]. Location of delivery wells (D), monitoring wells (MW), piezometers (P) and

other wells at the Schoolcraft plume A site.

87

'- .~‘_?i.'

 

 

gravel and medial (Donj eck-type) braided stream deposits are composed of anywhere
between 10 and 90 percent gravel. It can be further reasoned that distal (Platte—type)
braided stream deposits are composed of less than 10 percent gravel.

Along with mean grain size, bedding structures also helped when classifying the
outcrop analogs into proximal, medial and distal locations fiom the glacier. Miall (1977;
1996) stated that proximal braided streams deposits are commonly horizontal gravel
. sheets displaying weak imbrication. Trough cross-bedded gravels and planar cross
bedded gravels are also prominent (Miall 1977; 1996 and Benn and Evans 1996). Medial
deposits can be composed of a variety of facies e. g. massive gravels, trough cross-bedded
sands and gravels, planar cross-bedded sands and gravels, ripple cross-laminated sand
and sand drapes (Miall 1977; 1996 and Benn and Evans 1996). Distal deposits are
composed primarily of trough cross-bedded sands, planar cross-bedded sands and
horizontally bedded sands. Minor gravel lenses and overbank fines may be present
(Miall 1977; 1996 and Benn and Evans 1996).

A gravel pit located approximately 0.5 miles away from the Kalamazoo Moraine
on Ravine Road is composed of greater than 90 percent gravel and coarser material with
the remaining percent composed of sand. The gravel at this site displays cross-bedding
and horizontal bedding with some irnbricated clasts indicating minor channel fills or
bedforrns and channel lag deposits.

Approximately 2 miles fi'om the Kalamazoo moraine on Ravine Road is the
gravel and sand pit interpreted as a medial (Donjeck-type) braided stream deposit. This
sand and gravel pit, was a site of outcrop analog study for this project. The entire section

studied for the outcrop analog was approximately 10 meters in height and generally fined

88

upward from gravels at the base of the section up to medium sand at the top of the section
(Figures B.4- 3.7-Appendix B outcrop analogs). This section had less than 90 percent
gravel and greater than 10 percent gravel classifying it as a medial (Donj eck-type)
deposit. Gravels were cross-bedded to horizontally bedded and some displayed
imbrication suggesting minor channel fills, bedforms, and channel lags as locations of
deposition. Coarse and medium sands were mainly trough cross-bedded suggesting
deposition on linguoid bars, which are common in Donjeck-type streams (Miall, 197 7).

No analogs were found to distal (Platte-type) braided stream deposits in
Kalamazoo County, likely for the reason that it is not as cost effective to have a sand and
gravel pit in distal environments of deposition. This is most likely due to primarily two
factors 1) the cost of gravel is much higher than the cost of sand, so less money will be
made by the sale of sand and 2) the further from the Kalamazoo Moraine, the lower the
elevation and the higher the water table, making excavation of sands difficult and costly
when dewatering the area.

The Schoolcraft Aquifer is located approximately 6 miles away from the
Kalamazoo Moraine at Paw Paw Lake, which is considered to be the outlet of the river
whose sediments compose the Prairie Rhonde Fan and Schoolcraft Aquifer. The bottom
of the aquifer (27.5 to 20.5 m bgs) is dominated by gravel fining upward to medium sand,
suggesting an intermediate location from the glacier or Donjeck-type braided stream.
The medium sands, coarse sands and gravels are cross-stratified with coarser grains
deposited along beds.

The upper part of the aquifer (20.5 to 5 m bgs) is dominated by medium and fine

sand with some very fine sand, silt and gravel. This corresponds well to the Platte type

89

braided stream described by Miall (1977). Further evidence for a Platte type braided
stream is the presence of cross-stratified sands, massively bedded sands and silt. The
cross-stratified sands may have formed in linguoid and transverse bars. The massively
bedded sands may have stratification to fine to see at the core scale, or it may have been
deposited during high discharge when all of the channels generally merge to create a
sheet flow, as described by Krigstrom (1962). The silt described in the core is thought to
have deposited during the last stages of waning flow when deposition of thin silt or mud
drapes and channel fills occur in inactive areas (Miall, 1977; 1996 and Benn and Evans
1998)

Miall (1977) stated that the Donj eck and Platte types may merge subtly into one
another, supporting the degree of difficulty involved in identifying the boundary between
medial and distal outwash facies. Although much uncertainty exists when classifying
outwash into proximal, intermediate and distal zones, it is believed that the upper portion
of the aquifer represents distal outwash (Platte-type) and the bottom portion of the aquifer
represent intermediate outwash (Donj eck-type). This conclusion is based on the mean
grain size, proportions of grain size, types of bedding, thickness of beds, fining upward
cycles and distance from the Kalamazoo Moraine.

Below the outwash, lies a light gray clay that was inferred by Mayotte (1991) to
be laterally extensive beneath the Schoolcraft, thus marking the base of the aquifer for
this study. This is a till unit that occurs beneath the Village of Schoolcraft and acts to
retard the downward migration of plumes A-G (Lipinski, 2002). This till is likely the
Ganges till reported by Monaghan et al. (1986) to crop out at the Tekonsha moraine and

hence underlie sediments between the Tekonsha and Sturgis or Kalamazoo Moraines.

90

The gray till contains small quantities of sand and silt and was reported by Mayotte
(1991) to have been encountered in several test borings at depths ranging from 21.3 to

30.5 m below grade.

91

Key for core descriptions

Beddirg structure symbols and abbreviations

 

 

 

 

Sm Massively bedded sand
\\
g Cross Stratified bedding
‘2'}??3 Faint cross stratified bedding
//
fl

Pebbles or granules along cross stratified bedding

 

Faint laminar bedding

 

 

 

 

 

 

Pebbles
X-S beddigg cross stratified bedding
M bedding massively bedflg
Sorti abbreviations

-sorted

wm-sorted
m-sorted

well sorted
well sorted
well to sorted
sorted
to
soned

to

soned

sorted

sorted
Grain size abbreviations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VF sand veg fine sand

F sand fine sand

M sand medium sand

C sand coarse sand

F-s-c sand fine sand with high silt and clay content

F-m-vf sand fine sand with high medium and very fine sand content
F-msand fine sand with high medium sand content

F-m-c sand fine sand with high medium and coarse sand content
F-m-g sand fine sand with high medium sand aflgranule content
F-m-p sand fine sand with high medium sand and pebble content
F-m-vf-p sand fine sand with high medium, very fine sand and pebble content
M-f sand medium sand high fine sand content

M-f-c sand medium sand with high fine and coarse sand content
M-c sand medium sand witllgigh coarse sand content

M-c-f sand medium sand with high coarse and fine sand content
Mg sand medium sand with high granule content

C-m sand coarse sand with high medium sand content

£_-g sand coarse sand with higflgranule content

C-p:g sand coarse sand with higgpebble and granule content
Pebbles-c sand pebbles with higgcoarse sand content

Other abbreviations

"SKA same as above

HCL hydrochloric acid

 

Table A1. Key for core descriptions.

92

 

Well Name: P-18 Location: Plume A,Schoolcraft,Ml Page: _1_ of _§

Date: 1-5-01 thru 1-12-01 Described by: Susanne E. Biteman

 

 

m Kcm/sec
2N N N m n
(D l . I
g '0 >-DI.u ' uJ ' Comments
= g 984. 3 iii a iii
a) "’ mo" ‘7 " ‘° 32% M sand, vp-sorted, subrounded with pebbles composed

 

of limestone and quartzite, color dry 2.5Y 4/4 olive

brown, coal <1%.

' ,_ M sand, w-sorted. subrounded to rounded with pebbles
composed of quartzite, limestone and clumps of silt
color dry 2. SY 6/4 light yellowish brown, coal 1 -2%.

sand, m- sort ted, subrounded. pebbles increase
downward.

*1 M-c sand, w-sorted, subrounded, X-S bedding, coarser

grains in X-S beds, coal 1%, abrupt change from above.

~ — Gradual change, granules are limestone and quartzite.
~ Pebbles, composed of quartzite, limestone, and chert
vp- sorted, subrounded.

‘ SSA. except pebbles also granite and coal 1-1.5%.

" M-c sand, m-sorted, subrounded to rounded, color dry

_ 2.5Y 7/4 pale yellow.

SAA, except coal present 1%, and subrounded.

\'\ M-c sand with pebbles composed of quartzite, granite,
limestone and dolomite, subrounded, mp—sorted, color
2.5Y 6/4, no coal.

“ Clast supported rounded pebbles composed of granite,
limestone, quartzite and schist, largest pebble 25x32 mm.
\ “ C sand m-sorted, subrounded.

' C-g sand, mw-sorted, subrounded, clasts SAA , color
\ dry 2.5Y 5/2 grayish brown.

M-c sand, m-sorted, subrounded, color dry 2.5Y 6/3
. .. light yellowish brown.

\ P 1 f; Gravel with pebbles, pvp-sorted, subrounded, color SAA.

\ C-p-g sand. m-sorted, rounded to subrounded.
i M-c sand, m-sorted. subrounded.
‘Pebbles composed of clay, silt, quartzite, vp-sorted.

