PRAIRIE FEN HYDROLOGY
By
Hassan Abbas

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Civil Engineering
2011

ABSTRACT
PRAIRIE FEN HYDROLOGY
By
Hassan Abbas
Prairie fens provide habitat to more than 50 rare species, including the federally endangered
Mitchell‟s satyr butterfly. Substantial resources have been invested in their protection; however,
these conservation efforts have proceeded without an understanding of the underlying
groundwater flow regime that is critical to these fens. Regional scale hydrological investigations
could cost hundreds of thousands of dollars for an individual fen, and therefore, are rarely
undertaken in a management context – confining the conservation efforts to the local fen
community and adjacent lands. Without delineation of groundwater sources, the conservation
community has had limited ability to protect the fen groundwater quantity and quality.
In this research we explored the use of a data driven modeling approach for groundwater
modeling and to improve our ability to understand fen hydrology at different spatial scales. The
approach used in this study directly analyzes, filters, and processes water well records and
surface water elevations to estimate groundwater flow. We developed a steady state mean flow
model for the southern Michigan basin, regional scale models for 9 fen clusters/sites, and sitespecific vertical profile models for 19 fens. Vertical profile models facilitated understanding
groundwater delivery mechanisms from regional recharge mounds to individual fens. Our major
findings are: 1) occurrences of fens exhibit a clear systematic pattern on the regional scale; 2)
most prairie fens are located around or at the foot of several large groundwater “mounds” at the
intersection of major watersheds. These mounds are critical, regional source water areas for

aquifer systems, fens, and other groundwater-dependent ecosystems (GDE); 3) these critical
regional mounds play a disproportionally large role controlling the sustainability of groundwater
resources and GDE; 4) Groundwater flow patterns in the regional source water areas are complex
and most fens are recharged from multiple sources; 5) Prairie fens located seemingly far apart in
different watersheds, counties, or states may share the same regional source water; and, 6) Prairie
fens exhibit four distinct types of groundwater connections, which are: i) shallow connections –
groundwater recharge from local hills and surface water in lakes/wetlands at higher elevations
adjacent to a fen seeping into the fen through shallow, and relatively short, groundwater flow
paths; ii) deep connections – source waters recharge at distant locations coming in contact with
deeper glacial/bedrock formations before upwelling into the fen; iii) confined connections –a
distant source water recharge area becomes connected to a fen through older outwash beds
confined under recent till plains; and, iv) cascading connections – groundwater sources directly
linked to fens being dependent on other surface/sub-surface source waters.
These findings have significant practical implications and support the need to reconsider
the current priorities and restoration strategies. In particular: 1) our improved ability to use GIS
data creates new possibilities with basin-wide implications for the holistic management of fen
ecosystems; 2) fen management must move beyond water‟s edge to account for the impact of
regional flow systems; 3) because of the connectivity of the basin‟s groundwater systems, a few
“smart” actions in key locations could yield high ecological returns; 4) when we protect prairie
fens and their upstream sources, we are also protecting other ecosystems downstream; and, 5)
occurrence of fens in the regional aquifer recharge areas at the intersections of watershed
boundaries implies that coordination across the river basins as well as the political divides and
between water resources and ecological communities would be imperative for protection of fens.

To my parents
And my brothers Mohsin (1961-1995) and Hussain (1965-2010)

iv

ACKNOWLEDGEMENTS
I gratefully acknowledge the contributions of all those who made this research possible. I
express my thanks to Professor Shu-Guang Li, my Ph.D. advisor, Department of Civil and
Environmental Engineering at Michigan State University (MSU), for his continuous guidance,
advices and encouragement throughout the course of my graduate studies. I would like to extend
my very special thanks to Dr. Huasheng Liao, visiting professor, Department of Civil and
Environmental Engineering at MSU, for his relentless efforts in developing new tools and
making countless improvements in our groundwater modeling system, Interactive Groundwater
(IGW), as well as his ever available guidance and support which made all the modeling work
possible for this research. I also extend very special thanks to Mr. Adrain Zhou, research
specialist at MSU and the chief designer and developer of our data modeling software Michigan
Interactive Groundwater for Wellhead Protection (MIGWWP).
I thankfully acknowledge the support provided by the fen conservation community. In
particular, I express my special thanks to Mr. Jack Dingledine and Ms. Tameka Dandridge from
US Fish and Wildlife Service, Mr. Matthew Herbert, program manager at the Nature
Conservancy, Prof. Douglas Landis, Department of Entomology at MSU and Ms. Daria Hyde,
Conservation Scientist at Michigan Natural Features Inventory.
I also wish to extend my thanks to the other members of our research team who had been
supporters and/or participants in many of the research aspects presented in this dissertation. They
include Mr. Richard Mandle, Michigan Department of Natural Resources and Environment,
Professor Jon Bartholic, Institute of Water Research at MSU, Dr. Steven Hamilton, Kellogg
Biological Station at MSU, Professor Grahame Larson, Geological Sciences at MSU, and Dr.
Qun Liu, former Post Doc at MSU.
v

I am thankful to the organizations that provided funding for my research at MSU which
include U.S. Fish and Wildlife Service, Great Lakes Protection Fund, National Science
Foundation, Michigan Department of Environmental Quality, MSU College of Engineering, and
MSU Institute of Water Research.
Finally, I acknowledge the guidance and support from my parents and siblings who had
been a constant source of courage and motivation for me in the continued pursuit of knowledge
throughout my life.

vi

TABLE OF CONTENTS
List of Tables ................................................................................................................................. x
List of Figures ............................................................................................................................... xi
CHAPTER 1 INTRODUCTION ............................................................................................... 1
Motivation ...................................................................................................................................... 1
Scientific Challenges ..................................................................................................................... 4
Review of Prior Work ................................................................................................................... 6
Historical Perspective ............................................................................................................................. 6
Need for Multi-scale Thinking .............................................................................................................. 7
Need for a Quantitative Framework ................................................................................................... 11
Challenges Facing Multiple Scale Thinking on Flow Regime Integrity ....................................... 14
Aim and Objectives ..................................................................................................................... 15
CHAPTER 2 METHODOLOGY ............................................................................................ 17
Overview ...................................................................................................................................... 17
Statewide Model .......................................................................................................................... 18
Regional Models .......................................................................................................................... 19
Data-Driven Characterization of Groundwater Heads in Shallow System ................................... 19
Data-driven Characterization of Groundwater Heads in Glacial Drift .......................................... 25
Data-driven Characterization of Groundwater Heads in Bedrock .................................................. 29
Estimating Hydraulic Conductivity .................................................................................................... 29
Estimating Porosity ............................................................................................................................... 31
Tracking Source Water in Horizontal Plane ...................................................................................... 31
Vertical Profile Models for Groundwater Systems .................................................................. 33
Modeling Assumptions ........................................................................................................................ 33
Boundary Conditions for Profile Models ........................................................................................... 34
Hydraulic Conductivity ........................................................................................................................ 36
Vertical Anisotropy............................................................................................................................... 37
Estimating Water Table along the Profile.......................................................................................... 38
Estimating Discontinuities within the Glacial Drift ......................................................................... 38
Tracking Source Water in Vertical Plane .......................................................................................... 39
Sensitivity Analysis .................................................................................................................. 40
Hydraulic Conductivity ........................................................................................................................ 40
Vertical Anisotropy............................................................................................................................... 41
Discontinuities in Drift ......................................................................................................................... 41
Comparing with Traditional Approach .................................................................................... 44
Adopting Data Driven Modeling Approach ..................................................................................... 47
Data Used ..................................................................................................................................... 50
Pre-processing of Data ................................................................................................................ 52
Noise, Outliers and Bad Data .............................................................................................................. 52

vii

Including Surface Water Features ....................................................................................................... 56
Interpolation and Smoothing ............................................................................................................... 58
Generating Statewide Raster Maps for Static Water Level ............................................................. 59
CHAPTER 3 RESULTS ........................................................................................................... 61
Statewide Model .......................................................................................................................... 61
Statewide Hydrology and Occurrence of Fens .................................................................................. 62
Flow Regime and Fen Occurrence Compared to Glacial Landforms ............................................ 63
Statewide Regions of Significance for Source Groundwater .......................................................... 64
Selection of Fen Sites for Detailed Modeling ............................................................................ 65
Regional and Profile Models for Selected Sites ........................................................................ 70
Barry State Game Area Group of Fens .............................................................................................. 70
Liberty Fen and Skiff Lake/MacCready Fen ..................................................................................... 91
Ives Road Fen ...................................................................................................................................... 103
Sarett Nature Center Fen .................................................................................................................... 110
Paw Paw Lake Fen and Paw Paw Prairie Fen ................................................................................. 115
Monette St Fen .................................................................................................................................... 123
Springbrook Fen and Butterfield Lake Fen ..................................................................................... 128
Dew Road Fen ..................................................................................................................................... 138
Silver Lake Fen ................................................................................................................................... 143
CHAPTER 4 DISCUSSION .................................................................................................. 148
Basin-wide Groundwater Picture ............................................................................................ 148
Regional Upwelling ................................................................................................................... 148
Source Water Areas for Fen Sites ........................................................................................... 151
Local Sources ...................................................................................................................................... 151
Sub-Regional Sources ........................................................................................................................ 151
Regional Sources ................................................................................................................................. 152
Delivery Mechanisms of Groundwater at Fens Sites ............................................................. 152
Shallow Connections .......................................................................................................................... 152
Deep Connections ............................................................................................................................... 153
Confined Connections ........................................................................................................................ 156
Cascading Connections ...................................................................................................................... 157
Major Implications of Findings ............................................................................................... 164
Protection of Source Water Areas .................................................................................................... 164
Basin Wide Perspective ...................................................................................................................... 165
Policy, Management and Development ........................................................................................... 170
CHAPTER 5 CONCLUSIONS.............................................................................................. 171
Summary of Significant Findings ............................................................................................ 171
Summary of Implications ......................................................................................................... 173
Relating to Previous Work ....................................................................................................... 175
Strengths and Limitations of the Study .................................................................................. 178

viii

Models Validation ............................................................................................................................... 178
Limitations of Profile Models ........................................................................................................... 179
Recommendations for Further Study ..................................................................................... 181
Closing Remarks ....................................................................................................................... 182
REFERENCES .......................................................................................................................... 184

ix

LIST OF TABLES
Table 4.I Summary of modeled fen sites ................................................................................... 165

x

LIST OF FIGURES
Figure 1.1 Southern half of Michigan‟s Lower Peninsula showing locations of prairie fens and
Mitchell‟s satyr occurrence. The map was created by using DEM data (NED USGS 2006),
NHD data (NHD USGS 2010) and fen locations data (including occurrence of Mitchell‟s
satyr) from Michigan Natural Features Inventory. Shaded relief represents surface
topography. For interpretation of the references to color in this and all other figures, the
reader is referred to the electronic version of this dissertation. .............................................. 3
Figure 1.2 Conceptualization of multiples scales of groundwater flow systems [Modified from
Meyboom (1967)] ................................................................................................................... 5
Figure 2.1 Scatter points generated in surface water features (lakes, streams and wetlands) to
characterize shallow groundwater head using 30m/10m DEM around a fen site................. 22
Figure 2.2 Gaussian model fit to the variogram .......................................................................... 23
Figure 2.3 Contours at 2.12m interval represent shallow groundwater surface without smoothing
(top) and with 3 and 5 passes (mid and bottom) respectively............................................... 24
Figure 2.4 The relief and contours around fen sites represent the shallow groundwater surface
interpolated from the data generated in surface water features (lakes, streams and wetlands).
Contours at 2.12m interval show shallow water table varying in elevation from 176.636m to
239.196m. Contour interval is 3m. ....................................................................................... 25
Figure 2.5 Shaded relief and contours around fen sites represent the regional groundwater
surface interpolated from the water wells data. Black dots mark the data points. Contours
show regional groundwater heads varying in elevation from 177.5936m to 218.419m.
Contour interval is 2m. ......................................................................................................... 27
Figure 2.6 Groundwater head contour maps for deep system (top) and shallow system (bottom) .
...................................................................................................................................... 28
Figure 2.7 Groundwater capture zone delineation for a fen site based on reverse particle
tracking. The relief and contours represent regional groundwater surface. Contours show
regional groundwater heads varying in elevation from 177.5936m to 218.419m. Contour
interval is 2m......................................................................................................................... 32
Figure 2.8 A profile sections is selected between a regional source and a regional sink,
intersecting the fen site and its regional groundwater capture zone. Relief and contours in
the figure represent shallow groundwater surface or the water table. Contours show regional
groundwater heads varying in elevation from 177.5936m to 218.419m. Contour interval is
2m. ..................................................................................................................................... 35
xi

Figure 2.9 Typical boundary conditions for the profile models .................................................. 36
Figure 2.10 Boundaries, zones, and water table estimated for a profile model ........................... 38
Figure 2.11 Water well logs around the profile (above) projected in vertical plane (bottom) to
show clay occurrences .......................................................................................................... 39
Figure 2.12 Profile model showing the groundwater flow paths (pink lines) leading to the fen
site (purple lines). Light blue lines show groundwater head contours at 1.84m interval.
Vertical exaggeration in this figure and all other models presented in this study is 10x,
unless noted otherwise. ......................................................................................................... 40
Figure 2.13 Variable (top) versus uniform K (bottom) field in glacial drift profile. Orange flow
paths are fen capture zones. .................................................................................................. 40
Figure 2.14 Sensitivity to vertical anisotropy. Orange and purple flow paths represent capture
zone for the fen sites ............................................................................................................. 41
Figure 2.15 Effect of scattered discontinuities at Paw Paw Prairie Fen site. Orange flow paths
indication fen capture zones.................................................................................................. 42
Figure 2.16 Effect of scattered discontinuities at Paw Paw Lake Fen site. Orange flow paths
indication fen capture zones.................................................................................................. 43
Figure 2.17 Conceptual aquifer systems in Michigan ................................................................ 49
Figure 2.18 Water Well Drillers records in the Wellogic Database have a good coverage
throughout the state, especially in the southern half of Lower Peninsula. ........................... 53
Figure 2.19 A slice of wells selected across the state and the plot of their static water level
elevations. ............................................................................................................................. 54
Figure 2.20 Bad data and outliers were identified and removed. Rest of the data was used to
evaluate SWL trends through interpolation and smoothing. ................................................ 55
Figure 2.21 Surface water data sampling settings ....................................................................... 57
Figure 2.22 Different number of smoothing passes applied to a regional model after inverse
distance interpolation of water wells data for drift aquifer. Contour interval is 2m. Blue
color represents lower heads and orange higher. For this particular site, 10x smoothed
surface was selected for particle tracking. ............................................................................ 60

xii

Figure 3.1 Surface and groundwater hydrology in relation to occurrence of fens. Relief and
contours represent groundwater surface elevation which varies from 148.14m to 344.93m.
Contour interval is 7m. ......................................................................................................... 62
Figure 3.2 Occurrence of fens with respect to statewide groundwater hydrology and the glacial
landforms. Relief and contours represent groundwater surface elevation which varies from
148.14m to 344.93m. Contour interval is 7m. ...................................................................... 63
Figure 3.3 Source groundwater areas of regional significance for prairie fen sites across the
state. Relief and contours represent groundwater surface elevation which varies from
148.14m to 344.93m. Contour interval is 7m. ...................................................................... 64
Figure 3.4 Fens selected for detailed regional groundwater modeling. Shaded relief and
contours represent groundwater levels delineated at statewide scale varying from 148.14m
to 344.93m. Contour interval is 7m. ..................................................................................... 66
Figure 3.5 Selected fens in Jackson County. Contours at 7m interval represent groundwater
delineation from statewide model. ........................................................................................ 67
Figure 3.6 Ives road fen located in a narrow band of glacial outwash along Raisin River.
Contours at 7m interval represent groundwater delineation from statewide model. ............ 68
Figure 3.7 Cluster of fens between Barry and Allegan Mounds. Contours at 7m interval
represent statewide groundwater........................................................................................... 69
Figure 3.8 Cluster of fens in Barry State Game Area. The shaded relief represents topography
derived from 10m DEM which varies from 223.22m to 342.78m. ...................................... 72
Figure 3.9 The contours at 2m interval and shaded relief represent regional groundwater head
around Barry State Game Area in the drift aquifer, varying between 220 m and 279 m. .... 73
Figure 3.10 The contours at 2m interval and shaded relief represent regional groundwater head
around Barry State Game Area in the drift aquifer, varying between 221 m and 259 m. .... 74
Figure 3.11 Comparison of Bedrock and drift SWL surfaces in Barry State Game Area. Contour
labels represent elevation (m). .............................................................................................. 75
Figure 3.12 Distribution of clay in well logs across selected sections in model area ................. 77
Figure 3.13 The shaded relief represent surface topography in the region which varies in
elevation from 241.06m to 298.58m. .................................................................................... 78

xiii

Figure 3.14 Well location and well logs along Hill Creek Fen capture zone. Contours at 2m
interval show SWL for deep drift. ........................................................................................ 79
Figure 3.15 Profile Model for Hill Creek Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 1.86m interval.
Reverse particle tracking to estimate groundwater flow regime (pink) and likely fen‟s
source water (purple) [bottom]. ............................................................................................ 80
Figure 3.16 Wells and well logs along Turner Creek Wetlands Fen capture zone. Contours at 2m
interval show SWL for drift. ................................................................................................. 81
Figure 3.17 Profile Model for Turner Creek Wetlands Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours at
1.90m interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen‟s source water (purple) [bottom]. ......................................................................... 82
Figure 3.18 Well locations and well logs along Bowens Mill Fen capture zone. Contours at 2m
interval show SWL for drift. ................................................................................................. 83
Figure 3.19 Profile Model for Bowens Mill Fen with clay lenses in the glacial drift [bottom] and
without [top]. Light blue lines show groundwater head contours at 0.77m interval. Reverse
particle tracking shows estimated groundwater flow regime (pink) and likely source water
paths for the fen (purple)....................................................................................................... 84
Figure 3.20 Well locations and well logs along McDonalds Lake Fen capture zone. Contours at
2m interval show SWL for drift. ........................................................................................... 85
Figure 3.21 Profile Model for McDonald Lake Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours at
1.83m interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen‟s source water (purple) [bottom]. ......................................................................... 86
Figure 3.22 Well locations and well logs along Yankee Spring Fen capture zone. Contours at 2m
interval show SWL for drift. ................................................................................................. 87
Figure 3.23 Profile Model for Yankee Spring Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours at
1.92m interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen‟s source water (purple) [bottom]. ......................................................................... 88
Figure 3.2 4 . Well locations and well logs along Hall Lake Fen and McKibben Fen capture zone.
.................................................................................................................................... 89

xiv

Figure 3.25 Profile Model for Hall Lake Fen and McKibben Fen. Geological layers and
delineation of groundwater head and velocity [top]. Light blue lines show groundwater head
contours at 2.30m interval. Reverse particle tracking to estimate groundwater flow regime
(pink) and likely fen‟s source water (purple) [bottom]. ........................................................ 90
Figure 3.26 Liberty Fen. The shaded relief represent surface relief derived from 10m/30m DEM.
Surface elevation varies between 310.69m and 384.31m. .................................................... 91
Figure 3.27 Skiff Lake and MacCready Fens. The shaded relief represent surface relief derived
from 10m/30m DEM. Surface elevation varies between 294.77m and 346.75m. ................ 92
Figure 3.28 The contours at 2m interval and shaded relief represent regional scale delineation of
groundwater surface around Liberty Fen and Skiff Lake Fen which varies in elevation from
292.92m to 347.14m. ............................................................................................................ 93
Figure 3.29 Zone of regional upwelling around Skiff Lake/MacCready Fens and locations of
artesian wells in the area. Contours at 1.05m interval and relief represent groundwater head
surface delineated by interpolating bedrock wells data. ....................................................... 95
Figure 3.30 The shaded relief represents surface topography in the region which varies in
elevation from 296.27m to 387.36m. .................................................................................... 96
Figure 3.31 Local flow system around Liberty fen...................................................................... 97
Figure 3.32 Local flow systems around Skiff Lake/MacCready fens .......................................... 98
Figure 3.33 Selected sections for profile models. Shaded relief represents topography and
contours represent regional SWL at 2m interval. ................................................................. 99
Figure 3.34 Profile Model for Liberty Fen. Geological layers along the profile and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours at
1.85m interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen‟s source water areas (purple)[bottom]. ............................................................... 100
Figure 3.35 Profile Model for Skiff Lake/Macready Fens. Geological layers along the profile
and delineation of groundwater head and velocity [top]. Light blue lines show groundwater
head contours at 1.88m interval. Reverse particle tracking to estimate groundwater flow
regime (pink) and likely fen‟s source water areas for Skiff Lake (purple) and MacCready
(brown) fens [bottom]. ........................................................................................................ 102
Figure 3.36 Ives Road Fen along the Raisin River valley. The shaded relief represents
topography derived from 10m/30m DEM which varies from 218.98m to 253.64m. ......... 103

xv

Figure 3.37 The contours at 3m interval and shaded relief represent regional scale delineation of
groundwater surface around Ives Road Fen which varies in elevation from 217.40m to
295.15m. ............................................................................................................................. 104
Figure 3.38 The shaded relief represents surface topography, varying in elevation from 217.42m
to 332.32m. Glacial landforms in the area are represented by different colors. ................. 105
Figure 3.39 Thickness [top] and depth [bottom] of clay deposits interpolated from water wells
data. .................................................................................................................................. 107
Figure 3.40 Perspective visualization based on data [top] and conceptual profile [bottom] of Ives
Road regional groundwater capture zone. .......................................................................... 108
Figure 3.41 Profile Model for Ives Road Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 2.84m interval.
Reverse particle tracking to estimate groundwater flow regime (pink) and likely fen‟s
source water (purple) [bottom]. .......................................................................................... 109
Figure 3.42 Sarett Nature Center Fen along the Paw Paw River valley. The shaded relief
represents topography derived from 10m DEM which varies from 176.75m to 198.44m. 110
Figure 3.43 The contours at 2m interval and shaded relief represent regional groundwater
head/surface around Sarett Nature Center Fen which varies from 176.30m to 218.42m. .. 111
Figure 3.44 The shaded relief represent surface topography in the region which varies in
elevation from 176.31m to 256.95m. .................................................................................. 112
Figure 3.45 Well location and well logs along Sarett Nature Center Fen capture zone. Contours
at 2m interval show SWL for drift. ..................................................................................... 113
Figure 3.46 Profile Model for Sarett Nature Center Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours at
1.84m interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen‟s source water (purple) [bottom]. ....................................................................... 114
Figure 3.47 Paw Paw Lake Fen and Paw Paw Prairie Fen along Paw Paw River. The shaded
relief represents topography derived from 10m DEM which varies from 261.74m to
316.91m. ............................................................................................................................. 116
Figure 3.48 The contours at 1m interval and shaded relief represent regional scale delineation of
groundwater surface around Paw Paw Fens which varies in elevation from 243.60m to
273.48m. ............................................................................................................................. 117

xvi

Figure 3.49 The shaded relief represent surface topography in the region which varies in
elevation from 243.64m to 321.76m. .................................................................................. 118
Figure 3.50 Well location and well logs along Paw Paw Lake Fen capture zone. Contours at 1m
interval show SWL for drift. ............................................................................................... 119
Figure 3.5 1Profile Model for Paw Paw Lake Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours at
0.48m interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen‟s source water (purple) [bottom]. ....................................................................... 120
Figure 3.52 Well location and well logs along Paw Paw Prairie Fen capture zone. Contours at
1m interval show SWL for drift. ......................................................................................... 121
Figure 3.53 Profile Model for Paw Paw Prairie Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours at
0.84m interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen‟s source water (purple) [bottom] ........................................................................ 122
Figure 3.54 Monette St Fen. The shaded relief represents topography derived from 10m DEM
varying from 241.03m to 261.10m. .................................................................................... 123
Figure 3.55 The contours at 0.61m interval and shaded relief represent regional scale
delineation of groundwater surface around Monette St Fen which varies in elevation from
241m to 260m. .................................................................................................................... 124
Figure 3.56 The shaded relief represent surface topography in the region varying in elevation
from 241.06m to 298.58m. ................................................................................................. 125
Figure 3.57 Well locations and well logs along Monette St Fen capture zone. Contours at 0.61m
interval show SWL for drift. ............................................................................................... 126
Figure 3.58 Profile Model for Monette St Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 0.90m interval.
Reverse particle tracking to estimate groundwater flow regime (pink) and likely fen‟s
source water (purple) [bottom]. .......................................................................................... 127
Figure 3.59 Springbrook Fen. The shaded relief represent surface relief derived from 30m/10m
DEM. Surface elevation varies between 255.36m and 293.80m. ....................................... 128
Figure 3.60 Butterfield Lake Fen. The shaded relief represent surface relief derived from
30m/10m DEM. Surface elevation varies between 256.49m and 282.75m........................ 129

xvii

Figure 3.61 The contours at 2m interval and shaded relief represent regional scale delineation of
groundwater surface around Springbrook and Butterfield Lake fens. The SWL elevation
ranges from 240.36m to 281.64m. ...................................................................................... 131
Figure 3.62 The shaded relief represent surface topography in the region which varies in
elevation from 240.36m to 321.27m. .................................................................................. 132
Figure 3.63 The shaded relief represent bedrock surface in the region which varies in elevation
from 175m to 238m. Bedrock formations in the area are represented by different colors.
Both fens occur in drift deposits over Coldwater Shale. .................................................... 133
Figure 3.64 Well locations and well logs along Springbrook Fen capture zone. Contours at 2m
interval show SWL for drift. ............................................................................................... 134
Figure 3.65 Profile Model for Springbrook Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours at
1.02m interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen‟s source water (purple) [bottom]. ....................................................................... 135
Figure 3.66 Well locations and well logs along Butterfield Lake Fen. Contours at 2m interval
show SWL for drift. ............................................................................................................ 136
Figure 3.67 Profile Model for Butterfield Lake Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours at
0.99m interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen‟s source water (purple) [bottom]. ....................................................................... 137
Figure 3.68 Dew Road Fen. The shaded relief represent surface topography derived from
30m/10m DEM. Surface elevation varies between 309.46m and 363.80m........................ 138
Figure 3.69 The contours at 0.5m interval and shaded relief represent regional scale delineation
of groundwater surface around Dew Road Fen which varies in elevation from 306.59m to
321.97m. ............................................................................................................................. 139
Figure 3.70 The shaded relief represents surface topography around Dew Road Fen which varies
in elevation from 309.40m to 362.92m. The fen occurs in ice-contact till while the capture
zone extends south to glacial outwash deposits. ................................................................. 140
Figure 3.71 Section for profile model. Shaded relief represents topography and contours at 0.5m
interval represent regional SWL. ........................................................................................ 141
Figure 3.72 Profile Model for Dew Road Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 0.84m interval.
xviii

