NM_Thesis

ELECTRICAL RESISTIVITY AND GROUND-PENETRATING
RADAR AS TOOLS TO CHARACTERIZE GROUNDWATER-
SURFACE WATER INTERACTION AT MIRROR LAKE, NH
A Thesis
Submitted to the
Temple University College Graduate Board
In partial Fulfillment
Of the Requirements for the Degree
Master of Science
By
Natasha Mitchell
August, 2008
_______________________________
Dr. Jonathan E. Nyquist, Advisor
_______________________________
Dr. Laura Toran
_______________________________
Dr. Allison Tumarkin-Deratzian

2
TABLE OF CONTENTS
Abstract............................................................................................................................... 4
Acknowledgments............................................................................................................... 5
Introduction and background.............................................................................................. 7
Study Site.......................................................................................................................... 12
2.1 Study Site Overview ............................................................................................... 12
2.2 Study Site Hydrology and Geology........................................................................ 12
2.3 Road salt Contamination at Mirror Lake ................................................................ 17
Methods............................................................................................................................. 19
3.1 Electrical Resistivity Survey Overview.................................................................. 19
3.2 Towed-Cable Surveys at Mirror Lake .................................................................... 25
3.3 Stationary-Cable (Lake-Bottom) Survey................................................................ 28
3.4 Data Inversion......................................................................................................... 30
3.5 Inversion Error Analysis......................................................................................... 32
3.6 Resistivity Test Box................................................................................................ 35
3.7 Sediment Sampling................................................................................................. 37
3.8 Ground Penetrating Radar....................................................................................... 39
3.9 Seepage Meters ....................................................................................................... 46
3.10 Bathymetry............................................................Error! Bookmark not defined.
3.11 Data Coordinates................................................................................................... 48
Results and Discussion ..................................................................................................... 50
4.1 Introduction............................................................................................................. 50
4.2 Resistivity Survey Results Overview..................................................................... 50
4.3 Southwest Shore Resistivity And Seepage Results ................................................ 62
4.3.1 Resistivity Values 1500 to 3000 Ω–m and Seepage Rates .................................. 62
4.3.2 Low (200 Ω-m) Resistivity and Low Seepage Rates .......................................... 72
4.4 Eastern Shore Seepage and Resistivity................................................................... 72
4.5 Northeast Inlet: Resistivity and Road salt Contamination...................................... 73
4.6 Ground Penetrating Radar Surveys......................................................................... 76
Conclusions....................................................................................................................... 86
5.1 Electrical Resistivity............................................................................................... 86
5.3 Electrical Resistivity Field Techniques .................................................................. 86
5.3 Ground Penetrating Radar....................................................................................... 88

3
5.4 Ground Penetrating Radar Field Techniques.......................................................... 88
5.5 Using Electrical Resistivity and Radar In Tandem................................................. 89
5.6 Position Data........................................................................................................... 89
5.7 Suggestions for Further Study ................................................................................ 89
5.8 Implications for understanding groundwater-lake interaction................................ 90
Appendix A....................................................................................................................... 96

4
ABSTRACT
The results of electrical resistivity surveys using both surface-towed and stationary
(lake-bottom) cables at Mirror Lake, NH suggested that resistivity surveying can be useful for
characterizing geologic heterogeneities that control groundwater-surface water interaction, as
well as for imaging road salt contamination. The rapidly acquired towed-cable surveys can
guide placement of higher-resolution, more time-consuming stationary-cable surveys. The
use of stationary-cable resistivity surveys in the very near-shore environments (<2 m from
shore), which are generally inaccessible with a towed-cable survey, can guide seepage meter
placement. In-situ measurements of seepage coincident with resistivity surveys suggested a
relationship between resistivity values and seepage rates. Specifically, we observed that
seepage rates were low (averaging -20 cm/day) at Mirror Lake where resistivity values were
greater than or equal 3000 Ω-m and where they were less than or equal to 200 Ω- m. Low
(200 Ω-m) resistivity values were indicative of organic matter deposits. High (3000 Ω-m)
resistivity values were indicative of low porosity, poorly sorted till. Intermediate (~1500 Ω–
m) resistivity values were observed in the regions where seepage rates were highest
(averaging -80 cm/day). Core, modeling, and slug test data suggest that these intermediate
resistivity values reflect more well-sorted, higher-porosity drift. Resistivity surveys of the
suspected region of salt contamination revealed a plume-shaped feature of low resistivity.
Low resistivity and higher chloride content were confirmed by laboratory analysis of pore
fluid. The results of ground penetrating radar (GPR) surveys suggested that this technique can
also identify geologic features which relate to seepage.GPR surveys identified the bounds of
a blanket of organic matter which covers the bottom of Mirror Lake; previous work suggested
that this organic matter controls seepage there. Radar also confirmed the results of previous
work at Mirror Lake about the distribution of cobbles and boulders there.