Gravel with sand and pebbles, m-sorted, subrounded.
\\ Pebbles composed of sandstone, granite and quartzite,

vp-sorted, rounded.
f M-c sand m-sorted, subrounded, color SAA.
“ M sand vw-sorted, subrounded, color SAA.

 

Depth (m) a:
/
/ /

V

 

\
\ M-g sand, w-sorted, subrounded, X-S bedding, color dry
‘ 2.5 Y 6/3, coal 1-1.5%.
M sand vw-sorted, X-S bedding with coal along
x-beds, coal 2%, color SAA.
“ M sand w-sorted, subrounded, color SAA, coal 1-2%.
‘ SAA, except v. well sorted.
\ SAA, except coal 15.-2.5%.
‘ SAA, except color wet 2. SY 5/3 light olive brown, dry 6/3
and coal -2 %
‘ SAA, except coal 0.5- 1.5%.

* L M sand vw-to w- ~sorted, iubrounded, M bedded, color
dry 2. SY 6/3, coalO

i M-c sand, w-sorted, subrounded M bedding, color dry 2. SY
7/3, coal 1 .5.-2 5%

\

 

 

 

 

 

 

Figure A.2.a. Core description of well P18.

93

Well Name: P-18 Location: Plume A,Schooleraft,Ml Page: 2 of 5

Date: 1-12-01l 7-19-01 thru 8-12-01

 

    

 

Described by: Susanne E. Biteman

 

    

 

Depth (m)

 

K (cm/sec)
% “J “I‘ ‘1' ‘1’ "P Comments
3. Lu LU LlJ UJ L”
m V. °°. N O. 10.
T) N ‘— 4- (D 1—
\j / IL 114% M-c sand,w—sorted, subrounded, X—S bedding, color wet 2.5Y
‘ 6/3 light yellowish brown, dry 2.5Y 7/3 pale yellow, coal
_\7, _ \ concentrated along beds.
“\\w—F sand, vw-sorted, subrounded, X-S bedding, coal
\H concentrated along X-S bedding, coal 2-4%, color SAA.
’\\ F‘ 7\_-¥ SAA except color, color wet 5Y 5/2 olive gray, dry 5Y
\\\ 6.5/2 light gray to light olive gray.

\ SAA except color, color wet 2.5Y 5/4 light olive brown,
dry 6/3.

/ F sand with larger grains, w-sorted, subangular, X-S
bedding ,thickness between beds 3-15mm. Color wet
2.5Y 5/4 light olive brown, dry 6/3 light yellowish brown.
,. F sand, vw-sorted, subangular, thin laminations. Color
wet 2.5Y 5/4 light olive brown, dry 2.5Y 7/3 light
yellowish brown.

 

 

 

 

F sand, vw sorted, subrounded, X-S bedding darker
along beds. Color SAA
“ F-m-vf sand, vw-sorted, subrounded, X-S bedding (3-
12 mm thick) darker grains and finer grains
concentrated along beds. Color SAA
\ F-vf-s-sand, w-sorted, subangular, X-S bedding (2—15
mm thick). Color SAA

‘\‘ F sand. vw-sorted, subangular, X-S bedding (2-10 mm

thick), darker grains concentrated along beds, color
SAA

\\ F-m-s sand with pebble size grains at base, w-sorted,

subrounded, X-S bedding (0.5-3 mm thick). Pebbles
occur alon beds and are composed of calcium
carbonate HCL reaction) cementing sand grains
together. Color SAA

\ F-vf-s-m sand, vw-sorted, subrounded, X-S bedding

(0.5-2 mm thick). Color wet SAA

\ F-m sand, some cobbles and pebbles wm-sorted,
subrounded, X-S beddding 1mm thick 3mm apart,
color saa.

F sand, w-sorted, subrounded, bedding and color SAA.

SAA. alternating dark and light beds

\ F sand, w-sorted, subrounded, X-S bedding, 1-4 mm

thick, 2-5 mm apart, coal 2-3 %, color saa.

\ F sand, w—sorted, subrounded, X-S bedding, 1-3 mm

thick, 1-3 mm apart, color saa.

\ Color wet 2.5Y 5/2

\ Color wet 2.5Y 5/4. dry 2.5Y 7/3

~ F-m-vf sand, w-sorted, subrounded. faint X-S bedding,
color 2.5Y 5/4, dry 7/3

 

Figure A.2.b. Core description of well P18. Vertical K measurements shown in second column.

94

Well Name: P-18 Location: Plume A,Schoolcraft,Ml Page: 3 of 5

 

Date: 7-23-01 thru 8-17-01

 

 

  

 

Described by: Susanne E. Biteman

Comments

 

 

 

 

 

—l

 

 

 

//./:e F m- vf sand w- -,sorted subrounded, color wet 2. SY 5/4,

gr)!t light olive brown, dry 7/3 pale yellow, X- S bedding

aInt.

~SAA, beds 1-2 mm apart, 2-4 mm thick.

_ F-m sand, w-sorted, subrounded, thin X-S bedding

(0.5-2 mm thick, 1 to 9 mm apart) color SAA.

“SAA

\ SAA, except coal 1-2%

~~ F- m- -p sand, p- -sorted, subrounded, X- S bedding 0. 5-1
mm thick, 7mm apartdarker grains and iron oxides are
concentrated along beds, pebbles are sandcemented
with calcium carbonate (HCL reaction), color SAA

F- -m sand w-sorted, subrounded, X-S bedding,1%
coal, color S

\ F-m-vf sand, w-sorted subrounded, thin X-S bedding

(0.5-2 mm thick), color SAA

‘ SAA, except, M bedding

‘ SAA, except, X-S bedding, 4-5 mm thick, 1 mm apart

\ SAA, except X-S bedding, 0.5 thick. 2 mm apart

SAA, except X-S bedding faint

F- -m- -vf sand, w-sorled, subround dd

F- m- -vf-p sand, moderately sorted, faint X- S bedding,
color SAA

\ M- -c- f--sand, w-sorted, subrounded, faint X- S bedding,
color wet 2. SY 5/4 light olive brown, dry 6/3l ight
yellowish brown, coal fragments %

F m-g sand, wm-sorted, subrounded, color wet 2.5Y
5/4, dry 7/3, faint X- S- -bedding.

M- sand with some granules wm- -sorted, subrouinded,
coal wet 2. SY 5/4, dry 6/3 coal 1 %, X- Sb
M-sand, m- sorted, rounded, color wet 2. 5Ye 4/4n o ive

\\ brown, dry 7/3, weak X- S bedding

\ Pebbles-c sand, vp-sorted. pebbles rounded, sand

subrounded, color SAA
M-f-c sand, vw-sorted, subrounded, coarser grains

along X-S bedding contact at base of X-S bed, F sand
beneath contact, color 2.5Y 5/4, dry 7/3.

‘F-m sand, vw-sorted, subrounded, M bedding, color
wet 2.5Y 5/4, dry 7/4 pale yellow.

\ M sand, w-sorted, subrounded, faint X-S bedding,
color wet 2.5Y 4/4, dry 6/4 light yellowish brown,

F sand, w-sorted, subrounded, M bedding, color wet,
2.5Y 5/4, dry 7/3
. M c sand, p- -sorted, rosunded, M bedded, color wet,
\ 2. 5Y 4/3 (color, dS/4rg
M sand, w- -sorte, su rounded, X- S bedding, 1% coal
M sand, w-sorted, subrounded, faint X- S bedding,
appears v. homogeneous, color wet, 2.5Y 5/4, dry 7 3
M sand, w-sorted, subrounded, M bedding

\ SAA, except F-m sand, color wet 2.5Y 5/4, dry 7/3

SAA, except M sand

 

Figure A.2.c. Core description of well P18. Vertical K measurements shown in second column.

95

Well Name: P-18 Location: _P_l_ume A, SchoolcraftI MI Page: 4 of 5

Date: 7-31-01 thru 8-22-01 Described by: Susanne E. Biteman

 

 

 

 

 

 

[ K cmlsec I
—3 N N 0?
'8 ggE u', u‘, m Comments
5 to Q ‘1 “2 N O.
a: a) 0,8 1- 4- 0 '-
Sm F-m sand, w-sorted, subrounded, appears M bedded,

color dry, 2.5Y 7/3

a M-f sand, w-sorted, subrounded, X-S bedding, color wet,
2.5Y5/4 light olive brown, dry 7/4 pale yellow

\ F sand, w-sorted, subrounded, color wet 2.5Y 5/4

19.. \ F sand, w-sorted, subrounded, M bedding, color wet,
2.5Y 4/4 olive brown, dry 6/3 light yellowish brown,

\ M sand, w-sorted, subrounded, M bedding, color wet,
2.5Y 5/4, dry 7/3

\ SAA, except M—c sand
\ SAA, except, M-sand, one cobble 25x25x12 mm

, MS-f sand, w-sorted, subrounded, faint X-S bedding, color
AA.

M-f sand, w-sorted, subrounded, M-bedded or faint X-S
bedding, color SAA

F-s—c sand, w-sorted, subrounded, coal 1-2 %, color wet
2.5Y 5/4, dry 7/4 pale yellow.