Reverse particle tracking to estimate groundwater flow regime (pink) and likely fen‟s
source water area (purple) [bottom]. ................................................................................... 142
Figure 3.73 Silver Lake Fen. The shaded relief represents topography derived from 30m/10m
DEM which varies from 260.55m to 303.04m. .................................................................. 143
Figure 3.74 The contours at 1.16m interval and shaded relief represent regional scale
delineation of groundwater surface around Silver Lake Fen which varies in elevation from
240.50m to 266.36m. .......................................................................................................... 144
Figure 3.75 The shaded relief represent surface topography in the region which varies in
elevation from 240.51m to 310.94m. Glacial landforms in the area are represented by
different colors. The fen occurs in glacial till. .................................................................... 145
Figure 3.76 Well location and well logs along Silver Lake Fen capture zone. Contours at 1.16m
interval show SWL for drift. ............................................................................................... 146
Figure 3.77 Profile Model for Silver Lake Fen. Groundwater flow lines show estimated
groundwater flow regime (pink) and likely source water paths for the fen (purple). Light
blue lines show groundwater head contours at 0.74m interval. .......................................... 147
Figure 4.1 Strong regional groundwater flow convergence shown by water wells data around
Paw Paw Lake and Paw Paw Prairie fens (left) and Skiff Lake fen (right). ....................... 149
Figure 4.2 Strong regional groundwater flow convergence shown by water wells data around
Springbrook fen (top left), Monette fen (top right) and Sarett Nature Center fen (bottom). ....
................................................................................................................................... 150
Figure 4.3 Delivery mechanism – adjacent shallow connections. ............................................. 153
Figure 4.4 Delivery mechanism – deep connections. ................................................................. 154
Figure 4.5 Upwelling – local source.......................................................................................... 155
Figure 4.6 Upwelling – sub-regional source .............................................................................. 155
Figure 4.7 Upwelling – regional source ..................................................................................... 156
Figure 4.8 Delivery mechanism – confined connections. .......................................................... 156
Figure 4.9 Delivery mechanism – cascading connections. ........................................................ 157

xix

Figure 4.10 Example from Liberty Fen – the fen site exhibits sub-regional upwelling as well as
shallow connections. ........................................................................................................... 158
Figure 4.11 Example from Ives Road Fen – the fen site exhibits shallow, confined and cascading
connections. ........................................................................................................................ 159
Figure 4.12 Example from Skiff Lake Fen – the fen site exhibits adjacent, local upwelling and
sub-regional upwelling connections. .................................................................................. 160
Figure 4.13 Example from MacCready Fen – the fen site exhibits regional and sub-regional
upwelling connections. ....................................................................................................... 161
Figure 4.14 Michigan Recharge area proposed as Forest Legacy Area (MDNR 2003)[top];
Groundwater mounds in South-eastern Michigan [bottom] coincide with the proposed
Forest Legacy Area. ............................................................................................................ 166
Figure 4.15 Glacial landforms and Michigan‟s recharge areas ................................................. 168
Figure 4.16 Recharge dynamics at basin-scale. Vertical exaggeration is 50x. .......................... 169

xx

Chapter 1

Introduction

MOTIVATION
Prairie fens are geologically rare and ecologically unique groundwater dependent
terrestrial ecosystems in Michigan providing habitat to more than 50 rare Midwest plant and
animal species, including the federally endangered Mitchell‟s satyr butterfly and the Eastern
Massasauga rattlesnake, which is a candidate for federal listing. Due to their regional
significance as globally rare ecosystems and as hotspots for rare species, substantial resources
have been invested in the protection and management of prairie fens.

However, these

conservation efforts have proceeded without an understanding of the underlying groundwater
flow regime and associated groundwater contributing areas that are critical in maintaining these
fens. Detailed field evaluations of ground-watersheds generally cost hundreds of thousands of
dollars for an individual fen, and are therefore rarely undertaken in a management context. As a
result, land protection and conservation efforts have focused on only the local fen community
and the immediately adjacent lands. Without a delineation of ground-watersheds that supply a
given fen, the conservation community has had very limited ability to protect the fen
groundwater quantity and quality. A “Great Lake Expert‟s Roundtable Report” (GLPF 1998)
stressed that current conservation efforts [for most ecosystems] in the basin are too disjointed
and focused too much on local, visible symptoms and we must think in terms of protecting the
underlying processes, the functionalities, and the biological engine to enable sustainable
restoration.

1

This research aims at improving our understanding of subsurface hydrology at these
unique groundwater dependent ecosystems, particularly in terms of system connectivity, source
water delineation and delivery mechanisms of groundwater to the fens.
A prairie fen is a type of peat land through which a continuous supply of cold
groundwater, rich in calcium and magnesium carbonates, prevails throughout the year (Kost and
Hyde 2009) and sustains the ecological functions of the ecosystem.
Prairie fens in Michigan provide habitat to Federally Listed Mitchell‟s satyr (MS)
butterfly (neonympha mitchellii mitchellii) – one of the rarest in North America (Barton and
Bach 2005). Historically, the butterfly had existed at thirty plus locations across Michigan, Ohio,
Indiana, New Jersey and Maryland (Barton and Bach 2005; Hyde et al. 2001), but as per summer
of 2010 survey conducted by Michigan Natural Feature Inventory (MNFI) the butterfly‟s
occurrence was only reported at sixteen sites in Michigan (MNFI 2010) and just one in Indiana
(Kost and Hyde 2009). The specific conditions required for the viability of MS populations
remain uncertain, but the maintenance of quality and quantity of the groundwater feeding the fen
is considered vital to a viable satyr population (USFWS 1998). There are 150 known prairie fen
sites in Michigan as shown in Figure 1.1.
Besides providing critical habitat to rare species, fens also provide critical ecological
services to the environment and their protection is considered vital (USFWS 1998). Many of the
region‟s rivers and lakes have their headwaters emanating from the pristine waters of these fens,
supporting countless fish and aquatic species. They provide critically important ecological
services by delivering clean waters to streams and lakes (Bedford and Godwin 2003; Kost and
Hyde 2009).

2

Prairie Fens
100 KM

Prairie Fens with
Mitchell’s satyr
occurrence

0

Figure 1.1

Southern
locations

half
of

of

Michigan’s

prairie

Lower

fens

and

Peninsula

showing

Mitchell’s

satyr

occurrence. The map was created by using DEM data (NED
USGS 2006), NHD data (NHD USGS 2010) and fen locations
data

(including

Michigan

occurrence

Natural

Features

of

Mitchell’s

Inventory.

satyr)

Shaded

from

relief

represents surface topography. For interpretation of the
references to color in this and all other figures, the
reader

is

referred

to

the

electronic

version

of

this

dissertation.

Deficient information on groundwater conditions across fens may be preventing resource
managers from insights that would facilitate protection of fens and their rare species. As a result
of the current lack of understanding of the multiple scale groundwater context for most fens,
critical information is often missing for making decisions on where and how to most effectively
invest limited conservation resources for protection, management, and permitting.

3

SCIENTIFIC CHALLENGES
The fens generally occur in complex hydrogeological settings and while investigating the
hydrology at any particular site, one of the major scientific challenges faced by the fen
conservation community is the lack of knowledge about the source water areas of groundwater
feeding into the fens. Prairie fens generally occur in hummocky topographies of glacial
moraines, within an extensive network of small streams, lakes, and other wetlands, making the
subsurface hydrology very complex. Shallow groundwater recharges under “knobs” and
discharges in the “kettles”, and when superimposed by the regional topography it gives rise to
deeper regional groundwater flow systems (Meyboom 1967). The delivery mechanisms of
groundwater to the fens in complex hydrological settings of prairie fens are not well understood
for most sites.
Figure 1.2 shows conceptualization of multi-scale subsurface flow regime beneath a
hummocky landscape with a number wetlands, lakes and small streams. Large number of surface
water features and shallower water tables in the vicinity of fens may suggest that water tables are
topographically controlled and may be dependent on the complex topography in the region
(Haitjema and Mitchell-Bruker 2005). The figure illustrates that some groundwater discharge
zones depend on local flow systems only (marked A) while some others may be dependent on
both the local and the regional flow systems (marked B). While the recharge zone for a shallow
system may be from a few hundreds of meters to a couple of kilometers from its discharge zone
(marked F), the source water area for regional groundwater discharge could be tens of kilometers
away (marked E). Discharge zones are generally more visible in the landscape. Every discharge
zone can be associated to one or more source water areas such as a recharge area in the
landscape, a surface water body etc. However, finding and establishing connections between
4

discharge zones and corresponding source area is challenging, especially when the groundwater
systems supporting the fens are multi-scale and complex as described above. It requires
knowledge/data on subsurface hydrology and geology.
E
E
F

F

C

A

A

D

A

D

B

D

D

A

D
B

Discharge
Zones

Recharge
Zones

Groundwater
flow system

Extent of flow
system

A

Local only

C

Regional

Regional

E

Regional

B

Regional &
Local

D

Local

Local

F

Local

Figure 1.2

Conceptualization of multiples scales of groundwater flow
systems [Modified from Meyboom (1967)]

Fens occur in groundwater discharge zones within the local hydrological settings and are
mostly recognized by the type of vegetation/species. Quality and quantity of water are critical
factors determining the type of species and the socio economic values provided by the fens (SNH
2010). In order to synergistically, holistically and effectively protect, restore and manage fen
ecosystems, the efforts need to be coordinated at the scales compatible with the scale of
5

hydrological regime in which the fen occurs. Such coordination may be greatly helped if
ecologists, hydrologists and managers involved in the process know the science behind the
following questions:


How do basin wide groundwater systems and hydrogeology correlate to the occurrence of
prairie fens within the landscape?



Where are the local/regional

source-water

areas

for

fens

which must

be

protected/managed to ensure that fens keep receiving the influx of required quality and
quantity of groundwater?


What are the natural delivery mechanisms of groundwater from source areas to the fens
sites in different hydrogeological settings which must not be disrupted within a region to
ensure uninterrupted supply of groundwater to the fens?

REVIEW OF PRIOR WORK
Historical Perspective
Geologists and natural scientists have been wondering over the quality, quantity and
sources of water to the wetlands since the early 19

th

century. Dau (1823) observed some

wetlands (bogs) were fed „from mainly rain and dew of heaven‟ and „not fed by earth‟.
Dependence of vegetation on quantity of water flowing through the wetlands was observed by
C.A. Weber in 1902 (Couwenberg et al. 2002). Du Rietz (1949; 1954) differentiated bogs and
fens on chemistry of their water and observed dependence of vegetation on pH. Gorham (1957)
discussed major contributing factors responsible for the occurrence of fens which include
climate, geology, topography and biology. He suggested that geological factors may dominate
the other factors and discussed the importance of extreme fluctuations of water. He pointed that
6

climate sets the broad limits within which peat formation [including fens] takes place and it may
be considered first, but also pointed out that while climate sets the broad limits, the local
moisture conditions and salts leached to the surface are controlled by relief. Less porous till of
the hills and plateaus favors fens influenced by mineral soil water. The rainwater and its
composition after percolation through mineral soils materials become different, as it picks up
calcium and magnesium bicarbonates, becoming less acidic favoring the formation of fens. Such
early works led to the classification of fens as rare ecological systems, their dependence on
groundwater hydrology, and their relationship with the local and regional landscapes.
The current literature primarily discriminates prairie fens from other wetlands for their
dependence on stable and calcareous groundwater contributions, necessary to maintain saturated
substrate conditions throughout the year (Bedford and Godwin 2003; Spieles et al. 1999).
Hydrology has been recognized as one the most important factors for fen development and its
biological functions (Bowles et al. 2005; Dekker et al. 2005; Harvey et al. 2007a; Harvey et al.
2007b; SNH 2010).
Need for Multi-scale Thinking
In particular, many researchers pointed out that hydrology is a process and fens are only a
small part of larger groundwater systems. Batelaan (2003) stressed that knowledge of
groundwater flow at both local and regional scales is a pre-requisite for sustainable land and
water management. By combing the synergy of hydrologic modeling and vegetation mapping, he
stressed the need for protection of ecologically valuable areas and land-use that would support
integrated water management. Wassen et al. (1990) stressed that hydrological research that
integrates the results of regional and local studies is essential if the ecology of fen ecosystems is
to be understood. Grootjans et al. (2006) discussed that small groundwater-fed systems are very
7

vulnerable to hydrological changes and climate change due to the lowering of groundwater tables
in the vicinity, and stressed the need to study the hydrological systems that stabilize the site
conditions.
Siegel and others (Siegel 1988; Variano et al. 2009) discussed how local flow systems
controlled by the smaller scale features may be modified by flow systems at larger scales.
Variano et al. (2009) showed through kilometer-scale tracer studies in Everglades how basinscale forcing form water management structures/operation can override the effects of local
landscape features in guiding the flow. Siegel suggested that evaluation of recharge-discharge
function for wetlands is one of the most difficult things because of the interactions of multiple
groundwater flow systems. Many authors have emphasized the need for a broader approach at
catchment/landscape scale to understand upland-wetland linkages and the hydrological systems
(Bedford 1999; Grootjans et al. 2006; Price et al. 2005). Wassen et al. (1990) suggested that
extensive fen systems sustained by large hydrological systems are less vulnerable to local
drainage, but large scale interference with the regional hydrology eventually has split up the
large fen systems into many small fens, each with a small local hydrological system. More recent
publications also maintain the view that most small fens are the remnants of once more extensive
systems (SNH 2010).
Gorham (1957) suggested topography or relief dictates the flow of nutrients leached from
the soil and argued that while geological factors may almost entirely suppress other factors,
climate sets the broad limits within which the formation of fens may occur. Winter (1998) argued
that wetlands that receive discharge from regional groundwater flow systems will continue to
receive that discharge even if the wetland depression in the upland is drained. He also suggested

8

that modification of groundwater discharge areas generally have less significant impact on
wetlands than modification of recharge areas.
Some authors have strongly argued for shifting the focus from merely number of hectors
lost or degraded to the alteration in environmental and hydrological variables at regional scale
that cause ecosystems to form and be maintained through time (Almendinger and Leete 1998a;
Bedford 1999; GLPF 1998). Almendinger and Leete (1998a) separated the fen hydrogeology at
local and regional scales. They included factors such as climate, geologic deposits, regional
landforms, groundwater movement from areas of recharge to fen location, deep vertical
groundwater gradients, and groundwater age as the regional factors. They concluded that rate
vegetation in fens rely on the integrity of regional groundwater flow system up-gradient from the
fens. Bedford (1999) stressed the hydrogeological understanding of wetland-landscape linkages
and argued that restorations should begin with a cumulative impact analysis for the entire region
in which restoration is proposed.
Many large scale studies had been conducted over vast regional areas to establish the
connections between the occurrence of fens and regional systems. Amon et al. (2002) studied
more than 70 fens sites covering 11 states in the Midwestern United States, and 13 in other
regions including Alaska, Canada and Ireland. This study described how different fens were
sustained by forces of climate, landscape, and geology, which permit groundwater to seep
continuously into the root zone at a focused location. This study also presented hypothetical flow
mechanisms of groundwater interactions in the immediate vicinity of fens. Halsey et al. (1997)
presented a comprehensive study for the fens in the Canadian province of Manitoba. This study
correlated climate, topography, soil texture and bedrock geology to the occurrence of fen over
contiguous regional scale of the province and emphasized the importance of integrity of the
9

overall system to maintain the health of all fens. Meyboom (1967) discussed occurrence of fens
in hummocky topography of glacial moraines, conceptualizing, in a qualitative way, the
dependence of fens on multiple scales of shallow/deep groundwater systems, and concluded that
groundwater discharge areas in the landscape where fens occur, would have their sources linked
to one scale of flow system or the other. Bedford (1999) suggested the likelihood that a wetland
receives ground water from more than one flow systems can be deduced from its topographic
position in the watershed. Such large scale studies, involving large number of fen sites, however,
has not been investigated the sub-surface flow regime at regional scales (perhaps because of data
limitations) and its connections with the occurrence of fen sites.
Many authors, nevertheless, have hypothesized and/or investigated connections of fen
sites with the regional groundwater systems which may be crucial to the existence of particular
fen sites. Winter (1998) suggested that the seriousness of the impact on an ecosystem is
commonly related to scale. He argued that it may be appropriate to assume that regional
groundwater flow systems are recharged in uplands and discharged in lowlands, however, a
similar assumption commonly does not apply on a local scale, because of the spatial and
temporal dynamics of groundwater recharge. Siegel and Glaser (1987) in their study on Lost
River peat land in northern Minnesota identified at a beach ridge 10km from the study site as the
“probable” recharge zone for the groundwater seeping in peat land and regarded it as regional
connection. Most of their work was based on observed groundwater levels and chemistry in the
observation wells within the peat lands. They also concluded that small changes in volume of
groundwater discharge in peat land may significantly affect the vegetation. Almendinger and
Leete (1998b) reported „local and regional‟ hydrology of six fen sites in Minnesota. The local
scale hydrology at these fens was taken as the peat land-forms and the substrate within a few

10

hundreds of meters of the sites, while the regional hydrology for these sites was investigated
between 2~6 km. Connections of source water areas to the fen sites were identified but these
connections were based only on general landforms and geology in the area rather than detailed
hydrological data from the source water areas. Stein et al. (2004) conducted a regional study for
subsurface hydrology of 20 slope wetlands concentrated in the central portion of the San Juan
Creek Watershed, within an area of approximately 12 x 5 km. “Shallow” and “deep” systems
were investigated at 1m and 15m depths respectively. Vertical flow dynamics discussed for
various fen sites were investigated based on site geology. Groundwater flow along the vertical
profiles was illustrated using hydrologic intuition and not the actual data. Inferences were drawn
for the “probable recharge mechanisms” at regional scale based on groundwater monitoring only
within the local sites. Recharge zones within the watershed (approximately 30 x 35 km)
extending beyond the study area were assumed as the „probable‟ source water areas for the
wetlands. The study concluded that occurrence of wetland was dictated by local and regional
geologic settings and sustained by both local and regional groundwater sources.
Need for a Quantitative Framework
Managing the flow regime as single system at large scales is gaining popularity as a
preferred approach in many parts of the world, across the political divides and trans-boundary
basins (EU 2000; IJC 2009; NSW CMA 2004; SOGW 2009); however, there are significant
challenges in the implementation of this concept.
Groundwater models had been used as tool to systematically investigate fen hydrology at
many studies and to guide data collection for further monitoring. However, most existing models
lack sufficient resolution, are limited to one particular scale or a narrow range of scales, and/or

11

are often prohibitively expensive to apply in a system-oriented management setting that requires
modeling integrated processes and large numbers of fen sites (Price et al. 2005)
Batelaan et al. (2003) combined vegetation mapping and groundwater modeling to
identify discharge areas and delineate corresponding recharge areas respectively. Their models
were based on particle tracking in MODFLOW. The models were built on regional scale and
primarily delineated horizontal flow dynamics between expected source water areas and the
groundwater discharge zones. Bradbury and Cobb (2008) delineated groundwater contribution
areas of 11 springs in Wisconsin using reverse particle tracking in horizontal groundwater
models using MODFLOW. The scale of these models ranged from less than a mile to a couple of
miles. Gilvear et al. (1993) modeled regional and local flow dynamics for Bradley Moore fen in
UK. Their regional model, 17 x 8 km, was used to determine boundary conditions for the local
scale model. Vertical flow dynamics were conceptualized in the local model. Using MODFLOW
they built a 3-layer model to resolve vertical flow at local scale and used chemistry data from
piezometers to interpret results for more detailed vertical dynamics and to relate discharge zones
to corresponding recharge zones.
Grapes et al.(2006) used a simple one layer regional-scale model using MODFLOW for
Lambourn Valley, UK. They inferred local scale dynamics through site specific data. Combing
results from regional model and local data they demonstrate the use of the results of catchment scale ground water modeling to determine the hydrodynamics of „local‟ floodplain wetlands.
Their results demonstrate the importance of taking catchment-scale water flow into account
when managing isolated wetlands in a permeable catchment.
Acerman (2004) and Acreman and Miller (2006) suggested that one of the first steps in
understanding wetland hydrology should involve identifying which water transfer mechanisms
12

are operating at the wetland and which of these are the most important in maintaining the present
ecology. They also emphasized the need to understand groundwater interactions in a third
dimension, i.e., looking at vertical cross sections. Various studies have described how water may
be delivered from the source water area to a fen/wetland employing the vertical visualization
(Acreman 2004; Acreman and Miller 2006; Dall'O' et al. 2001; Gilvear et al. 1993; Toth 1972;
Winter 1998), but most of these studies have either employed hypothetical models and/or have
used hydrogeologic theories/intuition to define these mechanisms. Some authors have employed
use of vertical profile models to delineate vertical flow paths in more details. Most of these
studies have involved regional scales, and consequently have not studied more than a couple of
sites, perhaps because of data limitations. Such studies are therefore mostly site-specific. It may
be difficult to draw generalized conclusions from these.
Bleuten et al. (2006) used MODPATH to model vertical profile dynamics of groundwater
for Ob River Terrace Fen in Russia. Vertical flow dynamics in this model were calculated by
using groundwater heads from piezometers along a 13 km long profile as upper boundary
condition and calculating flow lines in vertical profile. Horizontal flow pattern was not modeled
but assumed parallel to the profile based on general groundwater flow gradients in the region.
van Belle et al. (2006) used FLOWNET to build two vertical profile models in Het Hol, The
Netherlands, using hydraulic head data along two transects as the top boundary condition. The
profiles were approximately 10 km each. The flow pattern delineated by the model was used to
infer discharge areas along the profiles. The groundwater system at the regional scale was not
modeled in the horizontal plain to determine if the profiles also aligned with the horizontal flow
paths. Loon et al. (2009) studied a large regional area along River Vecht, The Netherlands, to
investigate historical changes in groundwater regime and to delineate current source water areas

13

for fens in the region. They modeled vertical flow paths along two pre-determined profiles to
conclude that fens were being fed from the regional sources.
Wessen et al. (1990) investigated groundwater connections between the regional
groundwater flow system at an individual fen site in central Netherlands. They concluded that
nested groundwater flow systems existed within the fens and that regional and sub-regional
groundwater flow systems strongly influenced the local hydrology of the fens. This study is an
exception from most others in that groundwater data was used at both local and regional scales.
Vertical groundwater flow was also modeled in this study for a transect along regional
groundwater flow path using FLOWNET. However, this study was undertaken in a highly
modified landscape where surface water levels in polders were extensively available for the
artificially controlled regional hydrology. The techniques presented in this study are useful, but
its results may not be generalized for the fens occurring in more natural settings.
Challenges Facing Multiple Scale Thinking on Flow Regime Integrity
The following bullets provide a summary of specific data-related modeling challenges:


The river basins are large and require large volumes of hydrological, geological, landuse, soils, topography data to effectively characterize their flow regimes.



Collecting data is expensive and time consuming.



It is difficult to organize a coordinated data collection program at large scales, especially
when a basin is divided in multiple administrative jurisdictions.



Since data requirements are diverse between various agencies, it is difficult to collect,
utilize and share data in a synergistic way.

14



Many historical records exist in hard prints and extremely tedious to use. These records
may be lying in bits and pieces with different agencies and difficult to gather.



Data converted to or collected in electronic format by different agencies can have format
compatibility issues, requiring tedious preprocessing before they can be used.



Strict and standard QA/QC procedures had mostly been absent at the time many
historical records were collected.



Desired density and uniform coverage to characterize a basin scale system may not be
available in the existing data sets.

These issues make it difficult to access data even when it is available, requiring tedious
pre-processing before it could be effectively utilized.
Currently the subsurface hydrology for most of the 150 known prairie fen sites in
Michigan is not well understood. Some of these sites make candidates for potential
reintroduction sites for MS populations but the source water areas are not clearly known for any
of these sites. While identification of regional sources of groundwater would be pivotal for
prairie fens management, the understanding of local scale dynamics and delivery mechanisms of
groundwater to at the individual fens sites are also crucial for their effective management.