5
ACKNOWLEDGMENTS
This work was supported by the National Science Foundation Hydrologic Sciences
Program under Award No. 0609827.
Thanks to the Hubbard Brook Research Foundation for providing logistical support.
Thank you, first, to Dr. Jonathan Nyquist. It was a leap of faith (no pun intended...well,
OK, kind of intended) to take on a religion major. Thank you for all you have taught me.
Thank you, too, for making me laugh.
To Dr. Laura Toran: You have a keen scientific and editing eye. This thesis would have
been a different animal altogether without your input. Thank you for all of your help.
Dr. Allison Tumarkin-Deratzian. Thank you for letting me be your first thesis advisee. It
was an honor and a pleasure.
To Matthew Heaney – Thank you, for everything. In your broad footprints I have striven to
walk. You lent your time, your intelligence, your wisdom, and, most of all, your company.
If I emerged from Temple University with nothing besides your friendship I would be just
fine.
To Roger Cutitta- You helped carry me through. Your warmth and your presence meant the
world to me.
To Billy Zacharatos – I don’t think there is much I can say without blushing, so I won’t try.
To Jim Mikochik – my partner in arms. Thank you for your kind spirit.
To Paul Freyer, Jason Mintz, Gary Stinchcomb and Hallie Meighan: Thank you for getting
me started, and for the many episodes of our friendship.
To the other members of the class of 2008 – Jesse Thornburg, Holly Sobocinksi, Christian
Obassi and Paul Oliver: I think we’ve done well by one another. Thanks for being raft
mates.
To the undergraduates, old and young: Steve Peterson, Eva Gladish, Star George, Billy
Lukens, Kevin McGinn, Collin Miller, and, most recently, Sam Galenty. Your presence in
daily life became as important as the air I breathed. Thanks just for being around. I hope
our friendship far outlasts the time I’ve spent working on this thesis.
To Shelah Cox – for being brave and all that you do. We would have sunk without you.
To Mike Whitson, wherever you are – your lecture, that night, in the sand and under the
stars, was what brought me to geology. I never could have seen where that journey would
bring me, I only knew I wanted to take it. Thanks for sending me along.
To my parents – without you, I wouldn’t be here.
To Rich Nourrie – Of all the rafts I have held to, you are one of the warmest. As you wrote
to me once, we are blessed by the connection we have.
Jane, John Paul, Kate and Peter MacDuffie – I love you very much.

6
DEDICATION
To Dr. Max Tegmark:
You helped me believe in sparkles.

7
CHAPTER 1
INTRODUCTION AND BACKGROUND
The goal of this study was to develop the use of electrical resistivity and ground-
penetrating radar for the characterization of groundwater-surface water interaction.
Characterizing and quantifying the interaction of groundwater with surface water across a
watershed is a complex task; seepage rates (groundwater discharge to surface flow systems or
surface flow recharge of the groundwater system), which are controlled by geologic
heterogeneity, can vary by an order of magnitude or more over a distance of 1-2 meters.
Characterizing seepage is therefore difficult in any setting. Finding, characterizing, and
quantifying underwater seepage is all the more difficult because the groundwater-surface
water interfaces lies beneath a column of water. These zones of exchange, however, represent
important features of hydrologic systems. The temperature, solute content, and water-flow
conditions of seepage zones make them crucially important to many aquatic species (Kalbus
et al., 2006; Hayashi and Rosenberry, 2002; Brunke and Gonser, 1997; Asbury, 1990).
Seepage can also be a significant source or sink of water to or from lakes and streams
(Rosenberry, 2005; Lee, 1977; McBride and Pfannkuch, 1975). Watershed management is
therefore partially contingent upon an understanding of the interaction of groundwater with
surface water. The ability to model the effects of and treat a groundwater contaminant plume,
for example, is dependent upon understanding how that plume is likely to interact with
surface water (Shaban et al., 2005; Sophocleous, 2002; Ball et al., 2006; Cherkauer, 1991).
Predicting the effects of pumping on aquifer recharge, flow of local surface water bodies, or
salt-water intrusion is also contingent upon knowing how much water is supplied or lost to
those bodies via groundwater-surface water exchange (Taylor, 1982; Taylor and Cherkauer,
1984; Cherkauer and Carlson, 1997; Snyder and Wightman, 2002; Shaban et al., 2005). Tools