* SAA, except fine laminations
_ M sand, w-sorted, subrounded, M bedded, color SAA
* SAA, except color wet 2.5Y 5/4, dry 7/3

c M-f—c sand, w-sorted, subrounded, M bedded, color wet
2.5Y 5/3 (color), dry 7/3

 

 

 

 

Depth (m)

 

 

 

N
_x

 

P M-f-c sand, w-sorted, subrounded, M bedding, color wet,
2.5Y 5/4, dry 7/3

 

SAA, except rounded grains

.. SAA, except subrounded grains

— SAA except rounded grains

— M-c sand, wm-sorted, rounded, M bedding, color wet,
2.5Y 4/4 (olive brown7), dry 6/3 (color)

, x; 3', SAA, except faint X-S bedding

Sm

 

 

 

 

 

 

Figure A.9.d. Core description of well P18. Vertical K measurements shown in second column.

96

Well Name: P-18 Location: Plume A, Schoolcraft, MI Page: 5 of 5
Date: 7-18-01, 8-2_2l-01 thru 8-24-01 Described by: Susanne E. _Bitemg

K cmlsec

Comments

M-f-c sand, wm-sorted, subrounded, faint X-S bedding
at base, top is M bedded

M—c sand, w-sorted, subrounded, faint X-S bedding at
base, top is M bedded, color wet 2.5Y 5/4 light olive
brown, dry 7/3 pale yellow

M-c sand, w-sorted, rounded, M bedding,

C-m sand, m-sorted, subrounded, X-S bedding (1-1.5
mm apart), color SAA

C-m sand, wm-sorted, subrounded-rounded, coarser
grains along X-S beds, color SAA

M—c sand, w—sorted, subrounded, faint X-S bedding
SAA, except M bedding

23

N
A

F-m sand, w-sorted, rounded, M-bedding, color SAA

M-c-f sand, m-sorted, rounded, faint X-S bedding, color
wet 2.5Y 4/4 olive brown?), dry 7/3

C-m sand, m-sorted, rounded, coarser grains along X-S
bedding, fines upward, bed thickness 5-10 cm

C-m sand, ,p-sorted, rounded, faint X-S bedding, color
wet 2. SY 64 (color?), dry 5/6 (color), iron oxide present

C-m sand, m-sorted, rounded, M-bedded with randomly
placed pebbles, color wet 2.5Y 4/4, dry 7/4 (color)

SAA, except faint X-S bedding

M-c sand, m-sorted, rounded, no bedding preserved,
color dry 2.5Y 6/3

M-c sand, p-sorted, rounded, large pebbles along X-S
bedding, color wet 2.5Y 4/4, dry 6/3

M-c sand, p-sorted, rounded, faint X-S bedding, color
wet 2.5Y 5/4, dry 7/4

Depth (In)

M-c sand, m-sorted, rounded, M bedded, color wet 2.5Y
5/4, dry 6/4

SAA, except M-c sand

Predominatly pebble sized grains with cobbles and
sand, vp-sorted, rounded, X-S bedding thin 0.5-1 cm
thick.

Cobble (6 cm x 4.5 cm) with pebbles and finer grains,
vp-sorted, rounded, no bedding preserved

26

27

 

Figure A.9.e. Core description of well P18. Vertical K measurements shown in second column.

97

Well Name: E Lomtion: Plume A, Schoolcraft, Ml Page: _1_ of2
Date: 1-3-01 thru 1-4-01 Described by: Susaflie E. Biteman

Comments

11 F sand, vw-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4

light olive brown, dry 2.5Y 7/3 pale yellow, coal 2-3 %.

F sand, vw-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4,
coal 2-3 "/0.

F sand, vw-sorted, subrounded, X-S bedding, color wet 2.5Y 6/3
light yellowish brown, dry 2.5Y 7/3 pale yellow, coal 2-3 %.

SAA, except color wet 2.5Y 5/4, dry 2.5Y 7/4 pale yellow.

F sand, vw sorted, coal 3-4 %. Core was dry and loosely packed,
no bedding was preserved.

Depth (m)

—L

F sand, vw-sorted, subrounded, X—S bedding, color wet 2.5Y 5/4,
dry 2.5Y 6/3, coal 2-3 %.

14 SAA

SAA

 

Figure A.3.a. Core description of well P6.

98

 

Well Name: l3 Location: Plume A, Schoolcrgft, Ml Page: _2_ ofg
Date: 1-3-01 thru 1-4-01 Described by: Susa_nne E. Biteman

Comments

15 F sand, vw-sorted, subrounded, X-S bedding, color dry 2.5Y 7/4

pale yellow.

M sand, w-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4 light
olive brown, dry 2.5Y 6/4 light yellowish brown, coal 0.5%.

Depth (m)

18

 

Figure A.3.b. Core description of well P6.

99

Well Name: FIB-7 Location: Plume A, Schoolcraft, Ml Page: _1_ of;
Date: 12-29—00 Described by: Susanneig. Bitema_r1

Comments

M sand, w-sorted, rounded, M-bedding, color wet 2.5Y 5/3 light
olive brown, wet 2.5Y 7/3 pale yellow, coal 3 %.

10 F sand, vw-sorted, subrounded, X-S bedding, color SAA, coal 3 %.

F sand, w-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4, dry
2.5Y 7/4 pale yellow, coal 3 %.

11 F sand, vw-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4,
dry 2.5Y 6/3 light yellowish brown, coal 2-3 %.
E
I: F sand, vw-sorted, subrounded, X-S bedding, color wet 2.5Y 4/4
5 olive brown, dry 2.5Y 7/3, coal 2-3 %.
o
a
12 F sand, w-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4
coal 2-3 %. Gravel is poorly sorted.
M sand, w-sorted, subrounded-rounded, M bedding, color wet 2.5Y
5/4 olive brown, dry 2.5Y 7/3, coal 1 %.
13 F sand, vw-sorted, subrounded, X-S bedding, color wet 2.5Y 4/4,

dry 2.5Y 6/4 light yellowish brown, coal 2-3 %.

 

Figure A.4.a. Core description of well P7.

100

Well Name: PlB-7 Location: Plume A, Schoolcraft, Ml Page: 3 of_2_
Date: 12-29-01 thru 1-3-01 Described by: Susanne E. Biteman

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

in
u '52 Comments
I: > 1°
5 III ‘2 1°
(0 to 0:8
- :7 F sand, vw-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4
:;—// light olive brown, dry 2.5Y 7/3 pale yellow, coal 2 % coal
_/ concentrated along beds.

14 — F sand, vw-sorted, subrounded, X-S bedding, color wet 2.5Y 4/4
% olive brown, dry 2.5Y 6/4 light yellowish brown, coal 2 %, coal
§ concentrated along beds.

.1
- / F sand, vw-sorted, rounded-subrounded, X-S bedding, color SAA
Z coal 2-3 %, coal concentrated along beds.
15“ F sand, vw-sorted, subrounded, color wet 5/4, dry 6/3.
' SAA, except grain size is M-sand
E
5 —I
o.
o
o
1 6 -
_l
1 7-

Figure A.4.b. Core description of well P7.

101

Well Name: PlB-8 Location: Plume/L Schoolcraft, Ml Page: _1 of §
Date: 12-27-00 thru 12128-00 Described by: Susanne l_E. Biteman

Comments

Pebbles and cobbles, rounded, composed of sandstone and granite
C-sand, poorly sorted, subrounded, color wet 2.5 Y 5/4 light olive
brown, dry 2.5Y 6/4 light yellowish brown.

F-sand, vw-sorted, subangular to subrounded, X-S bedding, color
wet 2.5Y 5/4, dry 2.5Y 6/3 light yellowish brown, coal 3-4 %.

F-sand, w-sorted, subangular, X-S bedding, color wet 2.5Y 5/4, dry
2.5Y 7/4 pale yellow, coal 3 %.

F-sand, w-sorted, subrounded to subangular, X-S bedding, color
wet 2.5Y 5/4, dry 2.5Y 6/3 light yellowish brown, coal 2-3 %.

F-sand, w-sorted, subangular to subrounded, X-S bedding, color
wet 2.5Y 5/3 light olive brown, dry 2.5Y 7/3 pale yellow, coal 2-3 %
coal concentrated along beds.

Depth (m)

F-sand, w-sorted, subrounded to suban ular, X-S bedding, color
wet 2.5Y 5/3 light olive brown, dry 2.5Y 7 3 pale yellow, coal 2-3 %
coal concentrated along beds.

F-sand, w-sorted, subrounded to subangular, X-S bedding, color
wet 2.5Y 5/4, dry 2.5Y 7/3, coal 3—4 %, coal concentrated along
beds.

F-sand, w-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4,
coal 3-4 %.

 

Figure A.5.a. Core description of well P8.

102

Well Name: FIB-8 Location:
Date: 12-27-00 thru 12-28—00 Described by:

 

Silt

Sand

 

gravel

Icobbles

 

 

14

W

 

 

 

\W/

 

 

 

Depth (m)

16..

17-

Hill!
l|\\\l
lIH‘I

 

 

 

Plume A, Schoolcraft, Ml Page: 3 of §

Susgnne E. Biteman

Comments

F-sand, w-sorted, subrounded, X-S bedding, color dry 2.5Y 7/4
pale yellow.

F-sand, w-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4

light olive brown, dry 2.5Y 7/4, coal 4 %, coal concentrated along
beds.