AIM AND OBJECTIVES
Recent advances in GIS, GPS, and Remote Sensing have, however, opened new
possibilities. Michigan Public Act 148 (2003) required the Department of Environmental Quality
(DEQ) to create a “groundwater inventory and map”. Water well drillers‟ well records database,
called Wellogic, and the Groundwater Inventory and Map Project database, called GWIM,
turned out as the outcomes of the Public Act. These geo-referenced databases, besides other data,
15

also contain information on hydrological and other properties of the natural systems necessary to
construct conceptual groundwater models almost anywhere in the state. When systematically
utilized, this vast data source can significantly enhance our ability to characterize and model
Michigan‟s groundwater systems

The aim of this research is to enhance our understanding of prairie fen hydrology that
may be useful for conservation, protection and management of prairie fens.

The specific objectives of the research are:


to determine groundwater flow regime at the statewide scale around the 150 known
prairie fens sites in southern Michigan and to correlate the occurrence of prairie fens with
the basin-wide hydrological landscape;



to identify source water areas for selected fen sites at local, sub-regional and regional
scales in varying hydrogeological settings;



to investigate delivery mechanisms of groundwater from the source areas to the selected
fen sites; and,

 assess implications of hydrologic understanding on fens protection, conservation and
management.

16

Chapter 2

Methodology

OVERVIEW
Data-driven groundwater modeling approach adopted in this research directly analyzes,
filters, and processes water well records for static water elevations and hydraulic conductivity to
estimate groundwater velocity field. Hydraulic head is mapped using static water levels data.
Hydraulic conductivity estimates are derived from lithologic information. Groundwater velocity
is computed based on preprocessed hydraulic head and conductivity using Darcy‟s equation
(2.1):

2.1

Where

= velocity of groundwater flow
η = effective porosity
= hydraulic conductivity
= hydraulic head
= divergence of h or the hydraulic gradient

The approach capitalizes on recent national movement in geospatial data integration and
Michigan‟s statewide groundwater database (e.g. State of Michigan (2006),(2009), MNFI
(2010), GWIM (2006)). The approach also takes advantage of the fact that prairie fens occur in
topographically complex areas characterized by the presence of large numbers of scattered lakes,

17

wetlands, ponds, swamps, creeks, and drains that can be used approximately as “natural
piezometers” for shallow groundwater systems.
One of the most significant advantages of a data driven modeling approach lies in that the
data is available everywhere across the state in sufficient density to model groundwater systems
around most fen sites. The disadvantage, however, is that data interpretation alone may not
always allow transient simulations and estimating of water budgets. And since the models are not
process based, they may be directly employed for scenario testing.
Three types of groundwater models were built:


Statewide scale model – covering the entire southern Michigan basin;



Regional scale models – around the selected fen sites/clusters extending from 10 to 40
km;



Vertical profile models – across the selected profiles within the regional models with
profile lengths varying between 10 to 30 km

In this study we preferred data-driven groundwater modeling approach over the
traditional groundwater modeling approach which may allow us to build a full 3D numerical
model to resolve all the complexities in the system. However, if any such model is built, it would
be data hungry with a very complex structure. Moreover, this model would have to resolve flow
dynamics at multiple scales which would make the model even more complex.

STATEWIDE MODEL
At the statewide scale, the subsurface groundwater flow regime was delineated by
interpolating static groundwater level data acquired from water well records to identify

18

groundwater sources of regional significance. The 150 prairie fen sites across the state were
superimposed over the delineated pattern of groundwater to correlate their occurrence with the
pattern of subsurface flow regime. Furthermore, the occurrence of fens and subsurface flow
regime were also compared with the glacial landforms, surface hydrology and topographic data.
Inferences on the occurrence of fens were drawn based on these comparisons.

REGIONAL MODELS
Areas for regional scale modeling were selected around the groundwater sources of
regional significance identified at the statewide scale. The subsurface groundwater heads were
delineated for shallow, drift and bedrock aquifers. Specifically:


elevations from all surface water bodies and wetlands are used to delineate local, shallow
groundwater head patterns around fens;



water well records and sampled surface water elevations from major streams and lakes
are used to delineate regional head in the glacial aquifer; and,



records from water wells are used to delineate regional groundwater head in the bedrock
aquifer.

Finally, groundwater profile models were built for selected sites to investigate vertical
dynamics of subsurface flow in shallow and deep groundwater systems.
Data-Driven Characterization of Groundwater Heads in Shallow System
Data-driven characterization of shallow groundwater heads involves the estimation of
shallow groundwater table

and its gradient

in equation 2.1, using the data from the

19

statewide databases. However, enough directly measured data for shallow groundwater heads
were not available from the statewide databases because:


there were very few existing water wells records located within or near most fen locations
as they are generally protected and situated away from population centers;



even the few wells closer to the sites in the database did not adequately represent shallow
groundwater system as the wells are generally screened in the deeper aquifers and do not
represent the shallow system; and,



even when shallow monitoring piezometers at specific sites do exist they are generally
too few and too sparse to characterize the larger shallow system at regional scale.

Following assumptions were made with regards to heads in shallow groundwater systems
in order to manipulate surface water data to delineate shallow groundwater heads:


All streams are connected to the shallow system.



All lakes are connected to the shallow system.



All wetlands are connected to the shallow system.



Shallow groundwater head, even in the wetlands which are apparently dry, would be very
close to (or almost at) the ground surface.

Some wetlands may be perched and may not be connected to the shallow system
altogether. The available database on wetlands did not have this information so these could not
be automatically filtered out. The aim of using wetlands as part of shallow systems was only to
generate lots of „approximate‟ data. However, the uncertainty introduced by not being able to
filter out perched wetlands was mitigated by the following:

20



Since most fens happen to occur in coarser outwash materials, the wetlands in the vicinity
are less likely to be perched due to less pervious layer within such materials



Lithology data from well logs were used ascertain clay layers in the strata. Where
continuous clay layers were found, they were incorporated in the profile models.
Wetlands occurring on top of such layers were hence isolated from the system

Shallow water level in surface water features such as lakes, wetlands and streams, was
treated as intersection of shallow groundwater surface with the surface topography. Scatter
points were generated in the surface water features and their elevations were estimated using
30m/10m DEM raster data (NED USGS 2006). The density of points generated in surface water
features across the model domain was greatly helped by the presence of relatively large number
of wetlands, streams and lakes within the vicinity of the fen sites as shown in Figure 2.1 .
Water levels at the scatter points were interpolated using kriging and variogram modeling
as shown in Figure 2.2. Gaussian models were fit to the variogram data by adjusting nugget,
range and variance. Nugget of at least 2m was used in all variogram models to cater for „noise‟
in the data. The data generated for shallow groundwater was inherently noisy because of:


inaccuracy in DEM (most locations with 10m DEM were actually using 30m DEM
interpolated at 10m);



some wetlands not being wet at the surface;



perched wetlands not connected with the aquifer; and,



fluctuations in stage in lakes and streams.

21

Model Area

Fen

2819 Points

Figure 2.1

0

Scatter

points

(lakes,

streams

2500

generated
and

in

wetlands)

5000m

surface
to

water

characterize

features
shallow

groundwater head using 30m/10m DEM around a fen site.

22

Variogram
140

100

60

20
0
0

2000

4000

6000m

Lag Distance

Figure 2.2

Gaussian model fit to the variogram

Between 0 and 5 passes of smoothing, as explained on page 58, were applied to the
interpolated surface. A comparison of kriging surface without smoothing and 3 and 5 passes of
smoothing respectively are shown in Figure 2.3. It may be noted that up to five passes of
smoothing only removes „kinks‟ from the initial contours without significantly altering the
interpolated surface. The final shallow groundwater surface used in a model using 3 passes of
smoothing after kriging interpolation is shown in Figure 2.4.

23

No smoothing applied

4km

3x smoothing applied

5x smoothing applied

Figure 2.3

Contours at 2.12m interval represent shallow groundwater
surface without smoothing (top) and with 3 and 5 passes
(mid and bottom) respectively
24

Fen

4 KM

0

Figure 2.4

The relief and contours around fen sites represent the
shallow groundwater surface interpolated from the data
generated in surface water features (lakes, streams and
wetlands). Contours at 2.12m interval show shallow water
table

varying

in

elevation

from

176.636m

to

239.196m.

Contour interval is 3m.

Data-driven Characterization of Groundwater Heads in Glacial Drift
Data-driven characterization of groundwater heads in glacial drift implies the estimation
deeper groundwater heads

and its gradient

in equation 2.1, using the data from the

25

statewide water well records, based on the following assumptions for deeper groundwater
systems:


SWL data (Wellogic) from the water wells screened in the glacial drift represent
groundwater head



in deep glacial aquifer;

all third and higher order streams (NHD) are connected to deep groundwater flow
systems within glacial drift; and,



all large inland lakes greater than 100,000 square meters (NHD) are connected to deep
groundwater flow systems within glacial drift.

Figure 2.5 illustrates an example of regional flow pattern derived for an area 12 km by 10
km around a fen site.
Comparison of shallow groundwater heads of Figure 2.4 with the regional ones of Figure
2.5 is shown in Figure 2.6. The differences in shallow and deep systems can be seen here. This
was an expected outcome based on the multi-scale conceptualization presented in Figure 1.2.

26

Fen site

4 KM

0

Figure 2.5

Shaded relief and contours around fen sites represent the
regional groundwater surface interpolated from the water
wells data. Black dots mark the data points. Contours
show regional groundwater heads varying in elevation from
177.5936m to 218.419m. Contour interval is 2m.

27

4km

Figure 2.6

Groundwater head contour maps for deep system (top) and
shallow system (bottom)

28

Data-driven Characterization of Groundwater Heads in Bedrock
Data-driven characterization of groundwater heads in bedrock involves the estimation
groundwater heads

and its gradient

for the bedrock aquifer in equation 2.1, using the data

from the statewide water well records, based on the following assumptions:


SWL data (Wellogic) from the water wells drilled in the bedrock represent groundwater
head



in bedrock aquifer; and,

where bedrock are very close to the surface, the groundwater flow systems in the bedrock
are connected to large surface water features, i.e., third and higher order streams and
lakes greater than 100,000 square meters.

Water level data for bedrock wells in Wellogic was interpolated to determine h for each
cell in the model. In case of shallow bedrock areas, large surface water features were also taken
into account.
Estimating Hydraulic Conductivity
Lithology information for the glacial drift was available in the wells data. Average
horizontal hydraulic conductivity of the glacial drift, Kh, at each well location was estimated by
taking the arithmetic mean of the conductivities of all lithology layers in the sequence (GWIM
2006) using equation 2.2.

2.2

Where B = Total depth of the drilled well
th

i = i number of lithology record

29

n = total number of lithology records in the sequence
th

= assumed horizontal hydraulic conductivity of the i lithology record
th

= thickness of the i lithology record
estimate was based on the type of material recorded in the driller‟s log and the type of
glacial landform within which the well was drilled (GWIM 2006). Water well records that had
usable lithology data were extracted from the Wellogic database. The glacial land system
associated with each well location was determined using GIS overlay techniques. For each well,
a hydraulic conductivity value was assigned to each lithology layer based on the type of material
reported by the driller, glacial land system, and range of conductivity values of the given
material in literature (GWIM 2006). Values of B and

were available within the well record.

Using these values equation 2.2 was solved for each well location and hence an estimate for
was obtained at each data point. Lithology based

values for glacial drift were further

processed at statewide scale to create statewide raster at 540m resolution raster for drift

using

kriging interpolation (Oztan 2010). Glacial landform polygons were not enforced as boundaries
while interpolating for

across the state.

Similar to raster maps for the conductivity of glacial drift, statewide raster maps for the
transmissivity

of bedrock formations were produced at 500m resolution using the specific

capacity test data from bedrock wells across the state (Oztan 2010). The specific capacity data
were interpolated within the bedrock formation polygons to estimated T for the respective
bedrock formation. Kh for model areas was estimated using equation 2.3.

30

2.3
Where B = Assumed thickness of bedrock in the model
In order to determine horizontal hydraulic conductivity

for the drift aquifers within a

selected model area, the values of Kh were read from the 540m conductivity raster map and
interpolated for each cell in the model. In order to determine horizontal hydraulic conductivity
for the bedrock aquifers, the values of

were read from the existing 500m conductivity

raster map, interpolated for each cell in the model and converted to

using equation 2.3.

Estimating Porosity
No data is available for the aquifer‟s effective porosity η in the databases, however, there
had been studies in southern Michigan reporting value of η for the glacial drift to vary between
0.15 and 0.30 in southern Michigan (Lowry and Anderson 2006). Estimated net porosity

of

sandstone bedrock in southern Michigan was reported as 0.115 in other geological works (Barnes
et al. 2009). The upper Marshall formation [where most of the water wells are drilled] is a
medium to coarse grained porous sandstone (Stearns 1993). Based on these studies, value of
η=0.1 was assumed for both glacial drift and bedrock formations in 2.1.

Tracking Source Water in Horizontal Plane
Regional capture zones were delineated for selected fen sites using the reverse particle
tracking on the static water level surface generated for the regional system. Instead of running
the simulations for specified time periods, the reverse tracking was allowed to run until the flow
paths hit one or more regional sources. An example of regional capture zone for a fen site is
shown in Figure 2.7, showing the flow paths traced back to a sub-regional groundwater mound..
31

It does not necessarily mean that all flow paths within the capture zone are draining into the fen
site. Some flow paths within this zone may up-well into the fen and some may up-well into the
wetlands and lakes within the capture zone while still some other paths may completely bypass
fen underneath the fen surface and end up in a regional sink. The groundwater flow system is
truly 3D in areas where fens occur and is not fully captured in this 2D representation of capture
zone. Profile models across the capture zones, as explained in succeeding sections, will be used
to improve upon our understanding of vertical capture zones. The conceptual flow regime model
presented in Figure 1.2 may also be kept in mind.

Regional
groundwater
capture zone

Fen site

4 KM

0

Figure 2.7

Groundwater capture zone delineation for a fen site based
on

reverse

particle

tracking.

represent

regional

groundwater

regional

groundwater

heads

The

relief

surface.

varying

in

and

Contours

show

elevation

from

177.5936m to 218.419m. Contour interval is 2m.
32

contours

VERTICAL PROFILE MODELS FOR GROUNDWATER SYSTEMS
Groundwater capture zones delineated using the horizontal flow system as shown in
Figure 2.7 only show the possible flow paths that may lead to the fen site in plan view. As
illustrated in Figure 2.5 and Figure 2.4, groundwater flow gradients for regional and shallow
groundwater systems may be different from each other. In a full 3D sense, many of these flow
paths may bypass the fen by flowing underneath the fen or up-well in other water bodies within
the captures zone before they can reach the fen. To fully resolve 3D complexities of the system
we require full 3D models, but full 3D models a data hungry and expensive. Using a 2D vertical
profile model through the captures can give some insights into the vertical dynamics of the flow
system without having to build a 3D model. 2D profile models presented in this study, should, in
no way be taken as a substitute to full 3D models, because of their inherent limitation as
discussed ahead. The model may only be used for gaining insights rather than making
predictions and must the interpreted with their limitations in view. The conceptualization of 2D
multi-scale flow regime is already presented in Figure 1.2. The profile models were built to
qualitatively gain insights into the shallow and deep groundwater flow systems and the likely
delivery mechanisms of groundwater to the fen locations. Similar insights had been provided by
others using mostly hypothetical approaches (Acreman 2004; Amon et al. 2002). The profile
models in this employ the estimated groundwater table as a forcing function on the top model
boundary and use process based simulations to track the vertical flow paths. The profile models
were used primarily to gain insights on 3D flow dynamics and not for making predictions.
Modeling Assumptions
The profile models were built to resolve the vertical dynamics of groundwater flow with
the following assumptions:
33



Water table, taken as prescribed head, drives the flow in the system.



No flow takes place across the profile section‟s longitudinal boundaries.

Water table from the shallow system was enforced as top boundary condition and the
models were solved for h in the profile. Estimation of other boundary conditions and parameters
are explained in succeeding paragraphs.
The profile sections for models were selected along the regional flow paths from the
deeper systems. However, shallow groundwater flow direction may not always be parallel to the
deeper one. The latter of the above assumptions is, therefore, more valid for the deeper parts of
the profile and less so closer to the surface. This limitation must be kept in view while
interpreting the profile models.
Boundary Conditions for Profile Models
A profile was selected through the regional capture zone delineated for a prairie fen site
as shown in Figure 2.8. In most profile models built in the study, both ends of the profile were
kept at a location coinciding with a groundwater divide to obtain a fully contained system with
relatively more robust vertical boundaries. Where this was not possible, remote conditions
(usually constant head) were assumed along the vertical boundaries.

34

Extent of profile model
extended to a regional
source (groundwater
mound)

Extent of profile model
extended to a regional
sink (River)

4 KM

0

Capture zone
based on regional
groundwater regime

Figure 2.8

Profile section

A profile sections is selected between a regional source
and a regional sink, intersecting the fen site and its
regional groundwater capture zone. Relief and contours in
the figure represent shallow groundwater surface or the
water

table.

Contours

show

regional

groundwater

heads

varying in elevation from 177.5936m to 218.419m. Contour
interval is 2m.

Top boundary of the model was defined by shallow water table taken as prescribed head.
The depth of the profile models extended down to impervious bedrock formation (mostly shale)
to achieve relatively more robust no-flow boundary at the base of the profile. Figure 2.9 shows
the typical boundary conditions used for most profile models.

35

Prescribed head
(Water table)

No-flow
(Regional sink)

No-flow
(Regional Source)

Clay layer
Glacial aquifer
Bedrock aquifer
Projection of
fen site in drift

Figure 2.9

No-flow

Typical boundary conditions for the profile models

Top elevation, or the topographic surface, was estimated along the profile using 30m/10m
DEM raster along the profile. However, this DEM was only used for display purposes and not as
a modeling parameter. An example is shown in Figure 2.10. Bottom of glacial drift, or the top of
bedrock, was estimated using 500m raster for bedrock elevation along the profile. An example is
shown in Figure 2.10
In most cases, only one bedrock layer was encountered under the glacial drift. An
arbitrary minimum thickness of 100ft (30m) was assumed for these bedrocks. In cases more than
one bedrock layers were encountered within 100ft below the drift, the actual thickness of the
upper layer from 500m raster was used.
Hydraulic Conductivity
For all profile models presented in this study, mean hydraulic conductivity for the glacial drift
was taken as 50ft/day and for Marshall Sandstone 250ft/day. Clay lenses within the drift and

36

Coldwater Shale in the bedrocks were treated as less transmissive formation with very low
conductivities of 0.001ft/day and 0.0001ft/day respectively. These values were not based on data
but used primarily to represent order of magnitude contrast in different materials. The contrast
does not affect the qualitative assessment of vertical flow. Profile models were not used in any
kind of quantitative assessment in this study. Sensitivity analysis for variability in hydraulic
conductivity was also performed and explained in the next section.
Vertical Anisotropy
No data was available on the vertical anisotropy for the glacial drift or the bedrocks in the study
area. Vertical anisotropy is one of the least known parameters in groundwater flow (Winter
1978). Gillhan and Farvolden (1974) tested this ratio and found it particularly sensitive in areas
of recharge and discharge. Freeze (1969) concluded that
uncommon. Vecchioli and others (1974) found

values greater than 100 are not
= 500 in a drift section of Long Island,

New York. Bennett and Giusti (1971) showed that the ratio had to be close to 1000 for their
simulations to match data in the coastal plains of Puerto Rico. Siegel and Glaser (2006) reviewed
the values for KH being 1-2 orders of magnitude greater than KV. Vertical profile models were
found very sensitive to the anisotropy ratio. After testing many different values for anisotropy
ratios, we concluded that

= 1000 for glacial drift and

= 500 for transmissive

bedrocks matched well with the data for most situations. Final profile models presented in this
research are based on these values. Sensitivity analysis for Anisotropy was also performed and
explained in the next section.

37

Estimating Water Table along the Profile
Groundwater level along the profile was taken from the shallow groundwater surface
generated for the shallow system. An example is shown in Figure 2.10

Topographic profile

Water table
Clay layer

Drift

Projection of fen site in drift

Figure 2.10

Boundaries,

zones,

and

water

table

Bedrock

estimated

for

a

profile model.

Estimating Discontinuities within the Glacial Drift
Clay lenses or continuous clay layers within the glacial drift were estimated from the lithology
logs in the water wells data. The available water wells along the selected profiles were extracted
and plotted as shown in Figure 2.11. Sensitivity analysis for discontinuities was also performed
and explained in the next section.

38

4km
Estimated
continuous clay

Clay in well logs

Figure 2.11

Water well logs around the profile (above) projected in
vertical plane (bottom) to show clay occurrences

Clay layers recorded in the well logs were shown with darker color which gave an idea of
occurrence of clay within the drift. Clay lenses or continuous confining clay units were identified
based on subjective judgment from these logs.
Tracking Source Water in Vertical Plane
Profile models based on above mentioned parameters were built for the selected sites to
understand how deeper and shallower groundwater systems deliver water to the sites. Forward
and reverse particle tracking techniques were used to trace flow paths from local and regional
systems and to find where the water was coming from to the fen sites. An illustration of profile
model with flow paths is shown in Figure 2.12.
39

Local source water
area for the fen

Regional source water
area for the fen

Clay layer

250m

Elev

Fen

125m
9,222 m

Figure 2.12

Profile model showing the groundwater flow paths (pink
lines) leading to the fen site (purple lines). Light blue
lines show groundwater head contours at 1.84m interval.
Vertical exaggeration in this figure and all other models
presented in this study is 10x, unless noted otherwise.

SENSITIVITY ANALYSIS
Hydraulic Conductivity
A number of models were analyzed using variable K field and constant K field. It was
observed that the qualitative delineation of flow paths did not significantly alter when complex
variable K fields were used. For final presentation in this study, uniform K fields were used in all
models. One of the sensitivity results is shown in Figure 2.13 to as illustration.

9,222 m

Figure 2.13

Variable (top) versus uniform K (bottom) field in glacial
drift profile. Orange flow paths are fen capture zones.
40

Vertical Anisotropy
Different values of

were analyzed for profile models and the results were

compared qualitatively to match with the regional flow patterns. Higher values of
produced too many closed systems along the profile where the regional flow systems from data
interpolation had suggested continuous flow gradients. By reducing the

ratio, the flow

patterns in profile models conformed to regional patterns in horizontal flow models. The
illustration in Figure 2.14 shows

varying from 1/10 through 1/500 for a glacial outwash

aquifer. Final anisotropy ratio for glacial outwash for all models presented in this study is 1/500.

Figure 2.14

Sensitivity

to

vertical

anisotropy.

Orange

and

purple

flow paths represent capture zone for the fen sites

Discontinuities in Drift
Well logs were analyzed for each site to indentify less pervious materials in the lithology.
Some sites showed very clear patterns of continuous clay layers where as in most sites the
pattern revealed by well logs was random. In order to assess the impact of scattered
discontinuities within the drift layer several profile models with scattered discontinuities were
41

compared with models without scattered discontinuities. The comparison revealed that in most
cases, scattered discontinuities did not significantly alter the flow pattern. Consequently, all
models presented in this study are without scattered discontinuities. However, where continuous
discontinuities were ascertained from well logs, these were duly incorporated in the profile
models. Figure 2.15 and Figure 2.16 show the well logs in profile and scattered discontinuities
inferred from these logs. The figures also show the comparison of vertical profile models with
and without discontinuities. In some cases, as in Figure 2.15, the discontinuities qualitatively
impacted the flow paths leading to the fens, but in most cases, as in Figure 2.16, that was not the
case.

9,222 m

Figure 2.15

Effect of scattered discontinuities at Paw Paw Prairie
Fen site. Orange flow paths indication fen capture zones.

42

3,783 m

Figure 2.16

Effect of scattered discontinuities at Paw Paw Lake Fen
site. Orange flow paths indication fen capture zones.

The impact of discontinuities in a vertical profile may become exaggerated because flow
lines may go around the discontinuity in the transverse direction which is not captured by the
profile model. Consequently, impact of small discontinuities was not taken in consideration for
43

all the models presented in this study because qualitative assessment of flow paths presented by
these models was not being significantly impacted even if this additional complexity was added
to the models.

COMPARING WITH TRADITIONAL APPROACH
In the traditional groundwater modeling approach, the models are built and calibrated to
establish the current or existing conditions. This is a challenging task because most natural
systems are very complex, and multiple scales in fen hydrology adds yet another layer of
complexity to the system. Modeling the whole system requires knowledge about every
component, parameter and boundary conditions in the system. The resulting models, therefore,
are complicated and data intensive. Not only the data is expensive to collect, it is seldom the
case, even in „data-rich‟ situations, that modelers would have complete data for all the
parameters in the system. The unknown parameters in the model, therefore, have to be estimated
using inverse modeling techniques or intelligent guessing. Adhering to the principals of
parsimony while building the traditional models can reduce the data requirements and
uncertainty associated with parameters uncertainty, but adds to the uncertainty associated with
the structural uncertainty is illustrated by Silberstein (2006), where structural uncertainty is
reduced by increasing complexity but parameter estimation uncertainty is increased.
Advocates of simpler models suggest that simplicity is important to the development of
useful models (Hill 2006). However, models should be able to capture essential complexities of
the natural system, and, in case of fen hydrology, the systems are very complicated and multiscale and cannot be adequately resolved by simpler models. Gómez-Hernández (2006) has
summarized the typical difficulties one faces when using complex models:

44



Data requirements are extensive for making and calibrating the model. Data quality is
also an issue.



Extensive knowledge on aquifer properties is seldom available.



It is very difficult to match all aquifer boundaries with geological boundaries where
conditions are known. Either the boundaries have to be approximated or extended over a
larger area. Approximate boundaries increase uncertainty and far away boundaries
increase model area, and hence increase number of parameters and data requirements.



The more complex a model, the more parameters it has. Unknown parameters are
established through inverse modeling which makes the calibrated models non-unique and
difficult to constrain.



Prediction uncertainties add from data deficiencies (and errors), non-unique calibration
and complex boundary conditions.

The modeling results are always approximate within the bounds of some uncertainty. The
common sources of uncertainty in traditional models are discussed below:
Aquifer Boundaries:
The areas of interest to be modeled rarely conform to natural boundaries of an aquifer
system. If the model is truncated at arbitrary boundaries it introduces errors at the boundaries, if
the model domain is extended to natural boundaries, it increases the model area, thereby
increasing the data requirements and adding more components in the model. This is a dilemma
faced by modelers whenever they have to model an existing groundwater system.

45

Aquifer Stresses:
Background Pumping: Complete knowledge about the location and pumping rates water
wells in an area is seldom known. Their pumping rates are highly variable in time and may have
an impact on small changes at local scales.

Stage in Surface Water Bodies: Stage in surface water bodies keeps fluctuating. It would
seldom be the case that (fluctuating) stage data on all rivers/lakes is available in a regional
setting. Even if stages are measured, the measured levels only represent a time stamp of stage
and could be misleading for the average conditions.

Recharge: Recharge is highly variable over space and time. Even for the most precise
estimates of recharge simplifying assumptions are made. It is extremely difficult and expensive
to make accurate estimates of recharge over space and time. In most cases, uniform recharge
rates are applied to large zones in the groundwater models.

Aquifer Properties:
Aquifer properties such as hydraulic conductivity can vary several orders of magnitude
within a small scale. Storage coefficient similarly can vary significantly within the spatial scales
of the modeling domain. The knowledge about distribution (zones) of geologic materials,
responsible for the variability, is seldom known accurately. Lack of knowledge about these zones
can affect the model structure and required complexity.

Initial Conditions:
It is seldom the case in groundwater modeling that initial conditions are completely
known. Even for a moderately complex system, modeling the absolute initial conditions is data
46

intensive and requires simplifying assumptions. The uncertainty, non-uniqueness and errors
associated with modeling the initial conditions are consequently projected to the predictions
made for the stress scenarios for which the model is used.

ADOPTING DATA DRIVEN MODELING APPROACH
Using lots of low quality/noisy data would yield only approximate results. However,
prior studies have shown that if data from Michigan‟s groundwater databases is processed
correctly, it can closely match the traditional modeling results (Simard 2007). But since
predictions from the traditional models are also approximate, we may not lose much by building
the „approximate‟ data models.
The relationship between data, (i.e., recorded measurements of nature) and process based
traditional models is very complicated. The dilemma of uncertainty in deterministic models, due
to theoretical reasons and structural simplifications, is that the models require calibration against
data, while the data by itself neither exactly follows the theory nor perfectly aligned with the
model structure. But in the presence of abundant data, we can even skip the process of
calibration (Cunge 2003).

Hydrological science had long been divided with two approaches used independently by
the modelers [theory based] and the experimentalists [data based]. Often, the two groups had
little to do with each other (Silberstein 2006). In the current environment of large GIS data bases
and high speed computing, we need models which combine data analysis, visualization and
processing with the theory based simulations. Data-driven approximations can be useful in
problems related to hydrology especially where the relation between dependent and independent
variables are poorly understood(Garcia and Shigidi 2006).
47

Data driven modeling approach adopted in this study gives us the following advantages:


Multiples scales of hydrology can be investigated since the data is available at multiple
scales.



Almost any part of the state can be modeled since data is available everywhere.



Relatively simpler models, compared to the traditional models, afford the opportunity to
concentrate more on understanding the natural processes underlying the fen hydrology
rather than the issues related to numerical modeling of complex systems.
For this study the Michigan‟s groundwater was conceptualized as two or three layered

aquifers system as shown in Figure 2.17. The top layer being the shallow glacial drift aquifer,
fully connected with all surface water features; the middle layer being deeper glacial drift aquifer
connected to large lakes and streams; and, the bottom layer being the bedrock aquifer.
As mentioned earlier, we employed data driven modeling approach to compute h and K
in 2.1. The equation can be also be used to simulate the above mentioned conceptual model
using process based traditional approach in which
groundwater flow while

is obtained by solving the equation for

is calibrated through inverse modeling by matching h with the

observed data. Based on estimated

and calibrated , velocity field is calculated in the equation.

Calibrating a model would require information on sources and sinks, complex processes,
geological boundaries and other parameters such as leakance. Recharge is generally estimated
through inverse modeling. Most traditional models lack enough data on these variables and are
non-unique even after calibration. Conceptual layers can be added to resolve vertical dynamics
of subsurface flow. Once calibrated, the models can be run for different scenarios as well as for
transient simulations. Water budgets and fluxes can also be estimated. However, data

48

requirements for calibrating the models, especially across multiple scales, may be expensive and
time consuming.
In the data driven modeling approach adopted in this study, 2.1 was characterized based
on data for both

, and lithology based estimates of

, to estimate the velocity field .

Groundwater heads for shallow drift, deeper drift and bedrock aquifers were all obtained from
separate data sets. These data, as opposed to traditional data, are often noisy but abundant and
available from open sources. Moreover, the data are available virtually „everywhere‟ in the state
– giving the flexibility to zoom into any fen location for modeling. Interpolated heads were used
as „prescribed heads‟ in the models. Recharge estimates were not required. Models were not
calibrated as they were already based on data.

Recharge

Large Lakes
and Streams

Ditches,
Wetlands etc.

Small Lakes
and Streams
Shallow Drift Aquifer

Deeper Drift Aquifer

Bedrock Aquifer

Figure 2.17

Conceptual aquifer systems in Michigan

49

An advantage of data modeling approach over the traditional one is that it allows
modeling almost any fen in the state without having to visit the site for data collection – saving
time and money, as the approach employs publically accessible GIS data, available all over the
state – making possible multi-scale modeling of a large number of fen sites within a single study.
The disadvantage of data models as compared to the traditional models, however, is that
they cannot be used for transient simulations and estimating water budgets. And since the models
are not process based, they cannot be employed for scenario testing. The models only give the
existing or long-term average picture of hydrogeologic conditions prevailing around a fen site.
Such picture may be useful for identifying areas of critical hydrological processes and likely
source water areas to help guide planning for detailed hydrological studies at a particular site.

DATA USED
Michigan‟s Statewide GIS databases were used for all data requirements in this research.
These databases were compiled by Michigan Department of Environmental Quality (DEQ) under
Michigan State Public Act 148 which required the department to create a groundwater inventory
and map. The act also required the inventory and map to be accessible by public (GWIM 2006;
PA 148 2003). This study capitalized on these publically available databases to investigate
groundwater sources for prairies fens in Michigan.
Following statewide GIS data layers were used in this study to characterize groundwater
flow regimes at different scales of investigation:


Water Well Drillers Data – Wellogic (GWIM 2006), was used:
o to characterize deeper and regional groundwater flow dynamics;
o to estimate aquifer properties based on lithology information; and,
50

o to refine conceptual models.


National Hydrography Dataset NHD layer (NHD USGS 2010) was used to delineate
streams network in the models.



Inland lakes layer was used to delineate lakes in the models (GWIM 2006).



National Wetlands Inventory NWI data (GWIM 2006) was used to delineate wetlands in
the models.



Glacial Land Forms data (GWIM 2006) was used to refine conceptual models and was
correlated with the occurrence of fens and hydrologic regimes.



Digital Elevation Model DEM (NED USGS 2006) for topographic surface at 30m/10m
resolution was used:
o to estimate mean surface water elevation in lakes, streams and wetlands;
o to delineate aquifer top in the model areas; and,
o to estimate local surface water sheds around the fens.



A combination of DEM, NHD, Inland Lakes and Wetland layers were also used to
generate static water level data for shallow/local groundwater systems.



Statewide pre-processed raster (Oztan 2010) for bedrock elevation (500m resolution) was
used to delineate bedrock top in the models and to estimate thickness of glacial deposits
above the bedrock.



Bedrock geology data (GWIM 2006) was also used for bedrock aquifer properties.



Statewide pre-processed raster (Oztan 2010) for horizontal hydraulic conductivity, Kh,
(540m resolution) was used to delineate the Kh field for glacial and bedrock aquifers.

51

PRE-PROCESSING OF DATA
Although the data on static water levels (SWL) and aquifer properties was available in
high density from the Wellogic, it could not be directly used without preprocessing because of
inconsistencies and noise. Moreover, SWL data for shallow groundwater systems was not
available directly from any database. Different data layers (DEM, NHD and NWI etc.) had to be
manipulated to characterize shallow systems.
Noise, Outliers and Bad Data
Wellogic data was used in determining static water level SWL (for statewide and regional
scale data driven groundwater models). However, using SWL from Wellogic had its own
challenges because this data is very noisy, though available in high density throughout the study
area (Figure 2.18).
Geostatistical algorithms for data analysis were chosen and coded by Dr. Huasheng Liao
for the groundwater modeling software used in this study (Liao 2010, personal communication).
Using these codes, water wells data was preprocessed at the statewide scale. The preprocessed
data for SWL and hydraulic conductivity was used in this study for the selected model areas at
different scales. A very brief overview of data preprocessing is as follows.

52

Study Area

200 km

Figure 2.18

Water Well Drillers records in the Wellogic Database have
a good coverage throughout the state, especially in the
southern half of Lower Peninsula.

Depth to water level was recorded at the time a well was drilled and reported by the
driller. This information was used to determine SWL elevation by subtracting the depth to water
table from the 10m DEM. The evaluated SWL elevation for a selected band across the state is
shown in Figure 2.19. One can observe the density of coverage at this scale, but also obvious
from the plot is the variability and noise inherent in this data.

53

Data selected
in this band

100 km
400

SWL Elevation (m)

300
200
100
0
-100
-200
0

Figure 2.19

100

Distance (km)

200

300

A slice of wells selected across the state and the plot
of their static water level elevations.

54

Geostatistically interpolated
and smoothed groundwater
trend at statewide scale

Regional groundwater
mounds can be identified
from the statewide profile

400

SWL Elevation (m)

300
200
Bad data
Outliers

100
0
-100
-200
0

Figure 2.20

100

Distance (km)

300

300

Bad data and outliers were identified and removed. Rest
of

the

data

was

used

to

evaluate

SWL

trends

through

interpolation and smoothing.

Bad data and outliers are shown in Figure 2.20. These were identified and excluded in the
preprocessing stage. The process of identification and removal of outliers and bad data has been
automated in the tools developed for data processing (these tools include MIGWWP and IGW
Liao 2011, personal communications). The rest of the data was used in the models. However,
even after removal of bad data and outliers, the data was still very noisy. Some of major sources
of noise in Wellogic data can be explained by the following (the noise was handled by using
smoothing functions as explained on page 58):


The data covers a range from mid-sixties to date. Water levels recorded at the time of
well drilling. This introduces seasonal and temporal variability in well records.



No QA/QC guidelines were required when data were recorded by the drillers.
55



Well locations are based on address matching rather than actual well coordinates.



Lots of bad data.

Including Surface Water Features
Surface water features such as streams, lakes and wetlands have their influence on the
subsurface hydrology. To incorporate their influence on groundwater levels, additional SWL
data, in the form of scatter points, were generated for surface water bodies as narrated earlier.
Scatter points were generated inside the lakes and wetland polygons and along the
lengths of streams. Since these surface water features were assumed to be the intersections of
groundwater table with the topography, surface elevations from 10m DEM were assigned to
these points. An example of generated scatter points is shown Figure 2.1.
The density of points generated within lake or wetland polygons and along the length of
streams was based on the size of the feature. Bigger water bodies had higher density as they
would have bigger influence on groundwater as shown in Figure 2.21.
Different criteria were adopted to assign SWL values to the scatter points. First, the DEM
value from the 10m raster was measured at each scatter point generated in a polygon or along a
stream. For scatter points along the streams, the measured value was assigned as SWL at that
point.
For lakes and wetlands, different sampling schemes were available as shown in Average
of all measured values of DEM was chosen for wetland polygons as DEM elevations may vary
within the wetland polygons because all wetlands (and/or some parts within a wetland) may not
be wet and may have undulating relief. Lake elevations may not fluctuate as much, but to keep

56

the estimates closer to the mean lake levels, same scheme was used for lake polygons. Figure
2.21 shows the sampling settings used for all models in the study.

Figure 2.21

Surface water data sampling settings.

57

Interpolation and Smoothing
SWL data acquired from water-well logs and generated along the surface water features
was imported in the models as scatter points and interpolated for each grid cell in the model
using inverse distance method for regional scale models (which were based on data from water
wells) and kriging for the local scale models (which were based on data generated from surface
water features). Since interpolation was based on noisy data as explained earlier, smoothing was
applied to interpolated results for more intuitive estimation of SWL. Moving window smoothing
technique using four nearest neighbors was used as explained below:

2.4

2.5
Where

is the interpolated SWL representing the hydraulic head in a grid cell
and

are the hydraulic heads in the neighboring cells.

is the smoothed head after first pass of the moving window
is the weighting factor (taken as 0.5)
The general equation for several passes of moving window can be written as:
58

2.6
For the deep aquifer systems, between 10 and 20 passes of smoothing were applied, more
passes applied to larger systems. For the statewide model, covering the entire southern Michigan,
50 passes of smoothing were applied.
Figure 2.22 shows different levels of smoothing for a regional model. The contour
interval in all cases is 2m. Final surface for this particular site was based on 10 passes of
smoothing. Without smoothing the pattern looks noisy with many closed contours, while 20x
smoothed surface seems too smooth. The selection of final surface is based on the subjective
judgment which may differ from person to person. The judgment is based on examining the
contours such that they make sense when correlated to surface hydrology and topographic
features. For example, it is expected to have bends in contours along the large streams, ring
contours around lakes and topographic heights etc. Similarly, it is not expected to have a line of
ring contours along a river. Smoothing helped reduce noise and gain hydrologically more
consistent groundwater heads. For example, in Figure 2.22, either 5x or 10x smoothed surface
may be chosen. Choosing any of these surfaces will not significantly alter the quality of insights
gained from the models. The number of smoothing passes used for each model is mentioned so
that the model could be reproduced by others.
Generating Statewide Raster Maps for Static Water Level
Preprocessed data, along with additionally generated data was interpolated (using various
geostatistical techniques such as inverse distance and kriging) and smoothed to generate
statewide raster maps for static water level (SWL) at 540m resolution. These preprocessed and
pre-calculated raster maps were used to estimate SWL in the selected model areas.

59

5x smoothing passes after inverse
distance interpolation

No smoothing after inverse
distance interpolation

4km

20x smoothing passes after inverse
distance interpolation

10x smoothing passes after inverse
distance interpolation

Figure 2.22

Different

number

of

smoothing

passes

applied

to

a

regional model after inverse distance interpolation of
water wells data for drift aquifer. Contour interval is
2m. Blue color represents lower heads and orange higher.
For

this

particular

site,

10x

selected for particle tracking.
60

smoothed

surface

was

Chapter 3

Results

The data driven groundwater modeling approach was applied to delineate fen hydrology
at different spatial scales. We developed a steady state mean flow model for the entire southern
Michigan basin, 9 regional scale horizontal flow models around fen localities in varying
geological settings, and 19 site-specific profile models for selected fens. Results from each of
these models are presented in this chapter, with very brief discussions.
Statewide model has been discussed separately while the regional horizontal flow models
and site specific profile models are discussed together. Some commonalities in different site
models yielded insights into fen hydrology. Such commonalties have not been discussed in this
chapter. The discussions of common results and their implications are carried out in detail in
Chapter 4.

STATEWIDE MODEL
Lower half of Michigan‟s Lower Peninsula was modeled at the statewide scale which
included all 150 known sites of prairie fens in the state.
The model grid had 360 x 212 cells. The size of grid cells was approximately 1 km2
(1001.5m x 1001.5m). Source data for interpolation of SWL included water wells data with
outliers beyond 2 standard deviations removed and SWL data generated for big inland lakes and
streams (3rd and higher order) . The interpolation results were smoothed out by applying 50
passes of moving window as explained on page 58.

61

Statewide Hydrology and Occurrence of Fens
Figure 3.1 is a representation of groundwater surface, superimposed with the regional
network of 3rd and higher order streams. Qualitatively looking at the map:


Regional highs of groundwater are distinctly prominent around the headwaters of all
major streams in the area.



Most fen sites are distributed close to these regional groundwater highs.

North
KM
100
50
0
Prairie Fens
Prairie Fens with Mitchell’s Satyr (16)

Figure 3.1

Surface

and

occurrence

of

groundwater
fens.

hydrology

Relief

and

in

relation

contours

to

represent

groundwater surface elevation which varies from 148.14m
to 344.93m. Contour interval is 7m.

62

Flow Regime and Fen Occurrence Compared to Glacial Landforms
Except for 16 sites occurring in glacial till landforms and one occurring in coarse
lacustrine deposits, other 133 sites occur in proglacial or ice-contact outwash. The occurrence of
fen sites with respect to glacial landforms is illustrated in Figure 3.2. The relief surface
represents regional groundwater profile.

North
KM
100
50
0

Prairie Fens
Prairie Fens with MS

Figure 3.2

Outwash materials
Till materials

Occurrence of fens with respect to statewide groundwater
hydrology and the glacial landforms. Relief and contours
represent groundwater surface elevation which varies from
148.14m to 344.93m. Contour interval is 7m.

63

Statewide Regions of Significance for Source Groundwater
Significant sources of groundwater around the prairie fens across the state were identified
as nine regional groundwater mounds near the head waters of major rivers and stream in the
study area. The mounds are highlighted in Figure 3.3.
Allegan
Mound

Montcalm
Mound

Barry
Mound

Hillsdale
Mound

Jackson
Mound

Oakland
Mound

N

KM
100
50
0

Prairie Fens
Prairie Fens with MS
Groundwater Mounds

Van Buren
Mound

Figure 3.3

Source

Cass-Kalamazoo
Mound

groundwater

Calhoun
Mound

areas

of

regional

significance

for

prairie fen sites across the state. Relief and contours
represent groundwater surface elevation which varies from
148.14m to 344.93m. Contour interval is 7m.

64

For this study we named these mounds as 1) Montcalm Mound; 2) Allegan Mound; 3)
Barry Mound; 4) Calhoun Mound; 5) Cass-Kalamazoo Mound; 6) Van Buren Mound; 7)
Hillsdale Mound; 8) Jackson Mound; and 9) Oakland Mound.

SELECTION OF FEN SITES FOR DETAILED MODELING
Eighteen fen sites across the state were selected for detailed groundwater modeling. Each
site was first modeled at regional scale in 2D horizontal plane to delineate regional groundwater
flow paths leading to the fen site. A representative transect was then chosen along the regional
flow paths for each site and modeled as 2D vertical profile. Selected sites within the state are
highlighted in Figure 3.4.
The sites were selected based on their proximity to the regional groundwater mounds
identified in the statewide model (Figure 3.3), their occurrences in different glacial landforms,
and whether or not a site was occupied by MS. Some of the selected sites were either clustered or
very close to each other. Such sites were modeled together within a single regional model, but
their selected transects were still modeled separately for the vertical profiles.

65

Silver Lake Fen

Barry County
fens

N

Butterfield
Lake Fen
Springbrook
Fen
Paw Paw fens

Ives Road Fen
Skiff Lake Fen

Sarett Nature
Center Fen

Liberty Fen

Monette
Street Fen

Dew Road Fen

KM
100

Prairie Fens
Prairie Fens with MS

50
or

Fens selected for detailed modeling

0

Figure 3.4

Fens selected for detailed regional groundwater modeling.
Shaded relief and contours represent groundwater levels
delineated

at

statewide

scale

344.93m. Contour interval is 7m.

66

varying

from

148.14m

to

Skiff Lake/MacCready Fen

Dew Road Fen

Liberty Fen

Hillsdale Mound
Glacial Till
Glacial Outwash
Figure 3.5

0

4

8 KM

Selected fens in Jackson County. Contours at 7m interval
represent groundwater delineation from statewide model.

Fens selected in Jackson County occur in glacial outwash along the slopes of the
Hillsdale Groundwater Mound as shown in Figure 3.5. The sites include Liberty Fen, Skiff
Lake/MacCready Fen and Dew Road Fen. Liberty and Skiff Lake fens have MS occurrence.
Dew Road fen is one of the few fens occurring in glacial till landforms.

67

Ives road fen in Lenawee County was selected for detailed modeling because of its
unique location, much farther from the regional mounds, and occurring in a narrow band of
glacial outwash along Raisin River valley as shown in Figure 3.6.

Hillsdale Mound
Ives Road Fen

Glacial Till

0

4

8 KM

Glacial Outwash
Figure 3.6

Ives road fen located in a narrow band of glacial outwash
along

Raisin

River.

Contours

at

7m

interval

represent

groundwater delineation from statewide model.

A cluster of fens occur between Barry and Allegan groundwater mounds within glacial
outwash in Barry County as shown in Figure 3.7. Turner Creek Wetlands Fen, McKibben Fen,
Yankee Spring Fen, Hill Creek Fen, Hall Lake Fen, McDonald Lake Fen and Bowens Mill Fen
were selected from the cluster. First three fens in the group have MS occurrence.
68

Allegan Mound

Bowens Mill Fen
Turner Creek
Wetlands Fen

Hill Creek Fen
Yankee Spring Fen

McDonald Lake
Hall Lake Fen
Fen

McKibben Fen

Barry Mound

Sprinigbrook Fen

Butterfield
Lake Fen

Glacial Outwash

0

5

10 KM

and

Allegan

Glacial Till
Figure 3.7

Cluster

of

fens

between

Barry

Mounds.

Contours at 7m interval represent statewide groundwater.

Springbrook Fen and Butterfield Lake Fen in Kalamazoo County, also shown in Figure
3.7, were selected on the southern slope of Barry Groundwater Mound. Springbrook fen has MS
occurrence. Sarett Nature Center fen in Berrien County was selected for being the only prairie
fen occurring in coarse lacustrine deposits and relatively away from regional groundwater
69

mounds. Silver Lake fen in Kent County was selected for being one of the few fens occurring in
glacial till rather than coarse glacial deposits. Monette Street Fen in Cass County was selected
for its large size, high ecological ranking and its regional proximity to Cass-Kalamazoo Mound.
Paw Paw Lake Fen and Paw Paw Prairie Fen were selected in Kalamazoo/Allegan counties for
being in close proximity of each other, one with MS occurrence and one without. These two fens
are located very close to the apex of Cass-Kalamazoo Mound. Groundwater models for each of
these fen sites are presented as follows.

REGIONAL AND PROFILE MODELS FOR SELECTED SITES
Barry State Game Area Group of Fens
Located relatively closer to each other, there exists a cluster of fens between the Barry
and Allegan Groundwater Mounds. Following fens from this cluster were selected for regional
modeling. Ecological ranking or EORANK (NatureServe 2002) of the fens are given in
parentheses:


Hill Creek Fen (B)



Turner Creek Wetlands Fen (BC)



Bowens Mill Fen (C)



McDonalds Lake Fen (C)



Yankee Spring Fen (BC)



Hall Lake Fen (C)



McKibben Fen (C)

70

The cluster of above fens occurs in glacial outwash deposits overlaying Marshall
Sandstone. The average outwash thickness around the cluster is 60m, varying between 25m and
150m. The fens and their surface catchment areas are shown on 30m/10m DEM relief in Figure
3.8. MS occurrence has been reported at Turner Creek Fen, McKibben Fen and Yankee Spring
Fen.
Barry State Game Area site has shallow groundwater, deep glacial drift, and distinct
transmissive bedrock aquifer systems. Each of these systems was separately evaluated in the
regional model and then correlated in profile models for the individual fen sites.

71

Turner Creek
Wetlands Fen

Hill Creek Fen

McDonald
Lake Fen
Yankee Springs
Fen

Bowens Mill
Fen

N

McKibben
Fen
Wetlands

Hall Lake Fen
2 KM

Lakes

Outwash
0

Bedrock (Marshall)
Figure 3.8

Cluster of fens in Barry State Game Area. The shaded
relief represents topography derived from 10m DEM which
varies from 223.22m to 342.78m.

Regional Groundwater Model for Fens Cluster
Regional model was selected over an area of 15.77 km by 16.12 km which included all
selected fens. Model area was divided into the 300x313 grid cells, with the cell size of 51.59m.
The cluster of fens occurs in significantly thick glacial outwash aquifer overlying a transmissive
bedrock formation – Marshall Sandstone. Water wells data in the model area was available for
72

both the glacial drift as well as bedrock. SWL was separately delineated for drift and sandstone
formations, using respective wells data.
Figure 3.9 shows SWL delineation for drift aquifer, derived from water wells data of
drift wells and taking large surface water bodies into account. Groundwater capture zones were
estimated using reverse particle tracking using groundwater head surface in the drift aquifer. The
estimation suggests the fen sites may be connected to Allegan and Barry Groundwater mounds,
between 5 and 12 km from the fen sites.
Fens

Groundwater
capture zones
for fens

Allegan
Groundwater
Mound

Barry
Groundwater
Mound

4 KM

Bedrock
N

Figure 3.9

0

Drift wells

The contours at 2m interval and shaded relief represent
regional groundwater head around Barry State Game Area in
the drift aquifer, varying between 220 m and 279 m.