F-sand, w-sorted, subrounded, faint X-S bedding, color wet 2.5Y
4/4 olive brown, dry 2.5Y 7/3 pale yellow, coal 4 %.

Figure A.5.b. Core description of well P8.

103

Well Name: PlB-8 Location: Plume A, Schoolcra_ft, Ml Page: _3 of Q
Date: 12-28-00 thrufl-ZB-OO Described by: Susagne E. Biteman

 

 

 

 

 

 

 

in
1, E’ g Comments
5 5 e P
In a) m 8
16 -
‘5‘1 7 .-
5 l
a
o
D
:1”; M-sand, w to vw-sorted, subrounded to rounded, X-S bedding, color
/< wet 2.5Y 5/3 light olive brown, dry 2.5Y 7/3 pale yellow, coal 1-2 %.
Q
18 -

 

 

Figure A.5.c. Core description of well P8.

104

Well Name: PlB-8 Location: Plume A, Schoolcraft, Ml Page: _4_on

Date: 12-28-00 thru 12_-28-00 Described by: Susanne E. Biteman

 

 

 

 

 

 

 

in
E 9% Comments
E III 91°
‘0 Oi
19 -
q
20 -
_—-:::: F-m sand, w-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4
‘__../< light olive brown, dry 2.5Y 7/3 pale yellow, coal 1-2 %.
E \\
E
o.
o _
o 21
22-
M sand, moderately sorted, subrounded to rounded, faint X-S
‘ , bedding, color wet 2.5Y 5/4 light olive brown, dry 2.5Y 7/3, coal
5 5 , : : 1-2 %.

 

 

Figure A.5.d. Core description of well P8.

105

Well Name: PIE-8 Location: Plume A, SchoolcraflLhfl Page: _5_ of 6
Date: 12-29-00 thru 1;;9-00 Described by: Susanne E. Biteman

 

 

 

 

E 13% Comments

55' III 91°
0) a) m§

23-

244

E _

.C

‘5.

0 ..a

o

25-

26-

 

 

Figure A.5.a. Core description of well P8.

106

Well Name: PlB-8 Location: Plume A, Schoolcraft, Ml Page: _6_ of §
Date: 12-29-00 thru 12-29-00 Described by: Susanne E. Bitemgnfi

 

Silt

 

Sand

 

gravel
[cobbles

 

28-

Depth (In)

30—

 

 

 

Comments

Predominantly pebbles with granules, C sand and M sand, vp-sorted
rounded, faint X-S bedding, color wet 2.5Y 4/3 olive brown, dry 2.5Y
7/2 light gray, coal <1%. Pebbles composed of chert, limestone,
quartzite, and granite.

Figure A.5.f. Core description of well P8.

107

Well Name: PlB-9 Location: Plume A, Schoolcrgfl,Ml Page: _1 of 2
Date: _1_2-19-00 thru 12-21-00 Described by: Susanne g. Bitemgg

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_—

 

 

 

 

 

 

—8
m—
:._- E 58
a) a) a8)
§
10‘
\\
\
112%
//
E ___,.
5 _\
a \____
O
a
12 ______._
:::
£4:a
Sm
13 //
//\
42:.

 

 

 

Comments

—— M-c sand, w-sorted, subrounded, color wet 2.5Y 5/4 light olive

brown, dry 2.5Y 6/3 light yellowish brown.

— M-c sand with a large amount of cobbles and pebbles, m-sorted,

subrounded, color wet 2.5Y 4/4 olive brown, dry 2.5Y 5/3 light olive
rown.

M sand, w-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4
dry 2.5Y 7/4 pale yellow.

F sand, vw-sorted, subrounded, faint X—S bedding color wet 2.5Y
5/4, dry 2.5Y 7/3 pale yellow.

SAA

F sand, w-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4, dry
2.5Y 6/3, coal 1 %.

F sand, vw-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4,
coal 1 %.

Pebbles are composed of sand cemented by calcium carbonate
(HCL acid reaction).

F sand with gravel size clasts composed of sand cemented by
calcium carbonate, vp-sorted, subrounded, color wet 2.5Y 4/4, dry
2.5Y 6/4 light yellowish brown.

F sand with pebbles, p-sorted, subrounded, X-S bedding, color wet
2.5Y 5/4, dry 2.5Y 6/4, coal 3 %, coal concentrated along beds.

Figure A.6.a. Core description of well P9.

108

 

Well Name: FIB-9 Location: flgme A, Schoolcraft, Ml Page: _2 of_2_
Date: 12-21-00 thru 12-27-00 Described by: Susanne LBitemgg

 

Comments

F sand, vw-sorted, subrounded, X-S bedding, color wet 2.5Y 5/4

light olive brown, dry 2.5Y 6/4 light yellowish brown, coal 2-3 %. coal
concentrated along beds.

14

F sand, w to vw-sorted, subrounded, X-S bedding, color wet 2.5Y

4/6 dark yellowish brown, dry 2.5Y 6/4, coal 3 %, coal concentrated
along beds.

F sand, w-sorted, subrounded, faint X-S bedding, color wet 2.5Y
4/4 olive brown, dry 2.5Y 7/4 pale yellow, coal 2-3 %.

 

_s
01

Depth (m)

17

Figure A.6.b. Core description of well P9.

109

Well Name: PlB—11 Location: Plume A, Schoolcraft, MI Page: _1 of g
Date: _1_2-15-00 thru ‘lj2-18-00 Described by: Susanne E. Biteman

 

 

 

 

 

 

 

0)
1, ‘52 Comments
c >10
E «I 9‘3
co 0) 008
q
M sand, w-sorted, rounded to subrounded, color wet 2.5Y 4/3
olive brown, dry 2.5Y 7/4 pale yellow.
—.I
20 a
21-
E ..
.c
a
o.
o ..
O
22—
23-

 

 

Figure A.7.a. Core description of well P11.

110

Well Name: PlB-11 Location: Plume A, Schoolcraft, Ml Page: 3 ofg

Date: 41-15-00 thru 12-18-00 Described by: Susanne E. Biteman

 

 

 

 

 

 

 

in

1, 62 Comments
c > 1°
5' III 8 1°
(0 (0 0:8

M sand with larger clasts, w-sorted, subrounded, color wet 2.5Y 5/4
- light olive brown, dry 2.5Y 6/3 light yellowish brown.
24 _ 111'], - M-c sand with pebble size clasts, w-sorted, subrounded, faint X-S

312:1? bedding, color wet 2.5Y 4/3 olive brown, dry 2.5Y 6/3.

I
_—
_——_.

 

 

Depth (m)
re
01

n I

26-

 

 

Figure A.7.b. Core description of well P11.

111

 

Well Name: PIE-12 Location: Plume A, SchoolcraftI Ml Page: _1_ of_4_
Date: 12-18-00 thru 12-19-00 Described by: Susanne E. Biteman

Comments

13

Pm sand, vw-sorted, subrounded to rounded, X-S bedding, color
wet 2.5Y 5/4 light olive brown, dry 2.5Y 7/4 pale yellow.

14

Depth (m)

16

 

Figure A.8.a. Core description of well P12.

112

Well Name: PlB-12 Location: Plume A, Schoolcraft, Ml Page: _2_ of 5
Date: _g—18-00 thru 12-19—00 Described by: Susgnne E. Biteman

 

 

 

1, 0% Comments
5 5 ED
0) (0 0:8
17-
184
E
.r: ._
‘6.
0
o
1911
20"
.J

 

 

Figure A.8.b. Core description of well P12.

113

Well Name: PlB-12 Location: PlumeA,Schoolcraft, Ml Page: ioffl

Date: _1_2-18—00thru1g-19-00 Described by: Susanne g. Biteman

 

 

 

 

 

7 ll)
1, 9.3 Comments
:5 5 9°
(D a) 0%
21 1
22 -
..... F-m sand, w-sorted, subrounded to rounded, faint X-S bedding,
_ : ff; : color wet 2.5Y 4/3 olive brown, dry 2.5Y 7/3 pale yellow.
r~--:
E Z : : ~ 2
I ._
a
a
o
a
23—
24 _

 

 

Figure A.8.c. Core description of well P12.

114

Well Name: FIB—12 Location:
Date: 12-18-00 thru 1&19'00 Described by:

 

 

 

 

3

E 93

E III 91°

to (0 0:8

24 ‘
q
25-

Depth (m)
N
‘i’

 

27"

 

 

Plume A, SchoolcrafL Ml Page: i of g

Susanne E. Biteman

Comments

Predominantly pebble, grain size ranges from M sand to cobbles,
p-sorted, sand is subrounded, pebbles and cobbles are rounded,

faint X-S bedding, color wet 2.5Y 4/3 olive brown, dry 2.5Y 6/2
olive gray.

Figure A.8.d. Core description of well P12.