73

Figure 3.10 shows SWL delineation for bedrock aquifer. This delineation is based only
on the bedrock wells data only, with consideration of surface water features. Bedrock flow
system seems to converge to major surface water features along northeast-southwest direction
from eastern and western up hills. This pattern gives a strong indication of bedrock upwelling in
this area. Likely zone of upwelling from bedrock aquifer is highlighted in the figure. In Figure
3.11, both drift and bedrock SWL surfaces are presented for comparison.

Bedrock wells

4 KM

N
0

Likely zone of
bedrock
upwelling

Figure 3.10

The contours at 2m interval and shaded relief represent
regional groundwater head around Barry State Game Area in
the drift aquifer, varying between 221 m and 259 m.

74

SWL in
bedrock aquifer

4km

SWL in
drift aquifer

Figure 3.11

Comparison of Bedrock and drift SWL surfaces in Barry
State Game Area. Contour labels represent elevation (m).
75

For most parts of the model area SWL in drift seems higher than that of bedrock;
however, bedrock SWL seems systematically higher in the groundwater discharge zones across
the model. This pattern suggests that drift aquifer is recharging the bedrock in the higher
elevations and the bedrock is discharging groundwater in the lower elevations.

This also

suggests that the two aquifers are connected.
Figure 3.12 shows bedrock well-logs across a few cross sections through the model area.
There seem to be clay layers in some parts of the drift and missing in others. Corresponding to
the discharge zones the clay seems particularly missing in the well lithologies. This reinforces
the possibility that the two aquifers are connected and, despite occurrence of clay within the
drift, there seems no continuous layer separating the bedrock and drift aquifers.

76

Section AA

Model area

Section BB

Section CC

0

2

4km

Section AA

Clay in
well logs

Section BB

Section CC

Figure 3.12

Distribution

of

clay

in

sections in model area

77

well

logs

across

selected

In Figure 3.13, the groundwater contribution zones are superimposed on surface
topography and geological land forms. The groundwater capture zones seem to converge to the
regional groundwater mounds, crossing over one or two regional watersheds on their way.

Regional
Watersheds

Groundwater
capture zones

N

Glacial drift
4 KM

0

Bedrock
(Marshall
Sandstone)

surface

topography

Proglacial outwash
Ice-contact outwash
Figure 3.13

The

shaded

relief

represent

in

the

region which varies in elevation from 241.06m to 298.58m.

Groundwater Profile Model – Hill Creek Fen
Well logs along the capture zone of Hill Creek Fen are shown in Figure 3.14. The logs do
not reveal existence of continuous clay layers in any portion along the capture zone. A two-layer
profile model was conceptualized with glacial drift layer overlying Marshall Sandstone, Barry
Mound being the regional source at one end, and Hill Creek being the regional sink at the other.
The profile model is shown in Figure 3.15. The model suggests that delivery of groundwater to
the fen is through upwelling of deep groundwater flow system which has its source
78

approximately 6 km south of the fen site just across the surface regional watershed. The model
also shows that there is upwelling of shallower groundwater closer to the fen. This upwelling
water is feeding the wetlands and stream just south of the fen. Deep groundwater from Barry
Groundwater Mound is upwelling into the stream north of the fen and does not seem to directly
contribute to the fen.

Hill Creek
Hill Creek Fen

Wetlands

Selected wells

Glass Creek

Estimated regional
groundwater flow path

4km
Well Logs from Selected wells

Clay in well lithologies

Figure 3.14

Well location and well logs along Hill Creek Fen capture
zone. Contours at 2m interval show SWL for deep drift.

79

Hill
Creek

Projection of fen
in glacial drift

Glacial drift

Wetland

Bedrock

Glass
Creek

Elevation (m)

340

Barry
Mound

100

Likely Groundwater
contribution area

Elevation (m)

340

100
Figure 3.15

13,465 m
Profile Model for Hill Creek Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 1.86m

interval.

Reverse particle tracking to estimate groundwater flow regime (pink) and

likely fen’s source water (purple) [bottom].

80

Groundwater Profile Model – Turner Creek Wetlands Fen
Turner Creek Wetlands Fen comprises a major area attached to Turner Creek and three
smaller parts in the south. Well logs along the capture zone of the fen are shown in Figure 3.16.
The logs do not reveal any structural discontinuities in the drift. A two-layer profile model was
conceptualized with glacial drift layer overlying Marshall Sandstone, Barry Mound being the
regional source at one end and lower reach of Turner Creek being the regional sink at the other
end. The model, in Figure 3.17, shows groundwater upwelling into the smaller parts of the fen
from glacial drift, and in the larger part from the bedrock. The source water recharge comes from
a sub-regional groundwater mound between Glass Creek and Turner Creek, approximately 6 km
from the fen site. It appears that shallow systems are not contributing much to the fen.
Fen

Selected wells
Turner Creek
Glass Creek
Estimated regional
groundwater flow path

4km

Well Logs from Selected wells

Clay in well lithologies

Figure 3.16

Wells

and

well

logs

along

Turner

Creek

Wetlands

Fen

capture zone. Contours at 2m interval show SWL for drift.

81

Figure 3.17

Profile Model for Turner Creek Wetlands Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours

at 1.90m interval. Reverse particle tracking to estimate groundwater flow regime
(pink) and likely fen’s source water (purple) [bottom].

82

Groundwater Profile Model – Bowens Mill Fen
Well logs along the capture zone of Bowens Mill Fen are shown in Figure 3.18 which
reveals a pocket of clay within the glacial drift. Two separate profile models for the site were
made – one with clay pocket and one without. In both models, the profiles were conceptualized
with glacial drift layer overlying Marshall Sandstone, Allegan Mound being the regional source
at one end and Turner Creek with adjacent wetlands being regional sink at the other end. With
slight variations, Figure 3.19, both models predict that the fen may be receiving groundwater
from both the shallow and the deep systems. The shallow system may extend up to 600 m
westwards from the fen site. The deeper system may be receiving its recharge from the Allegan
Mound, approximately 6 km from the fen site, upwelling into the fen from the bedrock aquifer.
Barlow Lake

Selected wells

Fen
Estimated regional
groundwater flow path

Turner Creek
4km

Well Logs from Selected wells
Clay in well lithologies

Figure 3.18

Well

locations

and

well

logs

along

Bowens

Mill

Fen

capture zone. Contours at 2m interval show SWL for drift.

83

Figure 3.19

Profile Model for Bowens Mill Fen with clay lenses in the glacial drift [bottom]
and without [top]. Light blue lines show groundwater head contours at 0.77m
interval. Reverse particle tracking shows estimated groundwater flow regime (pink)
and likely source water paths for the fen (purple).

84

Groundwater Profile Model – McDonalds Lake Fen
Well logs along the capture zone of McDonalds Lake Fen are shown in Figure 3.20. The
logs do not reveal structural discontinuities in the drift. A two-layer profile model was
conceptualized with glacial drift layer overlying Marshall Sandstone, Barry Mound being the
regional source at one end and McDonald Lake with adjacent wetlands being the regional sink at
the other end. The model, shown in Figure 3.21, predicts that the fen may be receiving
groundwater from the deeper groundwater system connected to a sub-regional groundwater
mound between Deep Lake and Glass Creek, approximately 5 km from the fen location. The
model also suggests that upwelling water in the fen may not come in contact with the bedrock.
McDonald
Lake

Deep Lake
Glass Creek

Fen

Estimated
regional
groundwater
flow path

Selected
wells

Model
extent

4km
Well Logs from Selected wells

Clay in well lithologies

Figure 3.20

Well locations and well logs along McDonalds Lake Fen
capture zone. Contours at 2m interval show SWL for drift.

85

Figure 3.21

Profile

Model

groundwater

head

for
and

McDonald
velocity

Lake

Fen.

[top].

Geological

Light

blue

layers

lines

show

and

delineation

groundwater

of

head

contours at 1.83m interval. Reverse particle tracking to estimate groundwater
flow regime (pink) and likely fen’s source water (purple) [bottom].

86

Groundwater Profile Model – Yankee Spring Fen
Well logs along the capture zone of Yankee Spring Fen are shown in Figure 3.22. The
logs do not reveal any structural discontinuities in the drift closer to the fen site. A two-layer
profile model was conceptualized with glacial drift layer overlying Marshall Sandstone, Barry
Mound being the regional source at one end and Gun Lake being regional sink at the other end.
The model, in Figure 3.23, suggests that the fen receives water from the adjacent shallow system
which extends all the way to the apex of the sub-regional groundwater mound approximately 1.8
km from the fen site. The model also suggests that upwelling water in the fen does not come in
contact with the bedrock. The fen does not seem to receive water from the deeper systems.
Model Extent

Glass
Creek

Fen

Gun
Lake

Estimated
regional
groundwater
flow path

4km

Well Logs from Selected wells

Clay in well lithologies

Figure 3.22

Well

locations

and

well

logs

along

Yankee

Spring

Fen

capture zone. Contours at 2m interval show SWL for drift.

87

Figure 3.23

Profile Model for Yankee Spring Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 1.92m

interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen’s source water (purple) [bottom].

88

Groundwater Profile Model – Hall Lake Fen and McKibben Fen
Hall Lake Fen and McKibben Fen share the same regional capture zone as shown in
Figure 3.24. Well logs along the capture zone are also shown which do not reveal any structural
discontinuities in the drift closer to the fen site. A two-layer profile model was conceptualized
with glacial drift layer overlying Marshall Sandstone, Barry Mound being the regional source at
one end and Gun Lake being regional sink at the other end. The model, shown in Figure 3.25,
predicts that the McKibben Fen may be receiving groundwater from the adjacent shallow system
while Hall Lake Fen receives water both from the adjacent shallow system as well as deeper
upwelling. The groundwater feeding the fens does not seem to come in contact with bedrock.
Hall Lake

Model extent

Hall Lake Fen
McKibben Fen
Gun Lake
Selected wells
Estimated regional
groundwater flow path
2km
Well Logs from Selected wells

Clay in well lithologies

Figure 3.24

Well locations and well logs along Hall Lake Fen and
McKibben Fen capture zone.

89

Figure 3.25

Profile

Model

for

Hall

Lake

Fen

and

McKibben

Fen.

Geological

layers

and

delineation of groundwater head and velocity [top]. Light blue lines show groundwater

head contours at 2.30m interval. Reverse particle tracking to estimate groundwater
flow regime (pink) and likely fen’s source water (purple) [bottom].

90

Liberty Fen and Skiff Lake/MacCready Fen
Liberty Fen occurs in proglacial outwash at the headwaters of Grand River where the
river is only a first order stream. Bedrock formation under the glacial drift is Marshall Sandstone
– a very transmissive formation. Average thickness of glacial drift in the area is 21m, with
maximum 38m and minimum 12m. The fen and its surface catchment area are shown on 10m
DEM relief in Figure 3.26. EORANK of the fen is „A‟. MS occurrence has been reported at
Liberty Fen.
Fen’s surface
catchment
Liberty Fen

Lakes

2 KM

Wetlands

0

Grand River

Figure 3.26

Liberty Fen. The shaded relief represent surface relief
derived

from

10m/30m

DEM.

between 310.69m and 384.31m.

91

Surface

elevation

varies

Skiff Lake Fen occurs on the southwestern side and MacCready Fen occurs on the
northeastern side of Skiff Lake. Both fens occur in proglacial outwash. Bedrock formation under
the glacial drift is Marshall Sandstone – a very transmissive formation. Average thickness of
glacial drift in the area is 24m, with maximum 58m and minimum 6m. The fen and its surface
catchment area are shown on 10m DEM relief in Figure 3.27. A prominent esker runs along the
northwestern extent of MacCready Fen. EORANK of the fen is „BC‟. MS occurrence has been
reported at Skiff Lake Fen but not at MacCready Fen.

Fen’s surface
catchment

Skiff Lake Fen
Skiff Lake
Grand River

Esker
2 KM

Wetlands
MacCready Fen

Figure 3.27

Skiff

Lake

represent

0

and

MacCready

surface

relief

Fens.
derived

The
from

shaded

relief

10m/30m

Surface elevation varies between 294.77m and 346.75m.

92

DEM.

Regional Groundwater Model
Liberty Fen and Skiff Lake/MacCready Fen were both covered in the same regional
model. Model area of 16.12 km by 19.97 km was divided into the 300x375 grid cells, with the
cell size of 53.40 m. The model area consists of thin glacial outwash deposits overlying a
shallow bedrock formation – Marshall Sandstone, which is very transmissive. Marshall
Sandstone is underlain by impervious Coldwater Shale formation. Static water level (SWL) was
ascertained using bedrock wells and surface water data from large streams/lakes. Delineated
zones of groundwater contribution to the fens are shown in Figure 3.28.
Shallow
bedrock

Hillsdale Mound
Liberty Fen
capture
zone

Skiff Lake /
MacCready Fen
capture zone

6 KM

Liberty Fen
0

Skiff Lake Fen

Zone of regional
upwelling enlarged
in Figure 3.29

MacCready Fen
Figure 3.28

The contours at 2m interval and shaded relief represent
regional scale delineation of groundwater surface around
Liberty Fen and Skiff Lake Fen which varies in elevation
from 292.92m to 347.14m.
93

The exception was made to include surface water data while delineating SWL from
bedrock wells because of the shallow proximity of bedrock to the surface.
Reverse particle tracking was used to delineate groundwater capture zones for the
selected fen sites in the model. Flow paths leading to all fen sites in the area seemed to originate
from Hillsdale Groundwater Mound in the south-west of the model area, between 12 and 18 km
from the respective fen sites. Residence time for the regional flow paths were estimated between
70 and 100 years.
The regional model shows bending of contours in the north-eastern part of the model
which implies convergence of regional flow in this area. Since Marshall Sandstone is underlain
by less transmissive shale, the convergence of flow suggests possibility of upwelling in this area.
Regional Groundwater Upwelling
The likely upwelling area as highlighted in Figure 3.28 was modeled in more detail to
ascertain regional upwelling. SWL was separately delineated for this area using bedrock wells
only (i.e., without taking into consideration any surface water features). SWL delineated by
bedrock wells, as shown in Figure 3.29, also showed strong convergence of regional flow in this
zone. The flow paths on either sides of the fen complex were oriented almost perpendicular to
each other.
During a site visit to this area, three artesian wells were observed around Skiff Lake /
MacCready fen sites. Exact locations of these wells are marked in Figure 3.29 The presence of
artesian wells within the convergence zone suggested by the model provides qualitative
validation of regional groundwater upwelling in the area – suggesting that fens may also be
benefitting from this source.

94

Please also refer to Chapter 4, Figure 4.1 and Figure 4.2 for discussion on regional flow
convergence observed around other fen sites in this study.

Converging
regional flow
pattern

Converging
regional flow
pattern

3 KM
N

200 gpm Artesian Well
0

Bedrock

Figure 3.29

Artesian Wells

Zone of regional upwelling around Skiff Lake/MacCready
Fens

and

Contours

locations
at

groundwater

1.05m
head

of

artesian

interval

surface

bedrock wells data.

95

wells

and

delineated

in

relief
by

the

area.

represent

interpolating

In Figure 3.30, the groundwater contribution zones are superimposed on surface
topography and geological land forms. The two fen sites occur within the glacial outwash. The
capture zones extend beyond the regional surface water catchments.

Liberty Fen
capture zone

Skiff Lake/MacCready
Fens capture zone

Liberty Fen

6 KM

Regional
Watershed

Skiff Lake Fen
Glacial Till
0

MacCready Fen

Figure 3.30

Glacial Outwash

The shaded relief represents surface topography in the
region which varies in elevation from 296.27m to 387.36m.

Local Flow Systems
Estimation of contribution areas of local/shallow flow systems were also made for the
two fen sites as shown in Figure 3.31 and Figure 3.32. The models for local flow systems suggest
that local flow may be contributing to the fens from all around. The residence times for local
flow systems at the two sites was estimated by using range of effective porosity from 0.1 to 0.3

96

in Equation 2.1. The estimated times varied between 1 and 3 years for the local systems. In case
of Liberty Fen, local high groundwater areas to the south of the fen seem to be connected to the
fen. The local system to the north of the fen has relatively small gradients. However, it can be
readily seen that these local highs across the flatter gradients may be contributing to the flatter
areas right next to the fen site, and therefore, may be indirectly providing water to the fen.
The profile model explained in the next section only captures the local flow details along
the selected section. Detailed estimates for local flow systems for other modeled sites in this
study area are not presented.

Liberty
Fen

Liberty Fen
local capture
zone

Liberty Fen
regional
capture zone

Figure 3.31

Local flow system around Liberty fen.
97

MacCready
Fen

Skiff Lake
Fen
Local capture
zone

Regional
capture zone

Figure 3.32

Local flow systems around Skiff Lake/MacCready fens.

Groundwater Profile Model – Liberty Fen
The profile section selected across the fen and its regional capture zone is shown in
Figure 3.33. The section starts in the south at the top of Hillsdale Groundwater Mound, runs
through the delineated groundwater capture zone of Liberty Fen, and extends further north-east
of the fen, ending close to the Grand River - the regional sink in the area.

98

Hillsdale Mound
Section for
Skiff/MacCready
Profile Model

6 KM

Section for Liberty
Profile Model

Figure 3.33

Selected

0

sections

for

profile

models.

Shaded

relief

represents topography and contours represent regional SWL
at 2m interval.

Profile model for Liberty Fen is shown in Figure 3.34. The model shows that the regional
groundwater in the capture zone delineated in the horizontal model flows mostly in the bedrock –
Marshall Formation, underneath the glacial drift. Some upwelling from bedrock into the fen
seems likely and its source may be linked to a recharge area around a sub-regional surface
watershed. The shallow system and drift aquifer, extending approximately 4.8 km from the fen
seems to be contributing most groundwater system to the fen site.

99

Figure 3.34

Profile Model for Liberty Fen. Geological layers along the profile and delineation
of

groundwater

head

and

velocity

[top].

Light blue lines show groundwater head

contours at 1.85m interval. Reverse particle tracking to estimate groundwater flow
regime (pink) and likely fen’s source water areas (purple)[bottom].

100

Groundwater Profile Model – Skiff Lake/MacCready Fen
The profile section selected across the fen and its regional capture zone is shown in
Figure 3.33. The section starts in the south at the top of Hillsdale Groundwater Mound, runs
through the delineated groundwater capture zone of Skiff Lake/MacCready Fens, through the fen
sites and ending close to the Grand River - the regional sink in the area.
Profile model for Skiff Lake/MacCready Fens is shown in Figure 3.35. The model
suggests that Skiff Lake Fen may be receiving groundwater from both shallow and deep aquifer
systems. The shallow system in drift aquifer feeding the fen may extend from the fen site to
approximately 3.5km to the west. Groundwater feeding the fen from the deeper Marshall
formation may have its recharge area approximately 11 km from the fen sites in the south-west
direction. Recharge from this zone is also feeding MacCready Fen.
It appears from the profile model in Figure 3.35 that MacCready Fen is getting all its
water only from the deeper bedrock aquifer only. However, it is very important to keep in mind
the local horizontal flow model shown in Figure 3.32 which suggests that the direction of local
flow to MacCready fen is almost perpendicular to the regional flow direction along which the
profile model was built. The profile model, therefore, does not show the local flow component to
the fen. It appears from the local horizontal flow model that wetland and adjacent hills to the
north of the fen may be the sources of shallow groundwater contribution to the fen. This case
typically highlights the 3D nature of flow systems in the area.
The profile model shows regional connection and the possibility that part of groundwater
discharge at MacCready fen may even be linked to the recharge areas at Hillsdale Mound, 22 km
from the fen site. The model also shows that most of deeper groundwater flow from Hillsdale
Mound bypasses the fens and ends up in Grand River.
101

Figure 3.35

Profile Model for Skiff Lake/Macready Fens. Geological layers along the profile
and

delineation

of

groundwater

head

and

velocity

[top].

Light

blue

lines

show

groundwater head contours at 1.88m interval. Reverse particle tracking to estimate
groundwater flow regime (pink) and likely fen’s source water areas for Skiff Lake
(purple) and MacCready (brown) fens [bottom].

102

Ives Road Fen
Ives Road Fen is one of the few prairie fens occurring far away from regional
groundwater mounds and in relative isolation compared to nearby fens. It occurs in proglacial
outwash deposits along Raisin River. Bedrock formation under the outwash is Coldwater Shale.
Average thickness of outwash is 56m, with maximum 79m and minimum 37m around the fen.
The bedrock almost pinches to the surface about 800m to the west of the fen. The fen and its
surface catchment area are shown on 10m DEM relief in Figure 3.36. EORANK of the fen is
„A‟. MS occurrence has not been reported at this site.
Lakes

Ives Road Fen

Raisin River

Fen’s surface
catchment

1.5 KM

Coldwater Shale
0

Wetlands

Figure 3.36

Ives Road Fen along the Raisin River valley. The shaded
relief

represents

topography

derived

which varies from 218.98m to 253.64m.

103

from

10m/30m

DEM

Regional Groundwater Model
Regional model for Ives Road Fen was selected over an area of 23.26 km by 15.66 km.
The area was divided into the 450x304 grid cells, with the cell size of 51.69 m. Static water level
(SWL) and other modeling parameters were ascertained as explained in methodology. Reverse
particle tracking was used to delineate groundwater capture zone for the fen site in the model.
The delineated zone is shown in Figure 3.37.
Northeastern end of
Hillsdale Mound

Groundwater
capture zone
Raisin River

6 KM

Coldwater Shale

0

Ives Road Fen
Figure 3.37

Water well
data points

The contours at 3m interval and shaded relief represent
regional scale delineation of groundwater surface around
Ives Road Fen which varies in elevation from 217.40m to
295.15m.

104

The model shows that the contribution zone may be connected to the north eastern end of
Hillsdale Groundwater Mound, approximately 16 km to the northwest from the fen site. In
Figure 3.38, the groundwater contribution zone is superimposed on surface topography and
geological land forms. The fen occurs in a narrow strip of glacial outwash along the Raisin River
valley. The groundwater contribution area, approximately perpendicular to the river valley
extends across the till plains all the way to another zone of glacial outwash as shown in the
northwest corner of the model. This area coincides with the northeastern end of the Hillsdale
Groundwater Mound. A few first order streams crisscross the capture zone in the till plains. A
few wetlands and lakes also occur over the capture zone.
Groundwater
capture zone
Raisin River

Coldwater Shale
(bedrock)
6 KM

Outwash
Till

Figure 3.38

0

Ives Road Fen

The shaded relief represents surface topography, varying
in elevation from 217.42m to 332.32m. Glacial landforms
in the area are represented by different colors.

105

Groundwater Profile Model
Analysis of lithology from the well logs around Ives Road Site revealed a thick and
continuous layer of clay close to the ground surface, extending west of the fen location for
approximately 15 km. Geostatistical interpolation of thickness of the clay layer and its depth
from the surface is shown in Figure 3.39. The clay layer is close to the surface in the central
portion of the fen‟s groundwater capture zone while it is buried deeper in outwash deposits at its
either ends. Closer to Raisin River valley the top of the layer seems to have been eroded by the
river and replaced by outwash deposits. The layer becomes thinner, and even seems missing at
places, especially in the vicinity of the fen.

Thick deposits of glacial drift to the west of Raisin River overlay Coldwater Shale
formation which may be considered as impervious for modeling purposes. The presence of a
large number of water wells in the area indicates a productive aquifer underneath the clay layer
within more transmissive outwash deposits from earlier glaciations. Gaps in the clay layer closer
to Raisin River connect the older outwash with the recent outwash deposits in the river valley.

A representative vertical section for the capture zone was inferred from this data analysis
as illustrated in Figure 3.40. The conceptual profile model for Ives Road Fen site was based on
this section. North eastern extent of Hillsdale Groundwater Mound (coinciding with the north
western end of the capture zone) was taken as regional recharge area while Raisin River was
taken as the regional sink. Confined aquifer conditions were assumed under the continuous clay
layer. Gaps in clay layers were added within the outwash deposits along Raisin River.

106

A

Clay Thickness

0

2km

0

2km

Clay Depth

Clay thickness reported in wells logs is color coded as follows:
0-3m

Figure 3.39

3-6m

Thickness

[top]

6-15m

and

depth

>15m

[bottom]

interpolated from water wells data.
107

of

clay

deposits

Regional Recharge area

Raisin River
Valley
Fen in outwash
deposits
Coldwater Shale
Clay layer
Aquifer
Gaps in Clay
Thin clay layer with gaps
Regional
recharge area
Clay

Clay eroded and
replaced by outwash
along the river valley

Raisin River
(Regional Sink)
Outwash

Glacial Drift

Impervious Coldwater Shale

Figure 3.40

Perspective
conceptual

Gaps in Clay

visualization
profile

based

[bottom]

of

on
Ives

data
Road

[top]

and

regional

groundwater capture zone.