115

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

X Y Z LOG(K) (cmls) K (cmls) mean grain size sortirL
38.7 22.5 2.0 -1.49E+00 3.26E-02 v. fine sand v-poody
38.7 22.5 2.1 -1.18E+00 6.61 E-02
38.7 22.5 2.3 -1.05E+00 8.88E-02
38.7 22.5 2.5 -1.21E+00 6.20E-02 coarse sand poorly
38.7 22.5 2.9 -1.35E+00 4.42E-02
38.7 22.5 3.2 -1.64E+00 2.27E-02
38.7 22.5 3.4 -1.01E+00 9.69E-02
38.7 22.5 3.6 -1.16E+00 6.98E-02
38.7 22.5 3.8 -9.76E-01 1.06E-01
38.7 22.5 4.4 -1.57E+00 2.71 E-02
38.7 22.5 4.9 -1.17E+00 6.75E-02
38.7 22.5 5.1 -1.73E+00 1.86E-02 very fine sand moderately
38.7 22.5 5.3 -1.69E+00 2.04E-02
38.7 22.5 5.6 -1.5SE+00 2.79E-02
38.7 22.5 5.9 -1 .55E+00 2.79E-02
38.7 22.5 7.2 -1 .56E+00 2.75E-02
38.7 22.5 7.4 -1.52E+00 3.02E-02
38.7 22.5 7.8 -1 .42E+00 3.84E-02
38.7 22.5 8.2 -1.44E+00 3.59E-02
38.7 22.5 8.4 -1.41E+00 3.90E-02 med sand moderately
38.7 22.5 8.6 -1 .27E-I-00 5.38E-02
38.7 22.5 9.3 -1.46E+00 3.47E-02
38.7 22.5 9.6 -1.57E+00 2.67E-02
38.7 22.5 9.9 -1.20E+00 6.28E-02
38.7 22.5 10.1 -1.36E+00 4.37E-02 coarse sand v-poorly
38.7 22.5 10.5 -1.53E+00 2.98E-02
38.7 22.5 11.0 -1.43E+00 3.76E-02
38.7 22.5 11.2 -1.41E+00 3.85E-02
38.7 22.5 11.4 -1.35E+00 4.51 E02
38.7 22.5 11.8 -1.90E+00 1.26E-02
38.7 22.5 12.0 -1.84E+00 1.43E—02 fine sand m-well
38.7 22.5 12.3 -1.98E+00 1.04E-02
38.7 22.5 12.9 -1 .90E+00 1.25E-02
38.7 22.5 13.9 -1.94E+00 1 .15E-02
38.7 22.5 14.1 -1.88E+00 1.31 E-02 fine sand well
38.7 22.5 14.4 -1.88E+00 1.31E-02
38.7 22.5 14.7 -1.86E+00 1.38E-02 fine sand well
38.7 22.5 15.5 -1.92E+00 1.20E-02
38.7 22.5 15.9 -2.07E+00 8.60E-03
38.7 22.5 16.4 -1.85E+00 1.41 E-02
38.7 22.5 16.7 -2.05E+00 8.90E-03
38.7 22.5 17.2 -1.97E+00 1.06E-02 v. fine sand well
38.7 22.5 17.6 -1.96E+00 1.09E-02 v. fine sand well
38.7 22.5 18.0 -1 .4ZE+00 3.84E-02

 

 

 

 

 

 

 

Table A2. Well D2. X, Y and 2 location, Log K value, K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain size (Boggs, 2001)).

116

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

X Y Z LOG(K) (cmls) K (cmls) mean ggin size sortirg
38.8 24.4 3.5 -1.14E+00 7.24E-02
38.8 24.4 3.9 -1.37E+00 4.23E-02
38.8 24.4 4.3 -1.52E+00 3.00E-02
38.8 24.4 5.0 -1.34E+00 4.59E-02
38.8 24.4 5.4 -1.73E+00 1.86E-02
38.8 24.4 5.6 -1 .66E+00 2.20E-02
38.8 24.4 7.3 -1.54E+00 2.90E-02 fine sand well
38.8 24.4 8.1 -1.61E+00 2.46E-02 med sand m-well
38.8 24.4 8.5 -1.55E+00 2.83E-02
38.8 24.4 8.9 -1 .72E+00 1 .89E-02
38.8 24.4 10.9 -1.58E+00 2.64E-02
38.8 24.4 11.3 -1.53E+00 2.98E-02
38.8 24.4 11.7 -1.98E+00 1.04E-02 fine sand moderately
38.8 24.4 1 1.9 -1 .88E+00 1.32E-02
38.8 24.4 13.0 -1.90E+00 1.26E-02
38.8 24.4 13.4 -1.99E+00 1.03E-02
38.8 24.4 14.0 -2.09E+00 8.20E-03 fine sand m-well
38.8 24.4 14.4 -1 .80E+00 1.60E-02
38.8 24.4 14.7 -1.80E+00 1.60E-02
38.8 24.4 15.1 -1.88E+00 1.32E-02
38.8 24.4 16.3 -1.97E+00 1.07E-02 v. fine sand well
38.8 24.4 17.0 -2.37E+00 4.30E-03
38.8 24.4 17.4 -1.89E+00 1 .29E-02
38.8 24.4 17.8 -2.13E+00 7.40E-03 v. fine sand well
38.8 24.4 18.0 -1.81E+00 1.56E-02 fine sand moderately

 

 

 

 

 

 

 

Table A3. Well D4. X, Y and 2 location, Log K value, K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain size (Boggs, 2001)).

117

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

X Y Z LOG(K) (cm/s) K (cmls) mean grain size sorting
38.8 26.4 0.6 -1 .07E+00 8.59E-02 granules v-poorly
38.8 26.4 1 .0 -1.18E+00 6.62E-02
38.8 26.4 1.3 -1.04E+00 9.02E-02 v. fine sand v-poorly
38.8 26.4 1.5 -1.59E+00 2.55E-02 med sand poorly
38.8 26.4 3.9 -1 .20E+00 6.38E-02
38.8 26.4 4.2 -9.91E-01 1.02E-01
38.8 26.4 4.5 -1.29E+00 5.10E-02
38.8 26.4 4.7 -1.18E+00 6.59E-02 med sand moderately
38.8 26.4 5.0 -1.21E+00 6.19E-02
38.8 26.4 5.4 -1.66E+00 2.17E-02 fine sand moderately
38.8 26.4 5.9 -1.60E+00 2.54E-02
38.8 26.4 7.3 -1 .52E+00 3.03E-02
38.8 26.4 7.7 -1.53E+00 2.93E-02
38.8 26.4 8.1 -1.59E+00 2.59E-02 fine sand moderately
38.8 26.4 8.5 -1.61E+00 2.47E-02
38.8 26.4 9.4 -1.49E+00 3.22E-02 fine sand m-well
38.8 26.4 9.8 -1.55E+00 2.79E-02 med sand moderately
38.8 26.4 10.2 -1.40E+00 3.95E-02 med sand poorly
38.8 26.4 11.0 -1 .52E+00 3.05E-02
38.8 26.4 11.3 -1.50E+00 3.18E-02
38.8 26.4 11.7 -1.63E+00 2.33E-02
38.8 26.4 12.1 -1 .83E+00 1 .48E-02
38.8 26.4 13.0 -1 .89E+00 1.28E-02
38.8 26.4 13.4 -1.84E+00 1.44E-02
38.8 26.4 13.8 -2.14E+00 7.20E-03
38.8 26.4 14.2 -1.91E+00 1.22E-02
38.8 26.4 14.6 -1.81E+00 1.55E-02
38.8 26.4 15.0 -1 .88E+00 1 .33E-02
38.8 26.4 15.8 -2.04E+00 9.10E-03
38.8 26.4 16.2 -1.90E+00 1 255-02
38.8 26.4 16.6 -2.14E+00 7.30E-03
38.8 26.4 17.0 -2.10E+00 7.90E-03
38.8 26.4 17.4 -2.04E+00 9.10E-03
38.8 26.4 17.8 -2.10E+00 8.00E-03
38.8 26.4 18.1 -1.77E+00 1.70E—02 coarse sand v-poorly

 

 

 

 

 

 

 

Table A.4. Well 06. X, Y and 2 location, Log K value, K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain size (Boggs, 2001)).

118

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

X Y Z LOGQQ (cm/s flemls) meangrain size sorting
38.8 28.4 1 .0 -1 .46E+00 3.46E-02
38.8 28.4 1.4 -1 .23E+00 5.91 E02
38.8 28.4 1.8 -1.23E+00 5.90E-02
38.8 28.4 2.2 -1.16E+00 6.87E-02
38.8 28.4 2.6 -1.18E+00 6.61E-02
38.8 28.4 3.0 -1 .45E+00 3.55E-02
38.8 28.4 4.1 -1 .63E+00 2.35E-02
38.8 28.4 4.5 -1.45E+00 3.57E-02
38.8 28.4 5.1 -1 .73E+00 1 .85E-02
38.8 28.4 5.7 -1 .41E+00 3.92E-02
38.8 28.4 7.1 -2.96E+00 1.10E-03
38.8 28.4 8.3 -1 .77E+00 1.70E-02
38.8 28.4 8.7 -1.52E+00 3.03E-02 med sand moderately
38.8 28.4 10.2 -1 .83E+00 1 .48E-02
38.8 28.4 10.6 -1.94E+00 1.14E-02
38.8 28.4 11.1 -1.50E+00 3135-02
38.8 28.4 11.6 -1.71E+00 1.96E-02
38.8 28.4 14.6 -1.79E+00 1.61 E02 fine sand well
38.8 28.4 14.9 -2.00E+00 9.90E-03 fine sand m-well
38.8 28.4 15.8 -1 .98E+00 1.04E-02
38.8 28.4 16.2 —2.39E+00 4.04E-03
38.8 28.4 16.6 -1 .49E+00 3.20E-02

 

 

 

 

 

 

 

Table A5. Well D8. X, Y and 2 location, Log K value, K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain size (Boggs, 2001)).