Groundwater profile model for Ives Road Fen is shown in Figure 3.41. The model
predicts groundwater flow from the northeastern end of Hillsdale Groundwater Mound, 15.5 km
from the fen, seeping beneath the clay layer and upwelling into the glacial outwash in Raisin
River valley. It seems possible that this regional flow is also upwelling into the fen. The model
also shows that the fen receives water from the outwash deposits, with capture zone extending up
to 3km west from the fen.

108

Figure 3.41

Profile Model for Ives Road Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 2.84m

interval.

Reverse particle tracking to estimate groundwater flow regime (pink) and

likely fen’s source water (purple) [bottom].

109

Sarett Nature Center Fen
Sarett Nature Center Fen is the only known prairie fen occurring in the coarse lacustrine
deposits overlying shale bedrock. Average thickness of glacial deposits around the fen is 54m,
with maximum 65m and minimum 45m. The fen occurs along Paw Paw River valley besides a
large complex of wetlands. The fen and its surface catchment area are shown on 10m DEM relief
in Figure 3.42. EORANK of the fen is „C‟. MS occurrence has been reported at this site.

Sarett Nature Center Fen
Fen’s surface catchment

Wetlands

300 M

Paw Paw
River

0

Bedrock (shale)

Figure 3.42

Sarett Nature Center Fen along the Paw Paw River valley.
The shaded relief represents topography derived from 10m
DEM which varies from 176.75m to 198.44m.

110

Regional Groundwater Model
Regional model for Sarett Nature Center Fen was selected over an area of 12.78 km by
10.91 km. The area was divided into the 300x261 grid cells, with the cell size of 41.14 m. Static
water level (SWL) and other modeling parameters were ascertained as explained in
methodology. Reverse particle tracking to delineate groundwater capture zone for the fen site
suggests the fen sites may be connected to a sub-regional groundwater mound 8km away. The
delineated zone is shown in Figure 3.43. Bending of regional contours closer to the fen location
suggests regional flow convergence and possibility of regional upwelling in the area. Please also
refer to Figure 4.2 for more discussion.
Sub-regional
groundwater Mound

Groundwater
capture zone
Sarett Nature
Center Fen

4 KM

Water well
data points
0

Figure 3.43

The contours at 2m interval and shaded relief represent
regional

groundwater

head/surface

around

Sarett

Center Fen which varies from 176.30m to 218.42m.

111

Nature

The model shows that the contribution zone may be connected to a sub-regional
groundwater mound approximately 8 km to the east. In Figure 3.44, the groundwater
contribution zones are superimposed on surface topography and geological land forms. The fen
occurs in the lacustrine coarse deposits while the capture zone extends through the glacial till and
outwash materials all the way to a regional watershed boundary. Many wetlands and a lake occur
within the capture zone.

Regional
Watershed

Groundwater
capture zone
Sarett Nature
Center Fen

N

4 KM

Outwash
Till
0

Lacustrine coarse

Figure 3.44

The

shaded

relief

represent

surface

topography

in

the

region which varies in elevation from 176.31m to 256.95m.

112

Groundwater Profile Model
Analysis of lithology from the well logs around Sarett Nature Center Fen in Figure 3.45
revealed a continuous layer of clay, coinciding with approximately 2.7 km of the glacial till to
the east of the fen. The estimated extent and thickness of this clay layer was incorporated in the
profile model as shown in Figure 3.46. The model predicts groundwater flow from the subregional groundwater mound beneath the clay layer and upwelling into Paw Paw River. It seems
possible that part of this regional flow is also upwelling into the fen. The model suggests that the
fen receives water from the sub-regional groundwater mound approximately 7.5 km east, through
deep upwelling from beneath the clay layer and also from shallow groundwater in the adjacent
coarse lacustrine deposits extending700m to the east from the fen.
Paw Paw River
Sarette Nature
Center Fen

Wetlands
Selected wells
Estimated regional
groundwater flow path
Well Logs from Selected wells

Extent of
profile
model

4km

Continuous clay

Clay in well lithologies
Figure 3.45

Well location and well logs along Sarett Nature Center
Fen capture zone. Contours at 2m interval show SWL for
drift.
113

Figure 3.46

Profile Model for Sarett Nature Center Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light blue lines show groundwater head contours

at 1.84m interval. Reverse particle tracking to estimate groundwater flow regime
(pink) and likely fen’s source water (purple) [bottom].
114

Paw Paw Lake Fen and Paw Paw Prairie Fen
Paw Paw Lake Fen and Paw Paw Prairie Fen occur very close to the northern apex CassKalamazoo Groundwater Mound. Located along headwaters of Paw Paw River, the fens occur in
glacial outwash where drift thickness is 100m or more. Paw Paw Prairie Fen occurs in proglacial
outwash while Paw Paw Lake fen occurs in ice contact outwash. The fens and their surface
catchment areas are shown on 10m DEM relief in Figure 3.47. EORANK of Paw Paw Lake Fen
is „B‟ and MS occurrence has been reported at this site. EORANK of Paw Paw Prairie Fen is
„BC‟ and MS occurrence has not been reported at this site. However, this site is a candidate for
re-introduction of MS.

Surface catchments for fens
Paw Paw
Lake Fen

Paw Paw
Prairie Fen

Paw Paw Lake
1 KM
0

Figure 3.47

Wetlands
Paw Paw River

Paw Paw Lake Fen and Paw Paw Prairie Fen along Paw Paw
River. The shaded relief represents topography derived
from 10m DEM which varies from 261.74m to 316.91m.

115

Regional Groundwater Model
Paw Paw Lake Fen and Paw Paw Prairie Fen were both covered in the same regional
model. Model area of 10.34 km by 8.90 km was divided into 300x285 grid cells, with the cell
size of 34.49 m. SWL for the model was ascertained from water wells records in the drift aquifer
and taking into consideration major surface water features in the area. The model area is situated
in thick glacial deposits overlying less transmissive shale. Strong bending of regional
groundwater in drift deposits suggests possibility of regional upwelling in the area.
Groundwater capture zones
Paw Paw Lake Fen

4 KM

Groundwater
capture zones
Paw Paw Prairie Fen

Water well
data points
0

Figure 3.48

The contours at 1m interval and shaded relief represent
regional scale delineation of groundwater surface around
Paw Paw Fens which varies in elevation from 243.60m to
273.48m.
116

The bending pattern of contours was reproduced along the same stretch even when no
surface water data was taken into consideration as shown in Figure 4.1. Although the upwelling
seems to be directed to surface water bodies in the area , there seems a strong possibility that the
fens, located within the likely upwelling zones, are also benefitting from the regional upwelling
which can make a more steady and robust source of groundwater for fens.
In Figure 3.49, the groundwater contribution zones are superimposed on surface
topography and geological land forms.
Groundwater
capture zones

Paw Paw
Lake Fen

Paw Paw
Prairie Fen
Paw Paw
River

Regional
Watershed
4 KM

Proglacial outwash
0

Ice-contact outwash
Figure 3.49

The

shaded

relief

represent

surface

topography

in

the

region which varies in elevation from 243.64m to 321.76m.

117

Likely zones of groundwater contribution on the northern side of Paw Paw River extend
beyond the river watershed. The fen locations seem to be connected to groundwater sources on
either sides of the river. The model shows that the contribution zones extend beyond the surface
water catchment areas of the fens and may be connected to sub-regional groundwater mounds
between 3 and 4 km from the respective fen locations.
Groundwater Profile Model – Paw Paw Lake Fen
Analysis of lithology from the well logs around Paw Paw Lake Fen within its delineated
capture zone did not reveal continuous clay lenses within the glacial drift. The profile section
was selected as shown in Figure 3.50. Cass-Kalamazoo Groundwater Mound at the eastern end
of profile was taken as regional source and Paw Paw Lake/River complex on the west as the
regional sink. Profile for the fen is shown in Figure 3.51. The model suggests that the fen
receives water from the adjacent shallow system as well as deeper regional and sub-regional
systems. The model also shows that most of the deeper flow from Cass-Kalamazoo Groundwater
Mound is going to Paw Paw Lake but part of this deep regional flow is also providing water to
the fen.

118

Estimated
regional
groundwater
flow path

Paw Paw
Lake Fen

Selected
wells

Paw Paw Lake

Extent of
profile model

2km
Wetlands

Well Logs from Selected wells

Figure 3.50

Well

location

Clay in well lithologies

and

well

logs

along

Paw

Paw

Lake

Fen

capture zone. Contours at 1m interval show SWL for drift.

119

Figure 3.51

Profile Model for Paw Paw Lake Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 0.48m

interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen’s source water (purple) [bottom].

120

Groundwater Profile Model – Paw Paw Prairie Fen (67th Street)
Paw Paw Prairie Fen symmetrically occurs on either banks of Paw Paw River. The
capture zones from either sides extend in opposite directions as shown in Figure 3.49. A single
profile model was built to cover both parts of the fen. Analysis of lithology from the well logs
along the chosen profile, as shown in Figure 3.52, did not reveal continuous clay lenses within
the glacial drift. The profile model is shown in Figure 3.53. Regional and sub-regional
groundwater mounds at either ends of the profile serve as sources and Paw Paw River in the
middle serves as the sink. The model predicts that wetlands adjacent to both parts of the fen
receive water from the shallow groundwater systems whereas the fen receives upwelling
groundwater from the deeper system connected to regional/sub-regional groundwater mounds.

Paw Paw
Lake Fen

Extent of
profile model
Selected
wells
Lakes

Paw Paw River
Estimated regional
groundwater flow
path

Wetlands
2km

Well Logs from Selected wells

Clay in well lithologies
Figure 3.52

Well location and well logs along Paw Paw Prairie Fen
capture zone. Contours at 1m interval show SWL for drift.
121

Figure 3.53

Profile

Model

for

Paw

Paw

Prairie

Fen.

Geological

layers

and

delineation

of

groundwater head and velocity [top]. Light blue lines show groundwater head contours

at 0.84m interval. Reverse particle tracking to estimate groundwater flow regime
(pink) and likely fen’s source water (purple) [bottom]

122

Monette St Fen
Monette St Fen is one of the largest prairie fens in Michigan occurring in proglacial
outwash overlying shale bedrock. The average drift thickness around the fen site is 87m, varying
between 112m and 83m. The fen and its surface catchment area are shown on 10m DEM relief in
Figure 3.54. EORANK of the fen is „A?‟. MS occurrence has not been reported at this site.

Surface catchment for fen
Monette St Fen

Wetlands

1 KM

0

Bedrock (Shale)

Glacial Drift

Figure 3.54

Monette St Fen. The shaded relief represents topography
derived from 10m DEM varying from 241.03m to 261.10m.

123

Regional Groundwater Model
Regional model for Monette Street Fen was selected over an area of 8.76 km by 10.22
km. Model area was divided into the 300x285 grid cells, with the cell size of 29.04 m. Static
water level (SWL) and other modeling parameters were ascertained as explained in
methodology. Reverse particle tracking was used to delineate groundwater capture zone for the
fen. Delineated zone of groundwater contribution to the fen is shown in Figure 3.55.

Southern high of CassKalamazoo Groundwater
Mound

Groundwater
capture zone
Monette St Fen

3 KM

0

Extent of
Surface water
catchments for fen

Figure 3.55

The

Monette St Fen
Water well
data points

contours

represent

at

regional

0.61m
scale

interval

and

delineation

shaded
of

relief

groundwater

surface around Monette St Fen which varies in elevation
from 241m to 260m.

124

The model suggests that the contribution zone may be connected the southern high of
Cass-Kalamazoo Groundwater Mound approximately 6 km to the north of the fen site. Strong
bending of regional contours closer to the fen sites location suggests regional flow convergence
and possibility of regional upwelling in the area. Please also refer to Figure 4.2 for more
discussion. In Figure 3.56, the groundwater contribution zones are superimposed on surface
topography and geological land forms. The fen occurs in proglacial outwash and has a few
wetlands within its surface water catchment area.
Regional
Watershed

Groundwater
capture zone

3 KM

0

Surface water
catchments
for fen

Proglacial outwash
Ice-contact outwash

Monette St Fen

Figure 3.56

The

Till
shaded

relief

represent

surface

topography

in

region varying in elevation from 241.06m to 298.58m.

125

the

Groundwater Profile Model
Well logs along the capture zone of Monette St Fen are shown in Figure 3.57 which do
not reveal any structural discontinuities within the drift. The drift is overlying shale formations
which were considered as impervious. The profile model was conceptualized as single layer of
drift as shown in Figure 3.58, with Diamond Lake as the regional source at the northern end and
the complex of Christina Lake and adjacent wetlands as the regional sink. The model predicts
that the fen may be receiving groundwater from a sub-regional recharge area approximately 7 km
to the north of the fen. The fen receives water through deep upwelling. Shallow groundwater
systems do not seem to contribute directly to the fen, but instead feed the adjacent wetlands
around the fen site.

Diamond Lake

Extent of profile
model

Estimated
regional
groundwater
flow path

Selected wells
Fen

Juno Lake
Eagle Lake
Christina Lake

2km

Well Logs from Selected wells

Figure 3.57

Clay in well lithologies

Well locations and well logs along Monette St Fen capture
zone.

Contours

at

0.61m
126

interval

show

SWL

for

drift.

Figure 3.58

Profile Model for Monette St Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 0.90m

interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen’s source water (purple) [bottom].

127

Springbrook Fen and Butterfield Lake Fen
Springbrook Fen occurs in proglacial outwash besides the eastern bank of Spring Brook –
a first order stream, within a wetlands complex which runs alongside the stream. The fen and its
surface catchment area are shown on 10m DEM relief in Figure 3.59. EORANK of the fen is „C‟.
MS occurrence has been reported at this site.

Fen’s surface
catchment

Spring Brook

Springbrook
Fen

Wetlands
2 KM

Proglacial outwash
Ice-contact outwash
0

Figure 3.59

Springbrook

Fen.

The

shaded

relief

represent

surface

relief derived from 30m/10m DEM. Surface elevation varies
between 255.36m and 293.80m.

128

Butterfield Lake Fen occurs in proglacial outwash adjacent to southern edges of
Butterfield Lake and along either banks of Gull Creek – a first order stream. The relief around
the fen is hummocky and dominated by a complex of wetlands along Gull Creek. The fen and its
surface catchment area are shown on 10m DEM relief in Figure 3.60. EORANK of the fen is
„CD‟. MS occurrence has not been reported at this site.

Butterfield Lake

Fen’s surface
catchment

Wetlands

1 KM

0

Gull Creek

Butterfield Lake Fen

Figure 3.60

Butterfield Lake Fen. The shaded relief represent surface
relief derived from 30m/10m DEM. Surface elevation varies
between 256.49m and 282.75m.

Regional Groundwater Model
Being in the close vicinity of each other, the two fens were covered in a single regional
model over an area of 12.94 km by 16.28 km. The model area was divided into the 250x313 grid
cells, with the cell size of 51.68 m. Static water level (SWL) and other modeling parameters
were ascertained as explained in methodology. Reverse particle tracking was used to delineate
129

groundwater capture zones for the two fen sites in the model. Delineated zones of groundwater
contribution to the fens are shown in Figure 3.61.
The model suggests that the contribution zone for Butterfield Lake Fen extends to Barry
Groundwater Mound, going through Gull Lake. Gull Lake in the region is a flow through lake
which may also be receiving groundwater input from Barry Groundwater Mound (from north)
and discharging water to the aquifer (from south). From the regional picture it appears that
Butterfield Lake Fen may be directly or indirectly linked to Barry Groundwater Mound. For
Springbrook Fen, the model suggests that it may be connected to Barry Groundwater Mound.
Bending of regional contours closer to the fen site suggests regional flow convergence and
possibility of regional upwelling in the area. Please also refer to Figure 4.2 for more discussion.

130

Butterfield Lake
Fen Groundwater
capture zone
upstream of Gull
Lake

Barry
Mound

Butterfield Lake
Fen Groundwater
capture zone
downstream of
Gull Lake

Springbrook Fen
Groundwater
capture zone

Butterfield
Lake Fen
Drift

Springbrook Fen
4 KM

Bedrock

0

Figure 3.61

The contours at 2m interval and shaded relief represent
regional scale delineation of groundwater surface around
Springbrook and Butterfield Lake fens. The SWL elevation
ranges from 240.36m to 281.64m.

In Figure 3.62, the groundwater contribution zones are superimposed on surface
topography and geological land forms. Groundwater capture zones of both fens remain confined
to their respective regional watersheds. The glacial drift thickness in the region varies between
24m and 123m with an average of 70m. The drift thickness is more in the northwest of the model
131

area and lesser in the southeast. The drift in the area comprises outwash deposits. Both fens
occur in proglacial outwash.

Butterfield Lake Fen
Groundwater capture zone
upstream of Gull Lake

Butterfield Lake Fen
Groundwater capture
zone

Springbrook Fen
Groundwater
capture zone

Regional
watersheds
Drift
4 KM

Proglacial outwash

Bedrock

Ice-contact outwash
0

Figure 3.62

The

shaded

relief

represent

surface

topography

in

the

region which varies in elevation from 240.36m to 321.27m.

Outwash deposits in the region are underlain by two different bedrock formations. The
formations are Coldwater Shale and Marshall Sandstone as shown in Figure 3.63. While
Marshall Sandstone is considered very pervious formation, Coldwater Shale is considered almost

132

impervious. Both fens occur over Coldwater Shale but their capture zones extend over to the
Marshall formation.
Butterfield Lake
Fen Groundwater
capture zone
south of Gull Lake

Springbrook Fen
Groundwater
capture zone

4 KM

0

Marshall Sandstone
Coldwater Shale
Figure 3.63

The shaded relief represent bedrock surface in the region
which

varies

formations

in

in

elevation

the

area

from

are

175m

to

represented

238m.
by

Bedrock

different

colors. Both fens occur in drift deposits over Coldwater
Shale.

Groundwater Profile Model – Springbrook Fen
Well logs along the capture zone of Springbrook Fen are shown in Figure 3.64 which do
not reveal any structural discontinuities within the drift. The profile model was conceptualized as

133

a layer of drift overlying bedrock (sandstone and shale formations), as shown in Figure 3.65,
with Barry Groundwater Mound as the regional source at the northern end. Steep regional
groundwater gradient (Figure 3.61) at the southern end of the profile was modeled as constant
head to allow boundary outflow. The model predicts that the fen may be receiving groundwater
from shallow systems immediately adjacent to it, extending up to 4 km northward from the fen.
The deeper systems from the Barry Groundwater Mound seem to bypass the fen.

A

Model extent

Selected wells
Estimated regional
groundwater flow path

Fen

4km

B
A

B

Well Logs from Selected wells
Clay in well lithologies

Figure 3.64

Well

locations

and

well

logs

along

Springbrook

Fen

capture zone. Contours at 2m interval show SWL for drift.

134

Figure 3.65

Profile Model for Springbrook Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 1.02m

interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen’s source water (purple) [bottom].

135

Groundwater Profile Model – Butterfield Lake Fen
Well logs along the capture zone of Butterfield Lake Fen are shown in Figure 3.66 which
do not reveal any structural discontinuities within the drift. The profile model was
conceptualized as a layer of drift overlying bedrock (sandstone and shale formations), as shown
in Figure 3.67, with Barry Groundwater Mound as the regional source at the northern end. Steep
regional groundwater gradient (Figure 3.61) at the southern end of the profile was modeled as
constant head to allow boundary outflow. The model predicts that the fen may be receiving
groundwater from deeper system, upwelling from Marshall Formation and connected to Barry
Groundwater Mound approximately 12 km northward from the fen, bypassing Gull Lake. The
shallow systems around the fen may be contributing to wetlands and lakes adjacent to the fen.

Model extent

Selected wells

Estimated regional
groundwater flow path

Gull Lake

Fen

4km

Clay in well lithologies

Well Logs from Selected wells

Figure 3.66

Well locations and well logs along Butterfield Lake Fen.
Contours

at

2m

interval

136

show

SWL

for

drift.

Figure 3.67

Profile Model for Butterfield Lake Fen. Geological layers and delineation of
groundwater head and velocity [top]. Light

blue lines show groundwater head

contours at 0.99m interval. Reverse particle tracking to estimate groundwater
flow regime (pink) and likely fen’s source water (purple) [bottom].

137

Dew Road Fen
Dew Road Fen is one of the few prairie fens occurring in ice marginal till. The fen is
adjacent to a wetlands complex.

Bedrock formation under the glacial drift is Marshall

Sandstone. Average thickness of glacial drift around the fen is 31m, with maximum 71m and
minimum 1m. The bedrock almost pinches to the surface about 800m to the west of the fen. The
fen and its surface catchment area are shown on 10m DEM relief in Figure 3.68. EORANK of
the fen is „C‟. MS occurrence has not been reported at this site.

Surface water
catchment for fen

Bedrock

Lakes

Dew Road Fen
4 KM

Wetlands
0

Figure 3.68

Dew

Road

Fen.

The

shaded

relief

represent

surface

topography derived from 30m/10m DEM. Surface elevation
varies between 309.46m and 363.80m.

138

Regional Groundwater Model
Regional model for Dew Road Fen was selected over an area of 5.94 km by 6.99 km. The
area was divided into the 120x140 grid cells, with the cell size of 49.89 m. Static water level
(SWL) and other modeling parameters were ascertained as explained in methodology. Reverse
particle tracking was used to delineate groundwater capture zone for the fen. Delineated zones of
groundwater contribution to the fens are shown in Figure 3.69.
Shallow bedrock
Groundwater
capture zone

Hillsdale Mound
(northern extent)

600 M

Bedrock
(Marshall
Sandstone)

Fen surface
catchment
Dew Road Fen

0

Figure 3.69

The contours at 0.5m interval and shaded relief represent
regional scale delineation of groundwater surface around
Dew Road Fen which varies in elevation from 306.59m to
321.97m.

139

The model shows that the contribution zone extends beyond the surface water catchment
area of the fen and may be connected to the northern end of Hillsdale Groundwater Mound,
approximately 5 km to the south from the fen site. In Figure 3.70, the groundwater contribution
zone is superimposed on surface topography and geological land forms. The contribution area
extends across the till plains all the way to a zone of glacial outwash in another surface
watershed. Fen‟s surface catchment is small and short compared to the delineated capture zone.

Groundwater
capture zone
Fen surface
for fen
catchment

Regional
Watershed

600 M

Outwash

Dew Road Fen
Till
0

Figure 3.70

The shaded relief represents surface topography around
Dew Road Fen which varies in elevation from 309.40m to
362.92m. The fen occurs in ice-contact till while the
capture zone extends south to glacial outwash deposits.

140

Groundwater Profile Model
The profile section selected across the fen and its regional capture zone is shown in
Figure 3.71. The section starts at Hillsdale Mound, being the regional source and ends at the
large wetlands complex just north of the fen, where a very steep regional gradient exists. This
end the profile was hence modeled as constant head to simulate boundary flow due to steep
gradient.
Section for
Dew Road Fen
Profile Model

1 KM

0

Figure 3.71

Section

for

topography

profile
and

model.

contours

at

Shaded

relief

0.5m

interval

represents
represent

regional SWL.

Profile model for Dew Road Fen is shown in Figure 3.72. The model shows that the fen
may not be connected to the regional system, drawing water only from the adjacent shallow
groundwater system extending approximately 1.5 km to the south from the fen. The regional
flow does not seem to up-well into the ice-marginal till or into the fen but bypasses the site
flowing through the Marshall formation.
141

Figure 3.72

Profile Model for Dew Road Fen. Geological layers and delineation of groundwater
head and velocity [top]. Light blue lines show groundwater head contours at 0.84m

interval. Reverse particle tracking to estimate groundwater flow regime (pink) and
likely fen’s source water area (purple) [bottom].

142

Silver Lake Fen
Silver Lake Fen is one of the few prairie fens in Michigan occurring in ice marginal till.
The average drift thickness around the fen site is 110m, varying between 85m and 140m. The fen
and its surface catchment area are shown on 10m DEM relief in Figure 3.73. EORANK of the
fen is „CD‟. MS occurrence has not been reported at this site.

Surface catchment for fen

Silver Lake Fen

Wetlands

1 KM

0

Bedrock (Redbeds)
Glacial Drift

Figure 3.73

Silver Lake Fen. The shaded relief represents topography
derived from 30m/10m DEM which varies from 260.55m to
303.04m.

143

Regional Groundwater Model
Regional model for Silver Lake North Fen was selected over an area of 8.68 km by 7.49
km, divided into 300x259 grid cells, with the cell size of 29.03 m. Static water level (SWL) and
other modeling parameters were ascertained as explained in methodology. Reverse particle
tracking was used to delineate groundwater capture zone for the fen site in the model. Delineated
zone of groundwater contribution to the fen is shown in Figure 3.74. The model shows that the
contribution zones may be connected to a sub-regional groundwater mound approximately 3 km
to the north from the fen location.

Sub-regional
groundwater mound

Groundwater
capture zone
for fen

Water well
data points

3 KM

Surface water
catchments for fen

Silver Lake
North Fen
0

Figure 3.74

The

contours

represent

at

regional

1.16m
scale

interval

and

delineation

shaded
of

relief

groundwater

surface around Silver Lake Fen which varies in elevation
from 240.50m to 266.36m.
144

In Figure 3.75, the groundwater contribution zones are superimposed on surface
topography and geological land forms. The model shows a narrow groundwater capture zone
extending north, culminating just across the regional watershed.
Regional
Watershed

Groundwater
capture path for
fen

Silver Lake
North Fen

3 KM

Glacial Till
Marginal Till
0
Figure 3.75

The

shaded

relief

represent

surface

topography

in

the

region which varies in elevation from 240.51m to 310.94m.
Glacial

landforms

in

the

area

are

represented

by

different colors. The fen occurs in glacial till.

Groundwater Profile Model
Analysis of lithology from the well logs around Silver Lake Fen did not reveal
continuous clay lenses within the glacial drift as shown in Figure 3.76. The model‟s profile
section starts in the north at a sub-regional groundwater mound acting as source and ends at the
wetlands complex just south of the fen, where a very steep regional gradient exists. This end the
145

profile was hence modeled as constant head to simulate boundary flow due to steep gradient. The
profile model is shown in Figure 3.77. The model predicts that the fen receives water from the
adjacent shallow system which extends all the way to the apex of the sub-regional groundwater
mound approximately 2.3 km from the fen site. The fen does not seem to receive water from the
deeper systems. The model also shows that the deeper regional flow of groundwater flow
bypasses the fen.

Lakes
Extent of
profile model
Selected
wells
Silver Lake Fen

Estimated regional
groundwater flow
path
Wetlands
2km

Well Logs from Selected wells

Clay in well lithologies

Figure 3.76

Well location and well logs along Silver Lake Fen capture
zone. Contours at 1.16m interval show SWL for drift.

146

Figure 3.77

Profile Model for Silver Lake Fen. Groundwater flow lines show estimated groundwater
flow regime (pink) and likely source water paths for the fen (purple). Light blue
lines show groundwater head contours at 0.74m interval.

147

Chapter 4

Discussion

BASIN-WIDE GROUNDWATER PICTURE
Statewide models revealed the larger picture of groundwater flow regime across the
southern half of Michigan‟s Lower Peninsula. Most fen sites seem to be occurring around the
regional groundwater mounds identified in the models. All major rivers in the modeled area also
seemed to have their headwaters emerging around the groundwater mounds. The placement of
mounds as well as the occurrence of fens generally coincides with the interlobate regions –
comprising mostly of pervious glacial outwash materials which have higher potential for
receiving groundwater recharge and contain more pore space for aquifer storage. Also refer to
Figure 3.2 and Figure 3.3.

REGIONAL UPWELLING
Horizontal flow models for some regions showed very strong convergence of regional
groundwater flow around the areas where fens were located. These models were first developed
using water wells data and taking large surface features into account. For some of the regions
showing particularly strong regional convergence, flow models were also made using only the
water wells data without taking into account the surface water features to ascertain if water wells
data alone also reveal a pattern of regional convergence. Some flow models, based entirely on
water wells data still suggested strong regional flow convergence around the fen locations. The
examples of 5 such regions are shown in Figure 4.1 and Figure 4.2 (also refer to Figure 3.28,
Figure 3.43, Figure 3.48, Figure 3.61, and Figure 3.55 for the same sites where regional flow

148

pattern was based on water wells data as well as large surface water data). Such strong signals
already suggest possibility of groundwater upwelling from the regional system. The flow
convergence though seems directed to a bigger surface water body such as a stream or a lake, the
fens sites within this „zone of convergence‟ also seemed benefitting from the regional system.

Paw Paw
Prairie Fen

Paw Paw
Lake Fen
Skiff Lake Fen

Converging
regional flow

Figure 4.1

Strong

regional

groundwater

flow

convergence

shown

by

water wells data around Paw Paw Lake and Paw Paw Prairie
fens (left) and Skiff Lake fen (right).

Many modeled sites which did not reveal clear „convergence zones of regional flow‟
around the fen locations still showed components of regional upwelling through profile models.
Profile models, therefore, were helpful in assessing the possibility of regional connections,
especially when they were not readily apparent from the horizontal flow models.
149

Springbrook
Fen

Monette
Fen

Sarett Nature
Center Fen

Converging
regional flow

Figure 4.2

Strong
water

regional
wells

Monette

fen

groundwater

data
(top

around
right)

(bottom).

150

flow

convergence

Springbrook
and

Sarett

fen
Nature

shown

(top

by

left),

Center

fen

SOURCE WATER AREAS FOR FEN SITES
This study identified likely groundwater sources feeding the prairie fen sites across the
state. Based on the modeling results, the source water areas for the fen sites may be divided into
three main categories as explained below. The models have also shown that a single site could be
connected to more than one source water areas.
Local Sources
Wetlands, lakes and stream close to a fen site (from a few 100 meters to a couple of
kilometers at the most) may act as source water for the groundwater seeping into the fen. Such
sources, though close to the fen, were found to be separated from the fen sites by small
topographic ridges. The ridges prevent the direct runoff from these surface sources into the fen.
Consequently, the surface runoff into these local sources turns into „groundwater‟ as it seeps
underneath the topographic ridges. The ridges are mostly made up of coarse glacial landforms
such as kames or eskers. Of the 19 fen sites modeled in this study, 13 sites seemed to be
receiving groundwater from their nearby local sources. Also see Table 4.I.
Sub-Regional Sources
Surface waters such as wetlands/lakes/streams or recharge areas in the uplands, far from a
fen site (from a couple of kilometers to about ten kilometers), may act as source water for the
groundwater seeping into the fen. Since these areas are not immediately adjacent to the fen sites,
they may not be easily identified as source water to a particular site without the knowledge of
regional groundwater dynamics. Of the 19 fen sites modeled in this study, 8 sites seemed to be
receiving groundwater from sub-regional sources. Also see Table 4.I.

151

Regional Sources
Regional recharge areas between the river basins with sizeable groundwater mounds, far
away from a fen site (tens of kilometers away), may act as source water for the groundwater
seeping into the fen. Generally such sources are huge and may be connected to more than one fen
site. Without the knowledge of basin-scale flow regime, as well as site specific details around a
fen site, it is difficult to ascertain whether a regional source might be connected to a particular
fen site. Of the 19 fen sites modeled in this study, 6 sites seemed to be receiving groundwater
from regional sources. Also see Table 4.I.

DELIVERY MECHANISMS OF GROUNDWATER AT FENS SITES
Profile models in the study implied distinct delivery mechanisms linking source water
areas to the fen sites. These mechanisms are explained as follows. Also refer to Table 4.I and
Figure 4.10 through Figure 4.13 for examples of delivery mechanism from the modeled sites.
Shallow Connections
Groundwater recharge from local hills and surface water in lakes/wetlands at higher
elevations adjacent to a fen may seep into the fen through shallow, and relatively short,
groundwater flow paths. Figure 4.3 shows the schematic representation of an adjacent shallow
connection.

152

Figure 4.3

Delivery mechanism – adjacent shallow connections.

The topography around the fen is often such that a natural „topographic dyke” forms on
their upstream sides, preventing significant direct runoff into the fen. Surface runoff due to the
dykes is mostly diverted away from the fen, generally accumulating in an adjacent lake/wetland.
The lake/wetland, if higher than the fen, may be hydraulically connected to the fen through a
shallow connection. We termed this special case of shallow connection as the “dyke effect”
which essentially converts local precipitation/runoff into “groundwater” before it enters the fen.
The source water in case of dyke effect is the local precipitation and runoff.
Deep Connections
A deep groundwater system, with its source far from the fen site, may be upwelling into a
fen as illustrated in Figure 4.4. In this study, the delivery of groundwater to a fen site through
deep upwelling is termed as „deep connection‟. The path that the groundwater follows may (or
my not) come in contact with deeper bedrock formations besides the deeper glacial drift deposits.
Figure 4.4 shows the schematic representation of a deep connection.

153

Figure 4.4

Delivery mechanism – deep connections.

Some fens seemed to have their groundwater source area tens of kilometers away from
the site and are linked to an “unexpected” strong connection to the regional flow in addition to
local flow. Shallow seeps/local hills may be more visible, but it is the regional flow paths that
may be “pumping” steadily into these fens, through the deep aquifer, providing
significant/continuous discharge of calcareous groundwater. Time scale of groundwater flow
along flow paths leading to fens can be decades and therefore impacts of changing land use or
climate on fens may be delayed, yet quite protracted once they are manifested.
The upwelling may be through deep glacial deposits and/or from the bedrock aquifers.
Deep connections may be further sub-categorized into the following:
Local Upwelling
Source waters closer to a fen, upwelling into the fen through relatively shallower drift materials
are termed as „local upwelling‟ in this study as shown in Figure 4.5. Such sources are generally
within 1 to 3 kilometers from the fen sites. The flow paths from the source to the fen run only
through shallow drift materials.

154

Figure 4.5

Upwelling – local source.

Sub-regional Upwelling
Distant sources of groundwater, upwelling through deeper drift materials and/or bedrock
formations into a fen site, are termed as „sub-regional upwelling‟ in this study as shown in Figure
4.6. These sources are generally between 5 to 10 kilometers from the fen sites. The flow paths
from the source to the fen may run through deeper glacial drift and/or the bedrock formations.

Figure 4.6

Upwelling – sub-regional source.

Regional Upwelling
Groundwater upwelling into a fen from a regional source after moving through deep glacial
and/or bedrock formation is termed as „regional upwelling‟ in this study as shown in Figure 4.7.
These sources are generally over 10 km from the fen sites.

155

Figure 4.7

Upwelling – regional source.

Confined Connections
Glacial landforms in Michigan classified as till plains, at some locations, may be
overlying pervious outwash from the earlier glaciations, creating sub-regional confined aquifers.
As shown in Figure 4.8, while a source water area and a fen site may be located over unconfined
aquifers, the groundwater flow path (or part or the path) between the source and the fen may be
channeled through a confining clay layer before upwelling into the fen. Such a groundwater
connection is termed as a „confined connection‟ in this study.

Figure 4.8

Delivery mechanism – confined connections.

Some fens seem spatially isolated from the „usual‟ fen clusters in interlobate outwash
region, but may be hydrologically connected to the regional recharge sources through confined
connections over tens of kilometers. The residence times in confined connections can be in

156

decades. The till plains also provide natural protection to groundwater flow paths from land-use
perturbations.
Cascading Connections
A groundwater source connected to a fen site may be dependent on one or more sources
of surface water and/or groundwater. The fen connected to such a source would hence be
indirectly dependent on other sources too as shown in Figure 4.9. Such indirect groundwater
connections between the sources and the fen sites are termed as „cascading mechanisms‟ in this
study.

Figure 4.9

Delivery mechanism – cascading connections.

Just identifying a source area feeding the fen may not be enough when the identified
source is further dependent on other sources. If a local source is dependent on a regional
recharge zone far from the fen, then the regional zone cannot be ignored while managing the fen.
Similarly, a surface water source connected to a fen is dependent on its surface catchment,
making its entire catchment critical for the fen.
Following figures illustrate some examples from of the delivery mechanisms discussed
above from the prairie fens investigated in this study.

157

Figure 4.10

Example from Liberty Fen – the fen site exhibits sub-regional upwelling as
well as shallow connections.

158

Figure 4.11

Example from Ives Road Fen – the fen site exhibits shallow, confined and
cascading connections.

159

Figure 4.12

Example from Skiff Lake Fen – the fen site exhibits adjacent, local upwelling
and sub-regional upwelling connections.

160

Figure 4.13

Example from MacCready Fen – the fen site exhibits regional and sub-regional
upwelling connections.

161

Table 4.I shows the summary of investigated sites. Flow paths through drift aquifer are
common to all 8 sites investigated with MS occurrence. Moreover, all but one of these sites
(Turner Creek Fen) is not connected to the local shallow system. And all but one of the MS sites
(McKibben Fen) does not seem to be receiving water from an upwelling connection, but this site
is located very close to major sub-regional source.
Of the 19 modeled sites, six sites show little connection with the local shallow systems
and seem mostly dependent on regional or sub-regional upwelling. Interestingly, none of these
sites is connected to multiple sources of groundwater.
Of the 150 prairie fens in Michigan, only seven occur in glacial till. Two of these sites
were included in detailed modeling. Both sites seem to exhibit only shallow connections to the
local groundwater sources.
From the 8 modeled sites with MS occurrence, there seems no particular similarity in
source water areas or delivery mechanisms. The ecological ranking (EO Rank) too does not
exhibit any particular pattern/similarity with the groundwater delivery mechanism or source
water areas.

162

Table 4.I

Summary of modeled fen sites

Upwelling

Flowpaths in
Aquifer

Geological
Landform

EO
Rank

Satyr

L

S

D,B

PGO

A

Y

L

S,L

D,B

PGO

BC

Y

MacCready

L

R,S

D,B

PGO

-

Dew Road

L

D

IMT

C

Ives Road

L

D

PGO

A

R

B

PGO

CD

Liberty
Skiff Lake

Butterfield

Confined

Fen Site

Cascading

Adjacent

Types of Sources & Connections

R

R,L

Spring-brook

L

L

D

PGO

C

Y

Paw Paw Lake

L

R

D

PGO

B

Y

R

D

ICO

BC

D

CL

C

S

D

PGO

A?

Paw Paw Praire
Sarett

L

Monette

S,L

S

Silver Lake

L

L

D

IMT

CD

Bowens M

L

R

D,B

PGO

C

Turner Ck

S

D,B

PGO

BC

Hill Creek

S

D

PGO

B

McDonald

S

D

ICO

C

D

PGO

BC

D

ICO

C

D

PGO

C

Y

Yankee Spring

L

L

Hall Lake

L

S

McKibben

L

S,L

Sources:

L = Local

S = Sub-regional

Aquifers:

D = Drift

B = Bedrock

Landforms:

R = Regional

ICO = Ice-contact outwash

IMT = Ice marginal till
EO Ranks

PGO = Proglacial outwash

CL = Coarse lacustrine

As defined by Nature Serve (NatureServe 2002)

163

Y

Y

Y

MAJOR IMPLICATIONS OF FINDINGS
The above findings have various management implications. The regional sources of
significance identified in the study need to be protected to preserve fens even when those are tens
of kilometers away from the actual sites. Understanding of hydrological regime at statewide and
regional scales can help coordinate statewide conservation efforts in a synergistic way to protect
and manage prairie fens. The methodology presented in this research could also assist identify
data needs and knowledge gaps at local and regional scales. The approach presented here for
prairie fens may as well be applied to other groundwater dependent ecosystems.
Protection of Source Water Areas
Surface water bodies such as lakes and wetlands which make a groundwater source for a
fen site need to be protected to protect the fen. A surface water body may be relatively small, but
its surface water catchment may be large which might imply more regulatory limitations in the
entire surface catchment of the source.
Groundwater contribution area for the source waters also need to be protected, however,
unlike the surface catchment areas, which can be readily ascertained through topography,
delineation of groundwater contribution areas would require hydrogeological investigations
and/or modeling – making it expensive and time consuming.
Some regional and sub-regional groundwater sources may be linked to more than one fen
site. Protecting such sources would protect multiple fen sites.
Fens connected to distant sources through distant and regional upwelling may have their
source water areas in a distant and disjoint (from the fen) location. Yet, the capture zone in the

164

plan view may need to be protected to disrupt quality and quantity of groundwater within the
path leading to the fen from the source.
Groundwater flow paths through the confined connections may be naturally protected
through impervious layers between the fen and source water areas.
Basin Wide Perspective
Oakland Mound, Jackson Mound and parts of Hillsdale Mound identified in this study
coincide with the Michigan Recharge area for southeastern Michigan as identified in a report by
Michigan Department of Natural Resources and proposed as one of the Forest Legacy areas in
the state (MDNR 2003). The comparison of delineation of groundwater mounds with proposed
Forest Legacy area is shown in Figure 4.14. The areas delineated for mounds, therefore, are not
only important of fens but also coincide with the known regions of ecological significance.
It is interesting to note in Figure 4.14 (bottom) that groundwater mounds are sitting
across the regional surface water divides and are large enough to spread out in more than one
political jurisdiction. The surface area over these mounds makes a contiguous region of
hydrogeological and ecological significance overlapping political, topographic and hydrological
boundaries. Hillsdale Mound, for example, spans seven counties in three states plus intersects
watersheds of five major rivers in three different states (Grand, Kalamazoo, St. Joseph, Tiffin
and Raisin rivers).

165

0

60 KM

Oakland
Mound

Jackson
Mound
Hillsdale
Mound

Figure 4.14

Michigan
(MDNR

Recharge

2003)[top];

Michigan

[bottom]

area

proposed

Groundwater
coincide

Legacy Area.

166

as

Forest

mounds

with

the

in

Legacy

Area

South-eastern

proposed

Forest

Current paradigm of resource management along political boundaries is already outdated
while the emerging trend of basin scale management (Environment Canada 2009; EU 2000; EU
2006; GLPF 1998; IJC 2009; NSW CMA 2004; Price et al. 2005; SOGW 2009) may also have to
be given a thought for the situation where what happens within a buffer on the basin boundaries
may not be best managed by dividing the critically important buffer zone along the topographic
boundaries.

The holistic management of these contiguous areas may pose administrative

challenges whether managing areas along river basin boundaries or the political boundaries.
Another significance of the groundwater mounds identified in this study shows when
glacial landforms are correlated to the occurrence of these mounds. The mounds generally
coincide with the pervious glacial outwash deposits, flanked on either sides by less pervious
surfaces made up by till plains and fine lacustrine landforms as shown in Figure 4.15. However,
the presence of significant number of groundwater wells in the areas of till and fine lacustrine
deposits indicates presence of more pervious glacial or bedrock formation under the till plains.
While direct recharge to these aquifers through less pervious till layers may be limited, the
groundwater mounds sitting between them in more pervious deposits and on higher elevations
may be acting as the recharge „window‟ to the aquifer systems in bedrocks/confined coarse
glacial aquifers under the till plains and fine lacustrine surfaces. This possibility along Section
AA‟ in Figure 4.15 is shown in Figure 4.16 and can be explored further with more data and
detailed modeling. The section was derived from GIS data.

167

A

Oakland
Mound

Mostly till deposits

A’

Jackson
Mound

Mostly till and fine
lacustrine deposits

Hillsdale
Mound

0
Outwash

Figure 4.15

50 Km

Till

Lacustrine Fine

Glacial landforms and Michigan’s recharge areas

168

Figure 4.16

Recharge dynamics at basin-scale. Vertical exaggeration is 50x.

169

Policy, Management and Development
With source water areas of significance identified at regional scales, the Policymakers
can make more informed decisions with regard to setting and enforcing laws and regulations for
water resources management and to evaluate future land use management plans related to zoning
and new developments. The resource managers and planners can become more effective in
identifying/prioritizing areas/sites for monitoring, development, conservation, or protection.
They can also be more effective in engaging the general public and informing high-level decision
makers about the implications of a proposed development and the transport of contamination on
source water areas. The consultants will be able to design more focused, cost effective analysis
and monitoring networks to protect state‟s water resources and environment.

170

Chapter 5

Conclusions

This research represents first systematic modeling study of prairie fen hydrology in
Michigan, spanning across local, regional, and basin-wide scales, taking into account a large
number of sites with diverse hydrogeological characteristics. We developed a basin-scale model
for the entire southern Michigan basin, regional scale models for 9 regions, and site-specific
models for 19 selected fens. We also developed vertical profile models to understand the
groundwater delivery mechanisms from source waters to the fens sites. Following is the
summary of significant findings from these multi-scale models.

SUMMARY OF SIGNIFICANT FINDINGS


Although at local scale prairie fens look scattered, unrelated, or even accidental, on a
regional scale they exhibit a clear, systematic pattern.



The vast majority of the prairie fens are located around or at the foot of several large
groundwater “mounds” at the intersection of major watersheds. These extensive mounds
are critical, regional source water areas for Michigan‟s aquifer systems, fens, and other
groundwater-dependent ecosystems (e.g., cold water trout streams, and many unique
plant and animal species);



These critical regional mounds may play a disproportionally large role controlling the
sustainability of groundwater resources and groundwater-dependent ecosystems and
biodiversity in the Michigan basin;



Groundwater flow patterns in the regional source water areas are complex and most fens
are recharged from multiple groundwater sources.
171



Prairie fens located seemingly far apart in different watersheds, different counties, or
even different states may share the same regional source water areas.



The data driven models reveal four distinct mechanisms of groundwater delivery into the
fens:
o shallow connections – groundwater recharge from local hills and surface water in
lakes/wetlands at higher elevations adjacent to a fen may seep into the fen through
shallow and relatively short groundwater flow paths. Time scale of groundwater
flow along flow paths leading to fens can be up to a couple of years. The
topography around the fens is often such that a natural „topographic dyke” forms
on their upstream sides, preventing significant direct runoff into the fen. This
“dyke effect” essentially converts local precipitation/runoff into “groundwater”
before it enters the fen.
o deep connections - where source water recharge comes in contact with deeper
glacial/bedrock formations before upwelling into the fen; Some fens seemed to
have their groundwater source area tens of kilometers away from the site and are
linked to an “unexpected” strong connection to the regional flow in addition to
local flow. Shallow seeps/local hills may be more visible, but it is the regional
flow paths that may be “pumping” steadily into these fens, through the deep
aquifer, providing significant/continuous discharge of calcareous groundwater.
Time scale of groundwater flow along flow paths leading to fens can be decades.
o confined connections - where a distant source water recharge area gets connected
to a fen through older outwash beds confined under recent till plains; Some fens
seem spatially isolated from the „usual‟ fen clusters in interlobate outwash region,

172

but these fens may be hydrologically connected to the regional recharge sources
through confined connections over tens of kilometers. The residence times in
confined connections can be in decades. The till plains also provide natural
protection to groundwater flow paths from land-use perturbations.
o cascading connections - where groundwater sources directly linked to fens are
dependent on other surface/sub-surface source waters; Just identifying a source
area feeding the fen may not be enough when the identified source is further
dependent on other sources. If a local source is dependent on a regional recharge
zone far from the fen, then the regional zone cannot be ignored while managing
the fen. Similarly, a surface water source connected to a fen is dependent on its
surface catchment, making its entire catchment critical for the fen.


groundwater flow times from regional/sub-regional source areas (5 to 25 km from the fen
sites) to the fen sites ranged between 10 and 100 years; groundwater flow times from
local source water areas (1 to 5 km from the fen sites) showed times ranging between
1and 4 years; with timescales as these, the impacts of changing land use or climate on
fens may be delayed, yet quite protracted once they are manifested.

SUMMARY OF IMPLICATIONS
Since water well records and other GIS data are now available for free virtually
“anywhere”, our improved ability to use this data source creates new possibilities that have
basin-wide implications for the management of fen ecosystems. By systematically making use of
this readily available data, we have the potential to significantly reduce the cost of site
characterization in the context of source water delineation and expand our “world view” by
informing management practices from a “local site” to a “region”, a “watershed”, and ultimately
173

a “basin”, transforming “isolated, site-based actions” into “synergistic, multi-scale
conservation”.
This research has significant practical implications and supports the need to reconsider
the current priorities in prairie fen restoration strategies. In particular, the research suggests:


Fen management must move beyond water‟s (or ecosystem‟s) edge to account for the
impact of regional flow systems, if it is to be effective.



Because of the connectivity of the basin‟s groundwater systems, a few “smart” actions in
key locations (e.g., regional source areas) could yield tremendous ecological return, with
gains being more than additive.



When we protect one strategic component(s) of a system, we are protecting many. When
we protect fens and their upstream sources, we are also protecting other fens, aquifers,
streams, lakes, fishes, and ultimately the Great Lakes



A “damaged” or regionally-altered groundwater mound (e.g., by pollution, urbanization,
agriculture, mining, climate change) could impact the entire water resource and
ecological systems. Stressed fen communities can be an early warning for potentially
much more grave consequences to come.



The current often incongruent perspectives and fragmented responsibility of agencies
make it difficult, if not impossible, to manage efficiently and holistically the entire fen
ecosystem – the knowledge on regional hydrologic integrity of the systems would help
inter-agency synergy.

Occurrence of fens in the regional aquifer recharge areas at the intersections of watershed
boundaries implies that coordination across the river basins as well as the political divides and
between water resources and ecological communities would be imperative for protection of fens.
174

This implies that management jurisdictions for protection of fens can neither be based on
political boundaries nor can be based on (environmentally much preferred) river basin
boundaries. A management approach across the river basins and political boundaries has to be
thought of to protect the natural integrity in a holistic manner.

RELATING TO PREVIOUS WORK
Following is the comparison of the review of previous work, presented in Chapter 1, to
the work conducted in this study, highlighting the improvements/additional contributions this
study has made in the knowledge base related to modeling of fen hydrology.


Both local and regional scales of groundwater dynamics were investigated in this study
using the actual data. The nature of fen hydrology has long been recognized as multiscale (Almendinger and Leete 1998b; GLPF 1998; Gorham 1957; Meyboom 1967; Siegel
1988; Wassen et al. 1990), however, groundwater models used to investigate hydrology
of fens and wetland systems in most prior studies had been applied either at regional or at
local scales only (Acreman et al. 2007; Acreman and Miller 2004; Batelaan et al. 2003;
Cobb and Bradbury 2008; Gilvear et al. 1993; Grapes et al. 2006). In cases where
regional dynamics had not been modeled, data limitation at large scales had been one of
the major factors. The scales not addressed by models had mostly discussed by the
authors only in an intuitive sense.