119

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

X Y Z LOG-(K) (cmls) K (cmls) mean grain size sorting___l
38.7 30.4 0.9 -1 .34E+00 4.60E-02 granules v-poorlL
38.7 30.4 1.0 -1.29E+00 5.12E-02 pebble poorly
38.7 30.4 1 .2 -9.87E-01 1 .03E-01 pebble v-poorly
38.7 30.4 3.3 -1.54E+00 2.87E-02 coarse sand v-poorly
38.7 30.4 3.7 -1.18E+00 6.64E-02
38.7 30.4 4.1 -1.18E+00 6.62E-02
38.7 30.4 5.3 -1.34E+00 4.56E-02 med sand poorly
38.7 30.4 5.8 -1.71E+00 1.93E-02
38.7 30.4 6.5 -2.68E+00 2.10E-03
38.7 30.4 6.9 -2.60E+00 2.50E-03 fine sand moderater
38.7 30.4 7.3 -2.20E+00 6.30E-03
38.7 30.4 7.9 -1.59E+00 2.59E-02
38.7 30.4 8.3 -1.55E+00 2.82E-02 med sand m-well
38.7 30.4 8.7 -1.59E+00 2.56E-02
38.7 30.4 9.0 -1 .83E+00 1 .49E-02
38.7 30.4 9.8 -1.55E+00 2.82E-02 fine sand well
38.7 30.4 10.2 -1.51E+00 3.11E-02 med sand poorly
38.7 30.4 11.3 -1.53E+00 2.98E-02 med sand moderately
38.7 30.4 11.7 -1.53E+00 2.96E-02
38.7 30.4 12.1 -1.72E+00 1.89E-02
38.7 30.4 13.4 -2.00E+00 1 .01E-02 fine sand m-well
38.7 30.4 14.2 -1.96E+00 1.10E-02 fine sand m-well
38.7 30.4 14.6 -1.81E+00 1.56E-02 fine sand moderately
38.7 30.4 15.0 -1.89E+00 1.29E-02 fine sand poorly
38.7 30.4 15.9 -2.30E+00 5.00E-03
38.7 30.4 16.2 -1 .99E+00 1 .03E-02
38.7 30.4 16.6 -1.91E+00 1.24E-02 fine sand moderately
38.7 30.4 17.4 -2.17E+00 6.70E-03
38.7 30.4 17.7 -1.95E+00 1.12E-02
38.7 30.4 18.2 -2.07E+00 8.60E-03 med sand poorly

 

 

 

 

 

 

 

 

Table A6. Well D10. X, Y and 2 location, Log K value, K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain size (Boggs, 2001)).

120

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

X Y Z LOG(K) (cmls) K (cmls) mean grain size sorting
38.7 32.4 3.8 -1.55E+00 2.85E-02 med sand poorly
38.7 32.4 4.2 -1.19E+00 6.49E-02
38.7 32.4 5.2 -1.53E+00 2.94E-02
38.7 32.4 5.5 -1.57E+00 2.68E-02
38.7 32.4 5.9 -1.61E+00 2.47E-02 fine sand m-well
38.7 32.4 7.0 -2.82E+00 1.50E-03 very fine sand m-well
38.7 32.4 7.4 -1 .74E+00 1 .84E-02
38.7 32.4 8.3 -1 .41 E+00 3.92E-02 med sand moderately
38.7 32.4 8.8 -2.28E+00 5.20E-03 med sand moderately
38.7 32.4 9.4 —1 .58E+00 2.63E-02
38.7 32.4 9.8 -1.54E+00 2.90E-02
38.7 32.4 10.2 -1.49E+00 3.21 E-02 fine sand poony
38.7 32.4 10.9 -1.52E+00 3.00E-02
38.7 32.4 11.3 -1.48E+00 3.33E-02 fine sand m-well
38.7 32.4 11.7 -1.54E+00 2.86E-02
38.7 32.4 12.8 -1 .85E+00 1 .40E-02
38.7 32.4 13.2 -1 .97E+00 1 .08E-02
38.7 32.4 13.6 -1.94E+00 1.14E-02 fine sand m-well
38.7 32.4 14.0 -2.15E+00 7.00E-03
38.7 32.4 14.4 -1.74E+00 1.82E-02
38.7 32.4 14.8 -1.85E+00 1.41E-02 fine sand poorly
38.7 32.4 15.4 -2.09E+00 8.10E-03
38.7 32.4 15.9 -1.95E+00 1.12E-02
38.7 32.4 16.3 -1.90E+00 1.25E-02 fine sand well
38.7 32.4 16.9 -1.97E+00 1.06E-02
38.7 32.4 17.2 -1.83E+00 1 .48E-02
38.7 32.4 17.6 -1.78E+00 1.67E-02
38.7 32.4 18.0 -1.91E+00 1.23E-02 pebble poorly

 

 

 

 

 

 

 

Table A]. Well D12. X, Y and 2 location, Log K value, K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain size (Boggs, 2001)).

121

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

X Y Z LOG-(K) (cmls) K (cmls) mean grain size sorthg
38.8 34.3 0.5 -1 .47E+00 3.40E-02 pebble v-poorly
38.8 34.3 0.7 -1 .40E+00 3.97E-02 Jebble v-poorly
38.8 34.3 0.9 -1 .04E+00 9.02E-02 granules poorly
38.8 34.3 1.1 -1.57E+00 2.72E-02 granules poorly
38.8 34.3 1.3 -1.37E+00 4.28E-02 med sand poorly
38.8 34.3 1.7 -1.38E+00 4.17E-02 coarse sand poorly
38.8 34.3 2.2 -1.24E+00 5.77E-02 coarse sand poorly
38.8 34.3 2.6 -1.17E+00 6.76E-02
38.8 34.3 3.1 -1.25E+00 5.58E-02
38.8 34.3 3.3 -1.25E+00 5.67E-02
38.8 34.3 3.5 -1.21E+00 6.24E-02 med sand m
38.8 34.3 3.9 -1.15E+00 7.03E-02
38.8 34.3 6.8 -1.64E+00 2.28E-02
38.8 34.3 8.1 -1.58E+00 2.64E-02 fine sand m
38.8 34.3 8.5 -1.43E+00 3.70E-02 med sand moderately
38.8 34.3 9.0 -1.57E+00 2.67E-02
38.8 34.3 9.6 -1.65E+00 2.25E-02 med sand poorly
38.8 34.3 10.0 -1.36E+00 4.34E-02 fine sand m-well
38.8 34.3 10.5 -1.69E+00 2.04E-02 fine sand m-well
38.8 34.3 10.9 -1.68E+00 2.08E-02
38.8 34.3 11.1 -1.71E+00 1.96E-02
38.8 34.3 11.9 -1.94E+00 1 .15E-02
38.8 34.3 13.2 -2.10E+00 7.95E-03 fine sand m-well
38.8 34.3 13.6 -1.90E+00 1.27E-02
38.8 34.3 14.9 -1.92E+00 1.20E-02 veryfine sand moderately
38.8 34.3 15.7 -2.01E+00 9.70E-03
38.8 34.3 16.1 -2.06E+00 8.70E-03
38.8 34.3 16.6 -1.91E+00 1.22E-02
38.8 34.3 17.0 -1.80E+00 1.58E-02 fine sand well
38.8 34.3 17.4 -2.03E+00 9.30E-03
38.8 34.3 17.8 -1.91E+00 1 .24E-02

 

 

 

 

 

 

 

 

 

Table A8. Well D14. X, Y and Z location, Log K value, K value, mean grain size (determined
by sieve analysis), and sorting (determined by standard deviation of grain size (Boggs, 2001)).

122

 

APPENDIX B:
OUTCROP ANALOGS

abbreviation

number

trends

 

Facies abbreviation

 

 

 

 

 

 

 

 

 

St trough cross bedded sand
Sm massively bedded sand
Sr sand with ripple marks
Fsm massively bedded silt or clay
Gt trough cross bedded gravel
Gh horizontally laminated gravel
Gu gravel undifferentiated (no distinct bedding evident)

 

Insert Key for Outcrop analogs

Table B.1. Key for Photomosaics

123

 

.58 E “395on 2 cams. .ofiom L8 .665 .moomtam 9.658 520 826. “commas.
98.. 5:52... .mcomtzm mcficaon E6859 8:: Bo=o> .gq co~<~m mecca 8850 .6 2808905.“. .mm 659“.

«m

I.-..IIԤI

m um

 

124

.560 E @2535 m_ $95 .295 .8 .055 .mmomtam 9:958 590 .026. E898.
8:: 5:55. .mmomtam @5953 E0352 3:: Bo=m> an. co~< 8 00.95 8.230 *0 2808905.“. .N.m 2:9“.