This study is one of the first to focus on a larger number of fen sites in a large region
making up a geographically contiguous flow regime, analyzing regional scales and also
zooming in to local details for a large number of fen sites. Many previous studies have
investigated hydrology for larger number of fens (Acreman 2004; Almendinger and Leete

175

1998b; Amon et al. 2002; Godwin et al. 2002; Halsey et al. 1997), however, these studies
have generally investigated large scale regional dynamics surrounding many fen sites
without zooming into the local details at specific sites. Few studies have covered both
local and regional scales for multiple fen sites (Almendinger and Leete 1998b), but
mostly these sites happen to be in disjoint regions, and therefore, a flow regime picture
around a large number of sites had mostly been missing from these studies.


To understand vertical dynamics of fen hydrology in a 3D sense, it is one of the first
studies to choose vertical profiles based on horizontal delineation of flow paths, using
data, rather than choosing them arbitrarily. Careful selection of boundary conditions for
the profile models in this study were also based on the insights gained from the regional
models for horizontal flow. Most prior studies which employed the use of vertical profile
models have mostly chosen the profile sections arbitrary manner and have not connected
the vertical flow dynamics to the regional flow systems (Bleuten et al. 2006; Grootjans et
al. 2006; Loon et al. 2009).



This study is one of the first to describe delivery mechanisms of groundwater to the fen
sites using actual data at multiple scales, and have identified the fen sites exhibiting those
mechanisms. Delivery or transfer mechanisms of groundwater from source water areas to
the fens or wetlands have been discussed in previous studies (Acreman 2004; Amon et al.
2002; Meyboom 1967) but most of these studies have either used hypothetical situations
in simulation models to explore the possibilities or have discussed the mechanisms with
hydrogeologic intuition only.



In this study we used statewide data to delineate subsurface hydrology across the river
basins. The results from this study led to on-ground identification of source water areas

176

of regional significance. It has been well recognized that the focus of conservation of
ecosystems should be shifted from the number of hectors lost or degraded to the
alteration in environmental and hydrological variables at regional scale (GLPF 1998).
Very few studies, however, have actually investigated connection of fen sites to the
groundwater dynamics at river-basin scales. Most regional studies had been based on
hydrologic intuition and judgments rather than the actual data at larger scales (Acreman
2004; Amon et al. 2002; Halsey et al. 1997). In many prior studies the conservation
potential has been considered higher for the sites with less significant modifications in
the recharge areas of their source waters (Bedford 1999; Koerselman 1989; Wassen et al.
1990; Winter 1998), but few studies have actually identified those source water areas,
especially if they are regional in nature. Data limitations at multiple scales had mostly
been the major limiting factor (Price and Waddington 2000).


In this study we not only delineated basin-wide regional groundwater system for southern
Michigan but also identified connections between regional and local groundwater sources
for 19 selected fen sites using actual data at every scale of investigation. The likelihood
that a fen receives groundwater from more than one flow systems has been intuitively
discussed in many previous studies (Almendinger and Leete 1998a; Amon et al. 2002;
Bedford 1999; Bedford and Godwin 2003), but it is difficult to find the work which
actually shows such hydrological connections and delivery mechanisms at specific sites
using actual data at multiple scales.



This is one of the first studies in Michigan (Great Lakes region) which have vividly
linked the local and regional dynamics of sub-surface flow regime to protect prairie fen
ecosystems across the state by seamlessly integrating the local and regional scales. A

177

“Great Lake Expert‟s Roundtable” (Great Lakes Protection Fund 1998) stressed to shift
the focus of current conservation efforts for ecosystems, which are too disjointed and
focused too much on local/visible symptoms, to protecting the underlying processes, the
functionalities, and the fundamental biological engine. Not many studies could be found
which have investigated the fundamental processes at multiple scales to help protect,
restore or enhance knowledge on the underlying hydrological processes on which the
ecosystems are dependent.

STRENGTHS AND LIMITATIONS OF THE STUDY
One of the major strengths of this study is the use of existing GIS-data which made multiscale hydrological investigations/modeling possible in a cost effective manner.

However,

following are the major limitations which could not be adequately addressed in the models
presented in this study:
Models Validation
Data from statewide GIS databases was used to built models and draw conclusions.
However, no independent datasets were available to validate modeling results. Of all the sites
presented in the study, site visits were made only to Liberty Fen, Skiff Lake Fen, MacCready
Fen and Ives Road Fen. The visits aimed only getting the „look and feel‟ of the sites rather than
elaborate data collection. There were signs of upwelling of groundwater in lakes and streams.
Artesian wells were observed in the area. Presence of marl deposits suggested groundwater from
bedrock formations feeding the streams and lakes. Vegetation observed in some lakes/wetlands
was typically „groundwater fed‟ (observed by Dr. Steve Hamilton – from Kellogg Biological
Station). Though these observations seemed consistent with the modeling results, it was not

178

possible within the study resources to further investigate and collect sufficient data that could be
used to validate modeling results.
Limitations of Profile Models
The profile models presented in the study are part „data-driven‟ and part „process-driven‟.
The boundary conditions and water-table profiles in the models were acquired from data
interpolation, while the particle tracking were based on process based simulations. The profiles
were selected along the regional flow trends with an underlying assumption that no flow takes
place across the profile. The shallow systems, however, were not always flowing parallel to the
deeper systems, rendering the profile models less certain for the shallow systems. Careful
consideration was given to this limitation while building the models and interpreting the results.
For example, a shallow water-table mound would flow in all directions, but within the profile
model, it is only allowed to flow parallel to the profile, thus forcing more water along the profile
than it would actually do. Attempts were made to identify such situations along the chosen
profiles. Where possible, the water-table profiles acquired from data were modified to mitigate
anticipated variations and/or considerations were made while interpreting the model predictions.
Consequently:


water balance estimates along the profiles were not done for the models, nor the
proportions volume of water from multiple sources seeping into the fens estimated as
they could be highly uncertain without independent validation data; and,



flow patterns presented in the profile models, therefore, are more qualitative in nature and
may be taken as „likely possibilities‟ along a profile rather than predicted flow.

179

Moreover, no measured data were available on porosity of glacial landforms or the
bedrocks which was critical to establish travel times of groundwater from the sources to the fen
sites. Using a range of porosity values between 0.1 and 0.3, travel times were estimated in the
models which only present a possible range. Without validation data, predicted travel times of
groundwater from a source to a site may be in a wide range of error.
Local or shallow flow systems connected to the fens were not adequately represented
along the profile models because the profiles were chosen along the major regional flow lines
while the local systems‟ flow paths were all around the fens. The profile models only captured
the local systems along the major regional flow paths.
Prescribed head boundaries driving the vertical flow in the profile models were obtained
from the delineation of local/shallow flow systems around the fen sites. These systems were
indirectly delineated by matching surface elevation data to a number of wetlands, lakes and
streams around the fens. This methodology may not be possible where surface water features are
not extensively available. Luckily for fen, they mostly exist in hummocky topographies
surrounded by lots of valleys and depressions filled with water.
Given all the shortcomings identified above, this study may be taken as a „first-cut‟
investigation which only employed the available data and revealed the probable source water
areas and possible delivery mechanisms of groundwater to the fen sites. The models from this
study can be used to guide multi-scale site specific investigations/data collections to validate
these models.

180

RECOMMENDATIONS FOR FURTHER STUDY
In this study we modeled only 19 of the 150 prairie fen sites in the state, out of which
only 8 were those with MS occurrence. It is recommended to model all of the remaining fen sites
and identify possible source water areas across the state. With all sites modeled, the hydrological
and other characteristics of sites with MS can be compared with other sites and statistical
inferences could be drawn to recommend sites for re-introduction of species and to protect the
existing sites.
The existing data from GIS databases can bring us thus far. Site specific data on fen
chemistry and outflow may be collected to validate source water locations and do the water
balance analyses to determine the share of groundwater contribution from multiple sources.
When all fen sites across the state are modeled, a complete inventory of fen sites can be
made with following information:


What is the groundwater contributing area for this fen?



Is there a strong regional source connected to the site?



How long does it take for water originating from the regional source to reach the fen?



How many other fens share this same regional source?



How is the regional water quality in this area?



Which fens seem to be hydrologically more sustainable?



Where should we monitor to be most effective in protecting a fen?



Where should we monitor to be most effective in protecting a large group of fens that
share the same regional source?



What is the land use pattern in the groundwater contributing areas? What potential
contamination sources exist in these areas?
181



How will the contamination spread if a spill occurs? How fast? Will it impact the fens?

For the fens which were historically occupied by MS, following may be investigated:


What has changed hydrologically at this site?



What common changes do fens historically occupied by MS share?

There are possibilities of discovering correlation of fen hydrology with its other
characteristics (e.g. EO Rank), but such correlations were not explored because the number of
modeled sites were too few to yield statistically significant conclusions. Better statistical
conclusions may be drawn from a full inventory of modeled sites, such as:


Is a high quality fen often associated with a very small surface water catchment?



What type of topography and geological conditions create the most stable groundwater
environment?



How does local and regional hydrology interact to control the ecological integrity of a fen
ecosystem?

CLOSING REMARKS
This study was aimed at investigating subsurface hydrology for prairie fens in southern
Michigan so that these unique ecosystems can be protected with more informed science. In the
process we investigated the hydrology at multiple scales and modeled 19 sites in detail. Existing
GIS datasets were used for groundwater modeling. The systematic, multi-scale study yielded
critical insights that support the need to consider linkages between local and regional flow
regimes.

182

The basin wide connections in sub-surface hydrology found in this study, on one hand,
pose new challenges and require a paradigm shift in the way the fen sites are currently being
managed, but on the other hand, these findings open up new opportunities for integrating and
coordinating conservation efforts between different agencies in a holistic and synergistic way.
Administratively more coordinated and scientifically more informed conservation activities will
benefit the regional resources on the whole.
With current advancements in databases and computing power, there lies an opportunity
that is unprecedented to protect and manage our ecological future and systematically enhance the
conservation effort of the Great Lakes States and beyond.

183

REFERENCES

184

REFERENCES
Acreman, M. C. (2004). "Impact assessment of wetlands: focus on hydrological and
hydrogeological issues." Phase 2 report, Environment Agency, Bristol (W6-091) and
Centre for Ecology and Hydrology, Wallingford (C01996).
Acreman, M. C., Fisher, J., Stratford, C. J., Mould, D. J., and Mountford, J. O. (2007).
"Hydrologic science and wetland restoration: some case studies from Europe." Hydrology
and Earth System Sciences, 11(1), 158-169.
Acreman, M. C., and Miller, F. (2004). "Hydrological impact assessment of wetlands."
International Symposium on Groundwater Sustainability (ISGWAS).
Acreman, M. C., and Miller, F. (2006). "Hydrological impact assessment of wetlands."
International Symposium on Groundwater Sustainability (ISGWAS), Centre for Ecology
and Hydrology,Wallingford, UK. Environment Agency, Solihull, UK, pp. 225-255.
Almendinger, J. E., and Leete, J. H. (1998a). "Peat characteristics and groundwater geochemistry
of calcareous fens in the Minnesota River Basin, U.S.A." Biogeochemistry, 43, 17-41.
Almendinger, J. E., and Leete, J. H. (1998b). "Regional and local hydrogeology of calcareous
fens in the Minnesota River Basin, USA." Wetlands, 18(2), 184-202.
Amon, J. P., Thompson, C. A., Carpenter, Q. J., and Miner, J. (2002). "Temperate zone fens of
the glaciated Midwestern USA." Wetlands, 22(2), 301–317.
Barnes, D. A., Bacon, D. H., and Kelley, S. R. (2009). "Geological sequestration of carbon
dioxide in the Cambrian Mount Simon Sandstone: Regional storage capacity, site
characterization, and large-scale injection feasibility, Michigan Basin." American
Association of Petroleum Geologists 16(3), 163-183.
Barton, B. J., and Bach, C. E. (2005). "Habitat use by the federally endangered Mitchell's satyr
butterfly (neonympha mitchellii mitchellii) in a Michigan prairie fen." The American
Midland Naturalist, 153, 41-51.
Batelaan, O., Smedt, F. D., and Triest, L. (2003). "Regional groundwater discharge:
phreatophyte mapping, groundwater modelling and impact analysis of land-use change."
Journal of Hydrology, 275, 86-186.
Bedford, B. L. (1999). "Cumulative effects on wetland landscapes: Linkages to wetland
restoration in the United States and Southern Canada." Wetlands, 19(4), 755-788.

185

Bedford, B. L., and Godwin, K. S. (2003). "Fens of the united states: distribution, characteristics,
and scientific connection versus legal isolation." Wetlands, 23(3), 608–629.
Bennett, G. D., and Giusti, E. V. (1971). "Coastal ground-water flow near Ponce, Puerto Rico."
United States Geological Survey, Professional Paper 750-D.
Bleuten, W., Borren, W., Glaser, P. H., Tsuchihara, T., Lapshina, E. D., Mäkilä, M., Siegel, D.,
Joosten, H., and J.Wassen, M. (2006). "Hydrological Processes,Nutrient Flows and
Patterns of Fens and Bogs." Ecological Studies, 190(Wetlands and Natural Resource
Management), pp. 183 - 204.
Bowles, M. L., Kelsey, P. D., and McBride, J. L. (2005). "Relationships among environmental
factors, vegetation zones, and species richness in a North American calcareous prairie
fen." Wetlands, 25(3), 685-696.
Bradbury, K. R., and Cobb, M. K. (2008). "Delineation of areas contributing groundwater to
springs and wetlands supporting the Hine's Emerald Dragonfly, Door County, Wisconsin.
Final report to the Wisconsin Coastal Management Program." Wisconsin Geological and
Natural History Survey, University of Wisconsin-Extension Open-File Report 2008-04.
Cobb, M. K., and Bradbury, K. R. (2008). "Delineation of areas contributing groundwater to
springs and wetlands supporting the Hine's Emerald Dragonfly, Door County,
Wisconsin." Open-File Report 2008-04, Wisconsin Geological and Natural History
Survey.
Couwenberg, J., Joosten, H., and Weber, C. A. (2002). "On the vegetation and development of
the raised bog of Augstumal in the Memel delta: Translation of the 1902 monograph by
Weber. IMCG/PPE „Friff & K‟, Tula, RU."
Cunge, J. A. (2003). "Of Data and Models." Journal of Hydroinformatics, 75-98.
Dall'O', M., Kluge, W., and Bartels, F. (2001). "FEUWAnet: a multi-box water level and lateral
exchange model for riparian wetlands." Journal of Hydrology, 250, pp. 40-62.
Dau, J. H. C. (1823). Neues Handbuch über den Torf, dessen Natur, Entstehung und
Wiedererzeugung: Nutzen im Allgemeinen und für den Staat [New Handbook on the turf,
the nature, formation and reproduction, Benefits in general and for the State]. J.E.
Hinrischssche Buchhandlung, Leipzig, DE
Dekker, S. C., Barendregt, A., Bootsma, M. C., and Schot, P. P. (2005). "Modeling hydrological
management for the restoration of acidified floating fens." Hydrological Processes
19(20), 3973-3984.

186

Du Rietz, G. E. (1949). "Huvudenheter och huvudgränser i svensk myrvegetation [Main units
and main limits in Swedish mire vegetation]." Sven. Bot. Tidskr, 43, 274-309.
Du Rietz, G. E. (1954). "Die Mineralbodenwasserzeigergrenze als Grundlage einer natürlichen
Zweigliederung der nord- und mitteleuropäischen Moore [The mineral soil water limit
pointer as the basis of a natural bipartition of the north and central Moore]." Vegetatio,
5(6), 571-585.
Environment Canada. (2009). "Integrated Watershed Management." Environment Canada.
EU. (2000). "Directive 2000/60/EC of the European Parliament and of the Council of 23 October
2000 establishing a framework for Community action in the field of water policy "
Official Journal of the European Union L 327, 0001 - 0073.
EU. (2006). "Directive 2006/118/EC of the European Parliament and the Council of 12
December 2006 on the protection of groundwater against pollution and deterioration."
Official Journal of the European Union, L 372, 19 - 31.
Freeze, R. A. (1969). "Regional groundwarter flow - Old Wives Lake Drainage Basin,
Saskatchewan." Canadian Dept of Energy, Mines and Resources, Inland Waters Branch,
Sci. Ser. no. 5, 245 p.
Garcia, L. A., and Shigidi, A. (2006). "Using neural networks for parameter estimation in ground
water." Journal of Hydrology, 318, 215-231.
Gillham, R. A., and Farvolden, R. N. (1974). "Sensitivity analysis of input parameters in
numerical modeling of steady state regional ground-water flow." Water Resources
Research, 10(3), 529-538.
Gilvear, D. J., Andrews, R., Tellam, J. H., Lloyd, J. W., and Lerner, D. N. (1993).
"Quantification of the water balance and hydrogeological processes in the vicinity of a
small groundwater-fed wetland, East Anglia, UK." Journal of Hydrology, 144, 311-334.
GLPF. (1998). "Report on the Meeting of Water Resources Roundtable." Great Lakes Protection
Fund, Chicago, Illinois.
Godwin, K. S., Shallenberger, J. P., Leopold, D. J., and Bedford, B. L. (2002). "Linking
landscape properties to local hydrogeologic gradients and plant species occurrence in
minerotrophic fens of New York State, USA: A Hydrogeologic Setting (HGS)
Framework." Wetlands, 22(4), 722-737.
Gómez-Hernández, J. (2006). "Complexity." Ground Water, 44(6), 782-785.

187

Gorham, E. (1957). "The Development of Peat Lands." The Quarterly Review of Biology, 32(2),
145-166.
Grapes, T. R., Bradley, C., and Petts, G. E. (2006). "Hydrodynamics of floodplain wetlands in a
chalk catchment: The River Lambourn, UK." Journal of Hydrology 320, pp. 324-341.
Grootjans, A. P., Adema, E. B., Bleuten, W., Joosten, H., Madaras, M., and Janáková, M. (2006).
"Hydrological landscape settings of base-rich fen mires and fen meadows: an overview."
Applied Vegetation Science, 9, 175-184.
GWIM. (2006). "Groundwater Inventory and Map Project", State of Michigan Public Act 148.
Haitjema, H. M., and Mitchell-Bruker, S. (2005). "Are water tables a subdued replica of the
topography?" Ground Water, 43(6), 781 - 786.
Halsey, L., Vitt, D., and Zoltai, S. (1997). "Climatic and physiographic controls on wetland type
and distribution in Manitoba, Canada." Wetlands, 17(2), 243-262
Harvey, F. E., Ayers, J. F., and DavidC.Gosselin. (2007a). "Ground Water Dependence of
Endangered Ecosystems: Nebraska‟s Eastern Saline Wetlands." Ground Water, 45(6),
736-752.
Harvey, F. E., Swinehart, J. B., and Kurtz, T. M. (2007b). "Ground Water Sustenance of
Nebraska‟s Unique Sand Hills Peatland Fen Ecosystems." Ground Water, 45(2), 218234.
Hill, M. C. (2006). "The Practical Use of Simplicity in Developing Ground Water Models."
Ground Water, 44(6), 775–781.
Hyde, D. A., Rabe, M. L., Cuthrell, D. L., and Kost, M. A. (2001). "Surveys for the recovery of
Mitchell's satyr butterfly (Neonympha m. mitchellii) in Michigan." U.S. Fish and
Wildlife Service, Region 3 Office, Ft. Snelling, Minnesota.
IJC. (2009). "The International Watersheds Initiative: Implementing a New Paradigm for
Transboundary Basins." International Joint Commission, Canada and United States.
Koerselman, W. (1989). "Groundwater and surface water hydrology of a small groundwater-fed
fen." Wetlands Ecology and Management, 1(1), 31-43.
Kost, M. A., and Hyde, D. A. (2009). Exploring the Prairie Fen Wetlands of Michigan, ANR
Communications Michigan State University, East Lansing.

188

Loon, A. H. v., Schot, P. P., Griffioen, J., Bierkens, M. F. P., and Wassen, M. J. (2009). "Palaeohydrological reconstruction of a managed fen area in The Netherlands." Journal of
Hydrology, 378, 205-217.
Lowry, C. S., and Anderson, M. P. (2006). "An Assessment of Aquifer Storage Recovery Using
Ground Water Flow Models." Ground Water, 44(5), 661-667.
MDNR. (2003). "Michigan Forest Legacy Program Assessment of Need." A Report from
Michigan
Department
of
Natural
Resources,
http://www.michigandnr.com/publications/pdfs/ForestsLandWater/ForestLegacy/ForestL
egacyAON_Complete.pdf.
Meyboom, P. (1967). "Mass-transfer studies to determine the groundwater regime of permanent
lakes in hummocky moraine of Western Canada." Journal of Hydrology, 5, 117-142.
MNFI. (2010). "GIS Data for Michigan's Prairie Fens and 2010 Survey for Mitchell's satyr
Butterfly sites." Michigan Natrual Features Inventory.
NatureServe. (2002). "Element Occurrence Data Standard." Network of Natural Heritage
Programs and Conservation Data Centers.
NED USGS. (2006). "National Elevation Dataset." US Department of Interior: United States
Geological Survey, http://ned.usgs.gov/Ned/about.asp.
NHD USGS. (2010). "USGS: National Hydrography Dataset." US Department of Interior:
United States Geological Survey, http://nhd.usgs.gov/index.html.
NSW CMA. (2004). "CMA Activities and Achievements." New South Wales Catchment
Management Authorities, Sydney, Australia.
Oztan, M. (2010). "GIS Enabled Modeling of Michigan's Groundwater Systems," PhD
Dissertation, Michigan State University, East Lansing.
PA 148. (2003). "State of Michigan Public Act 148 Groundwater Inventory and Map Project."
Michigan Department of Environmental Quality.
Price, J. S., Branfireun, B. A., Waddington, J. M., and Devito, K. J. (2005). "Advances in
Canadian wetland hydrology, 1999–2003." Hydrological Processes, 19, 201-214.
Price, J. S., and Waddington, J. M. (2000). "Advances in Canadian wetland hydrology and
biogeochemistry." Hydrological Processes, 14, 468 0478

189

Public Act 148. (2003). "State of Michigan Public Act 148 Groundwater Inventory and Map
Project." Michigan Department of Environmental Quality.
Siegel, D. I. (1988). "Evaluating Cumulative Effects of Disturbance on the Hydrologic Function
of Bogs, Fens, and Mires." Environmental Management, 12(5), 621-626.
Siegel, D. I., and Glaser, P. (2006). "The Hydrology of Peatlands." Ecological Studies, Vol.188
(Boreal Peatland Ecosystems ), pp. 289 - 311.
Siegel, D. I., and Glaser, P. H. (1987). "Groundwater flow in a bog-fen complex, Lost River
Peatland, Nothern Minnesota." Journal of Ecology, 75, 743-754.
Silberstein, R. P. (2006). "Hydrological models are so good, do we still need data?"
Environmental Modeling & Software, 1340-1352.
Simard, A. (2007). "Predicting groundwater flow and transport using Michigan's statewide
Wellogic database," PhD Dissertation, Michigan State University, East Lansing.
SNH. (2010). The Fen Management Handbook, Scottish Natural Heritage, Perth.
SOGW. (2009). "A National Framework for Ground-Water Monitoring in the United States."
Advisory Committee on Water Information, Subcommittee on Groundwater Monitoring.
Spieles, J. B., Comer, P. J., Albert, D. A., and Kost, M. A. (1999). "Natural community abstract
for prairie fen." Michigan Natural Features Inventory, East Lansing Michigan.
State of Michigan. (2006). "Michigan State Legislature Public Act 451 Natural Resources and
Environmental Protection Act 451 of 1994." Parts 327 and 328 as amended
State of Michigan. (2009). "Michigan State Legislature Public Act 451 Natural Resources and
Environmental Protection Act 451 of 1994." Section 324.32701 Michigan, USA.
Stearns, M. D. (1993). "The petrology of the Marshall Formation of Michigan." Journal of
Sedimentary Petrology, 3(3), 99-112.
Stein, E. D., Mattson, M., Fetscher, A. E., and Halama, K. J. (2004). "Influence on geologic
setting on slope wetland hydrodynamics." Wetlands, 24(2), 244-260.
Toth, J. (1972). "Properties and Manifestations of Regional Groundwater Movement." 24th
International Geological Congress Proceedings, Section 11, pp. 153 - 163.

190

USFWS. (1998). "Mitchell's Satyr Butterfly Neonympha mitchellii mitchellii Recovery Plan."
US Dept of Interior Fish and Wildlife Service.
Variano, E. A., Ho, D. T., Engel, V. C., Schmieder, P. J., and Reid, M. C. (2009). "Flow and
mixing dynamics in a patterned wetland: Kilometer-scale tracer releases in the
Everglades." Water Resources Research, 45.
Vecchioli, J., Bennett, G. D., Pearson, F. J., and Cerrillo, L. A. (1974). "Geohydrology of the
artificial-recharge site at Bay Park, Long Island, New York." United States Geological
Survey, Professional Paper 751-C.
Wassen, M. J., Barendregt, A., Schot, P. P., and Beltman, B. (1990). "Dependency of local
mesotrophic fens on a regional groundwater system in a poldered river plain in the
Netherlands." Landscape Ecology, 5(1), 21-38.
Winter, T. C. (1978). "Numerical Simulation Analysis of the Interaction of Lakes and Ground
Water." United State Geological Survey, Geological Survey Professional Paper 1001.
Winter, T. C. (1998). "A conceptual framework for assessment of cumulative impacts on the
hydrology of non-tidal wetlands " Environmental Management, 12, 605-620.

191