125

 

«IIIOOIUOOOIODOO
‘ O
, O

3.8 E 09585 m_ 39:.
.9595 xca 3 £505 9m ucm 299.550 5 9m flcmEmSmmoE =< .wmomtam @5958 .020 .626. 2865
8:: 8:55. .mmomtsm 9:958 .5352 3:: 26=o> .ga :02 8 mega 8.230 *0 0682.905; .m.m 0.59".

 

126

.._o_oo :_ 92:08.5 m_ 89:.
.288E NF >_2mE_xo.aqm m_ 8.050 :0 £92.. .mmomtsm 96:29... :85 .526. E898: 8:: :o::_:._. .8088
@5958 «88:8: 8:: Bo=o> .z: .885 ucm 2mm 869m «m mo_m:m aoBSo ho 068E905”. .v.m 2:9“.

 

127

:28 :_ 02:88.: E 88:... .N.m 9an :_ :96 8032.58: 2 288:8 9.855: 8m
.msotm :5: >9 5597. En ccm 9088580 :_ 08 3:862:88: .8083 95:30: .020 830. 288.59 8:: 8:55.:
.8828 96:30: E0852 8:: Bo=w> .g: .996 ccm ncmm 869w um moficm 8.650 :0 2806902.: .md 930E

 

128

.820 8. 80:8: 8~.8 :.8:0 0:8 80:89 :850
.868: 65:08 .8~.8 £85 :88E 89:3: 08m .m.m 8:39“. :_ $59.8 028:8 89930 :0 80.3.3880 .N.m 8.88...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

9:88-: 9 988-2 8 =83 28.278
8.0800 9 0:88-“. :9 2:00:
0:88-039“. 80:88:98: .8: 0 no _8E9:_ 5 :83
83:00 9 0:88-... 80883918: .8: 0 an .8E9:_ x" >_9.|8:800E
0:86 9 0:88-“. 80:89:02 8500.2. gm. 0: 5 :83
8.0800 9 0:88-.) 09:82:30 :9 >_U9|8:800::
0:.mm-o ouawi 80:80::0: 85008: _8::|9:. 0: 5 :83
8.8000 9 0:88.“ E93: 85:: 5 @809:
980.30: 9 :.m 0:830: 85:: 5 398808:

80.20: 9 0:88-: 90 >.:000-> 83:00 me

9.8-8.0 Emu. =8>> U.._8.>8_0 3v

8.9.80 9 0:“8H.> 9w 3:000 0:88-“. 8%

8.9.8: 0:8 8.0000 9 >8|_0 :5. :0 a9 98 8.8080 0:8 8.9.00 E8“. 3:00: 9. 8 me

0:88-”. 9 0:88-: 90:: :0 t8. 9 8:: 9m :82, 0:88- 2 3

8.8000 9 0:88;). 80:80.88: .m:_008na::9:_ 0: 80 2:000 888 0...

88: 8.8 585 :850 8.08:— m:_t08 8.8 985 :89: 08a

 

 

 

 

 

 

 

129

.:0_00 :. 09:88:: 8. 888E. .vd 8.89 :. :8>.m 8:03.880 9 0:088:00 880:5: 08m 830:8
v.:_a >0 550:8 8:8 0:8 8:995:80 :_ 8:8 9:8E8::888_2 8808038 9.0500 808 :830. 9:88:88: 88:: :8::_:...
.8898 0:650: 9:88:88: 88:: 30=8> 9.: .885 0:8 0:88 8.085 98 00.8:8 099:0 :0 2889:9020 .88 8:30.“.

 

130

.820 8. 80:8: 8N.8 :85 0:8 80:89 :850
88.08.. 85:08 .8N.8 E85 :88... 89:0: 088 8.8 8:30.". :. 559.8 00.8:8 09930 .0 8:09.880 .88 8.08...

 

 

 

 

80:8. .60.. .w :82. 0:88-: 8N
0:80 9 0:88.). 0:82.. 3 3E: .88.. .8 8.9.8. IM ..8309: 0:886 mm
88.9.00 9 0:88-... 8:... 0:8 88:800 .0 8:8 8. 838:8. :99. 5 2:00. 0:88 0.5. w I:

 

8.9.00 9 0:88-:
88.9.00 9 0:88.).
8.9.00 9 0:88-

8...... . o. ..8
88.008. 9 0:88-.)—

 

LL

 

 

 

8.9.8. 500:8: 5.2. 808:8 . . E9. IE

 

 

  

 

 

 

  

 

 

 

830:8 . . E9. a
8.9.8 9 0:88-: 830:8. . E9. .w .... . . 0:88 0.). n
8
8:... 0:8 88:88 .0 8:8 8. .888 :98... m

 

   

 

 

 

 

 

 

 

‘ -1. I‘ I‘ I‘ l‘ I‘ I' I‘ l"|"l"l"|"

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_ .8 0:88.). v
8.9.00 9 0:88-: 830:8 9:9. >000. 0:88-: 8
8 Elfin... N
.8>8: 9 ...8 .w .98:800E 0:88.). P
88.9.8 9 ...8 9m _ 0:88-20
8.9.8 9 ...8 t8. .8

.8>8: 9 8.9. .9. .: .8 8:... 0:8 8:80 .0 8:8 8. 808:8 :99. .m 98:800.: 0:88-: 8
8:... 0:8 88.800 .0 8:8 8. 838:8 .98.. .m :82: 0:88-: 8
0:88-“. 9 8.9.00 808:8 98.. 08.. .0 80.8 88. 0:88.). N
0:88-“. 98.9.00 808:8 0:98.. 8:8 8. :. 8.9.8 5 .:00 0:88-: 8
0:88-“. 9 89.00 .m :0 .8>8: 0:8 88.9.8 8

.8>8: 9 ...8 .m _:0 ..v

.8>8: 9 ._.8 2:0 0:88-: 8».
0:82. a 5:: :82. 0:88-: 8
89.0.. .8 :88:800 289%"

80:8: 8.8 58:0 :850 . 85:08 8.8 58:0 88:: 08..

 

 

 

 

 

131

.:o.00 :. 09:88:: 8. 888:... .¢.m 8.08... :. c828 8:93.580 9 0:088:00 8:895: 08m 8328
0.5.. .3 5598 8:8 0:8 8.98.5280 :. 8:8 8288858885. .8808t38 8505.9. 808 :830. E8858: 8:: :8::......
8808.58 @5059. 9:88:08: 88:: 30__8> .9... .885 0:8 0:88 8.09m .8 828:8 00.930 .0 0.8808905. Nd 8:08.".

 

132

  

8:80.:00-8:_.
88.800 .0 .0. 8 8.8030008...

 

 

 

0:88 8:: 9 E089 8.80.300
88.0000 0:8 88.008789.

.:8>.8 8. 88:8: 8N.8 :.8:8 0:8 .80:8.. .850
88.08. 8:...08 .8N.8 :.8:8 :88... 880:5: 08m .58 8.38.". :. 559.8 80.8:8 00.0.20 .0 8:008:08... 6.8 8.08 h

  

 

 

 

 

 

 

 

 

 

 

    

   

 

2.00

 

 

 

 

 

 

 

 

 

 

 

 

 

88:8 8...: 8 08:. 8. 00.8.: :0
82008.- 8.... 8 .8 =83
8.008. - o... .8 =8;
8.08.- m... .8 =8;
88.0000-0:88 08E .8 .8300...
88.0000-0:88 08E .8 .8.8:800E 0:88 9 .8>8: on
0:88 8:...> 88.0000 .8 0:88 8N
88.0000 0:8 88.008.-8:.. .00.. .80.... 88.008. E0058. .8 .98.800E 0:88-... 8N
88:. 08:. . 8. 88.800 8E08 ..0E0.. .806 .8 =83. 0:88 ..5. R
888:8. 88.8 58.8 _ .850 _88.08u.|_ 8:...08 _ 88.8 :.8.8 :88... _080

 

 

133

REFERENCES

Adams, E. E., and Gelhar, L., 1992, Field study of dispersion in a heterogeneous aquifer,
2. Spatial moments anaylsis: Water Resources Research, v. 28, no. 12, p. 3293-
3307

Anderson, M. P., Aiken, J. S., Webb, E. K., and Mickelson D. M., 1999, Sedimentary and
hydrogeology of two braided stream deposits: Sedimentary Geology, v. 129, p.
1 87- l 99.

Ashley G. M., 1990, Classification of large scale subaqueous bedfrorms; a new look at an
old problem: Journal of Sedimentary Petrology, v. 60, no. 1, p. 161-172

Barrese, P. G., 1991, A Hydrogeochemical Characterization of a Glacial Drift Aquifer
System in Southwest Michigan: Master’s Thesis, Western Michigan University,
Kalamazoo, MI.

Beard, D. C., and Weyl P. K., 1973, Influence of Texture on Porosity of Unconsolidated
Sand: The American Association of Petroleum Geologists, v. 57, no. 2, p. 349-
369.

Benn D. I., and Evans, D. J. A.,1998, Glaciers and Glaciation: New York, Oxford
University Press Inc., 734 p.

Boggs, M. J., Yound, S. C., Beard, L. M., Gelhar, L. W., Renfeldt, K. R., and Adams E.
E., 1992, Field study of dispersion in a heterogenous aquifer, 1., Overview and
site description: Water Resources Research, v. 28, no. 12, p. 3281-3291.

Boggs S. Jr., 2001, Principles of Sedimentology and Stratigraphy 2“d edition, Prentice
Hall, Upper Saddle River, New Jersey.

Brigham Young University, 2000, Department of Defense Groundwater Modeling
System (GMS) Version 3.1., Salt Lake City, UT.

Carle, S. F., LaBolle, E. M., Weissmann, G. S., Van Brocklin, D., and Fogg, G. E., 1998,
Conditional Simulation of hydrofacies architecture: a transition probability
approach, In: Fraser, G.S., Davis, J .M., (Eds). Hydrogeological Models of
Sedimentary Aquifers, Concepts in Hydrogeology and Environmental Geology
No. 1 Society for Sedimentary Geology Special Publication, p. 147-170.

Carle, S. F ., and Fogg, G. R., 1998, TPMOD: A transition probability/ Markov approach

to geostatistical modeling and simulation, User Documentation, University of
California, Davis.

134

Clement, T. P., 1997, A modular computer model for simulating reactive multispecies
transport in three-dimensional groundwater systems: PNNL-SA-11720, Pacific
Northwest National Laboratory, Richland, Washington.

Clement, T. P., and Jones, N. L., 1998, RT3D Tutorials for GMS Users, PNNL-11805,
Pacific Northwest National Laboratory, Richland, Washington.

Davis, M. J ., Wilson, J. L., Phillips, F. M., and Gotkowitz, M. B., 1997, Relationship
between fluvial bounding surfaces and the permeability correlation structure:
Water Resources Research, v. 33, no. 8, p. 1843-1854.

Deutsch, C. V., and. Joumel, A. G., 1992, GSLIB: Geostatistical Software Library and
User’s Guide, Oxford University Press, London.

Dorr, J. A. Jr., and Eschman, D. F., 1970, Geology of Michigan, Ann Arbor, Michigan,
University of Michigan Press, p. 476.

Davis J. M., Lohman, R. C., Phillips, F. M., Wilson, J. L., Love, D. W., 1993,
Architecture of the Sierra Ladrones formation, central New Mexico: Deposition
controls on the permeability correlation structure: Geological Society of America
Bulletin. V. 105, p. 998-1007.

Eschard, R., Lemouzy, P., Bacchiana, C., Desaubliaux, J ., Parpant, J ., and Smart 3.,
1998, Combining sequence stratigraphy, goestatistical simulations and production
data for modeling a fluvial reservoir in Chaunoy Field (Triassic, France):
American Association of Petroleum Geologists Bulletin, v. 82, p. 545-568.

Finbeiner, K. 1994. Subsurface Glacial Geology of the Area Between the Tekonsha and
Kalamazoo Moraines, Kalamazoo County, Michigan: Master’s Thesis, Western
Michigan University, Kalamazoo, MI.

F orstat, D. W., and Sorensen, H. O., 1982, Bedrock Geology of Kalamazoo County,
County Geologic Map Series, Michigan Geological Survey Division, Lansing,
MI.

Hess, K. M., Wolf, S. H., and Celia M. A., 1991, Estimation of Hydralic Conductivity in
a sand and Gravel Aquifer, Cape Code, Massachusetts: USGS Water Resources
Investigation report 91-4034, p. 15-22.

Hoard, C. J ., 2002, The value of Hydraulic Conductivity Data in Groundwater Flow and
Transport Models of Schoolcraft Plume A. Master’s Thesis, Michigan State
University, East Lansing, MI.

Hoeksema, R. J ., and Kitanidis, P. K., 1985, Analysis of the spatial structure of properties
of selected aquifers: Water Resources Research, v. 21, no. 4, p. 563-572.

135

Howell, P. D., and van der Pluijm, B. A., 1990, Early history of the Michigan Basin:
Subsidence and Appalachian tectonicszGeology v. 18, p. 1195-1198.

Howell, P. D., and van der Pluijm, B. A., 1999, Structural Sequences and Styles of
subsidence in the Michigan Basin: Geological Society of America Bulletin, v.
111, no. 7 p. 17

Hyndman, D. W., Dybas, M. J ., Fomey, L., Heine, R., Mayotte, T., Phanikumar, M. S.,
T atara, G., Tiedje, J ., Voice, T., Wallace, R., Wiggert, D., Zhao, X., and Criddle,
C. S., 2000, Hydralic characterization and design of a full-scale biocurtain:
Groundwater, v. 38, no. 3, p. 462-474.

Kehew, A. E., Straw, W. T., Steinmann, W. K., Barrese, P. G., Passarella, G., Peng, W.,
1996, Ground-Water Quality and Flow in a Shallow Glaciofluvial Aquifer
Impacted by Agricultural Contamination: Groundwater, v. 34, no. 3, p. 491-500.

Klingbeil, R., Klieneidam, S., Asprion, U., Ainger, T., and Teutsch, G., 1999, Relating
lithofacies to hydrofacies: outcrop-based hydrogeological characterization of
Quaternary gravel deposits: Sedimentary Geology, v. 129, p. 299-310.

Kolterman, C. E., Gorelick, S. M., 1996, Heterogeneity in sedimentary deposits: A
review of structure-imitating, process-imitating, and descriptive approaches:
Water Resources Research, v. 32, no. 9, p. 2617-2658.

Krigstrom, A., 1962, Geomorphological studies of sandur plains and their braided rivers
in Iceland: Geology Annaler, v. 44, p. 238-346.

Leopold, L. B., and Wolman, M. G., 1957, River channel patterns: braided, meandering,
straight: United States Geological Survey. Professional Paper, 282-B, p. 39-85.

Lipinski, B. A., 2002, Estimating Natural Attenuation Rates for a Chlorinated
Hydrocarbon Plume in a Glacio-fluvial Aquifer, Schoolcraft, Michigan: Master’s
Thesis, Michigan State University, East Lansing, MI.

Martin, H. 1957, Outline of the Geological History of Kalamazoo County: Michigan
Geological Survey Division, Open File Report.

Mayotte, T. J ., 1991, Village of Schoolcraft Kalamazoo County, Michigan Final
Remedial Investigation and Feasibility Study of Remedial Alternatives, Plume A

and Plume F: Halliburton NUS Environmental Corporation Project Number
4M84.

McDonald, G., and Harbaugh, A. W., 1988, A Modular Three-dimensional Finite

Difference Groundwater flow model: US. Geological Survey Techniques of
Water-Resources Investigations, Book 6, chapter A1, Reston, VA.

136

Miall A. D., 1997, A Review of the Braided-River Depositional Environment: Earth
Science Reviews, v.13, p. 1-62.

Miall AD. 1996. The Geology of Fluvial Deposits. Springer-Verlang, Berlin.

Miall, A. D., 1985, Architectural-Element Analysis: A New Method of F acies Analysis
Applied to Fluvial Deposits: Earth Science Reviews, v. 22, p. 261-308.

Moltyaner, G.L., R.W.D. Killey. 1988. The Twin Lake tracer tests: longitudinal
dispersion. Water Resources Research v. 24 no. 10 p. 1613-1627.

Monaghan, G. W., and Larson, G. J ., 1982, Surfical Geology of Kalamazoo County:
County Geologic Map Series, Michigan Geological Survey, Lansing MI.

Monaghan, G. W., Larson, G. L., Gephart, G. D., 1986, Late Wisconsinan Drift
Stratigraphy of the Lake Michigan Ice lobe in Southwestern Michigan: Geological
Society of America Bulletin, v. 97, p. 329-334.

Rehfeldt, K. R., Boggs, M. J ., and Gelhar, L., 1992, Field study in a heterogeneous
aquifer, 3., Geostatistical analysis of hydraulic conductivity: Water Resources
Research, v. 28, no. 12, p. 3281-3291.

Robinson J. W, and McCabe, P. J ., 1997, Sandstone-Body and Shale-Body Dimentions in
a Briaided Fluvial System: Salt Wash Sandstone Member (Morrison Formation),
Garfield County, Utah: American Association of Petroleum Geologists, v. 81, no.
8, p. 1267-1291.

Sleep, N. H., and Sloss, L. L., 1978, A deep borehole in the Michigan Basin: Journal of
Geophysical Research, v. 83, p. 5815-5819.

Smith N. D., 1985, Proglacial Fluvial Environment: Glacial Sedimentary Environments
SEPM short course eds. G.M. Ashley, J. Shaw, N.D. Smith, no. 16, p. 85-134.

Steinmann, W. K., 1994, The Geological and Hydrogeological Characterization of the
Schoolcraft Aquifer, Schoolcraft, Michigan: Master’s Thesis, Western Michigan
University, Kalamazoo, MI.

Sudicky, E. A., 1996. A natural gradient experiment on solute transport in a sand aquifer:
spatial variability of hydraulic conductivity and its role in the dispersion process:
Water Resources Research, v. 22, no. 13, p. 2069-2082.

Weissmann, G. S., and Fogg, G. E., 1999, Multi-scale alluvial fan heterogenity modeled

with transition probability geostatistics in a sequence stratigraphic framework:
Journal of Hydrology, v. 226, p. 48-65.

137

IIIIIIIIIIIIIIIIIIIIIIIIIIIIII

lllllflllllljflljlllWIZIMLHWLIMQIMII