8
that can characterize and quantify the interaction between groundwater and surface water are
therefore essential to watershed management strategies.
Researchers’ efforts to quantify seepage have been, historically, limited to point
measurement. Formal efforts to characterize seepage began in the 1940’s with seepage
meters, cylindrical drums that could be pressed into sediment to effectively seal the zone
beneath the drum from the external environment. Seepage was calculated by measuring the
amount and rate at which water was discharged or recharged within that sealed zone,
researchers measured seepage (Lee, 1977; Kalbus et al., 2006). Methods were also developed
to calculate seepage based upon proxies such as hydraulic and temperature gradients.
Hydraulic gradients, calculated via head measurements from wells within the region of
interest, can suggest seepage locations, as can temperature gradients (potentially caused by
the intrusion of groundwater which is of different temperature than the surface water to which
it discharges). Hydraulic conductivity can also indicate zones of seepage; where lake-bottom
sediment is particularly hydraulically conductive, as calculated from slug or pumping tests,
high rates of seepage may exist. Flow analysis via dye or isotope tracing can also point to
seepage (summarized by Kalbus et al., 2006). Though point techniques can measure seepage
directly, they are limited in that they are labor and time intensive. Moreover, estimates of
seepage will be over- or undervalued if spatial variation is not measured.
Geophysical surveying techniques, with which the researcher can rapidly characterize
a large area, yield information about geologic heterogeneities. Because geologic
heterogeneities can control seepage rates, geophysical characterization techniques offer a
means of guiding point measurement of seepage. Electrical resistivity profiling, which
responds to variations in porosity, saturation, pore fluid conductivity, and lithology, was one
of the first geophysical methods tested (Taylor, 1982; Taylor and Cherkauer, 1984;
Cherkauer, 1991; Cherkauer and Carlson, 1997). Taylor and Cherkauer (1984) and Cherkauer

9
(1991) built upon earlier work by Archie (1942), using electrical resistivity to identify
geologic heterogeneity in the sediment beneath Lake Michigan, and then drawing conclusions
about likely seepage rates based upon that lithologic information. In particular, Taylor and
Cherkauer (1984) showed that seepage rates at Lake Michigan varied with clay content.
Because clay is electrically conductive, regions where sediment resistivity was low were
indicative of regions with high clay content and therefore indicative of regions where seepage
rates were likely to be low. Taylor and Cherkauer confirmed this relationship via seepage
meter measurements, piezometer measurements, and by laboratory analysis of sediment
permeability. The authors concluded that further investigation ought be pursued at a small,
well-characterized lake to deepen understanding of the relationship between electrical
resistivity profiles, lithology, and seepage. Though several related surveys have been
conducted in the 20+ years since (e.g., Cherkauer, 1991; Panthulu, et al., 2001; Snyder and
Wightman, 2002; Manheim et al., 2004; Day-Lewis et al., 2005; Ball et al., 2006), no study
explicitly addressing understanding of the relationship between resistivity, geologic
heterogeneity and seepage rates had been published as of the spring of 2006.
The technology available for resistivity surveys has evolved markedly from what was
available to Taylor and Cherkauer in the 1980’s. Taylor and Cherkauer were limited to 1-
dimensional resistivity profiles; today, 2-D resistivity profiles are standard. GPS coordinate
positioning did not yet exist at the time of the Taylor and Cherkauer study; it was therefore
difficult to link geophysical observations with ground truth data. Finally, the computing
power available to control the collection, processing and inversion of apparent resistivity data
has grown substantially more powerful since the 1980s.
The work presented in this thesis represents the second stage of a project meant to
explore the relationship between resistivity, geologic heterogeneity and seepage in a means
similar to that suggested by Taylor and Cherkauer in 1984, with the more powerful tools

10
available 20 years later. The results of the first stage of that project were presented by Heaney
et al. (2006) and Heaney (2007). That stage of the project focused on Lake Lacawac, a small,
glacial lake in northeast Pennsylvania for which there were geologic and hydrologic data
available. Heaney et al. (2006) used a resistivity survey system that allowed them to gather 2-
D resistivity data, a GPS device to tie resistivity data directly to location, and computers and
software which allowed them to rapidly process these data. That technology represented
significant advance from the 1-D data gathered by Taylor and Cherkauer (1984), the strip
charts used to record those data, and the limited ability to tie resistivity data to location. In
short, Heaney et al. (2006) were not only able to follow up on the questions raised by the
work of Cherkauer and Taylor (1984) at exactly the sort of site the authors had suggested, but
they were able to do so with equipment which offered far greater resolution.
There were two significant outcomes of the work by Heaney et al. (2006) at Lake
Lacawac. First, they refined the field technique of electrical resistivity surveying for the
purpose of characterizing seepage: the equipment they employed for the project was new, and
its application to the characterization of lake seepage untested. Developing field techniques
for use of the equipment for the characterization of lake seepage represented an important
component of their work. Second, by sampling and analyzing sediment from areas where they
had conducted electrical resistivity surveys, they confirmed the efficacy of the technique in
identifying geologic heterogeneities. They were able to identify lithologic transitions and
likely seepage flow paths on the scale of 1-2 meters. They were unable, however, to
definitively link geologic heterogeneity with seepage rates because measured seepage rates at
Lake Lacawac were extremely low; maximum measured seepage flux was less than 0.19 cm
per day, a quantity within the bounds of measurement error. For a more definitive correlation
of the relationship between geophysical signature and seepage, a study site with higher
seepage rates was needed.

11
The work presented in this thesis represents such a continuation of the work begun by
Heaney et al. (2006). In this stage of the project, ground-penetrating radar (GPR) is added as
an additional survey technique. Under ideal conditions, GPR surveys yield high-resolution
images of the geometry of the subsurface environment (summarized by Neal, 2004). Using
GPR, the electrical resistivity field techniques developed by Heaney et al. (2006) and the
same resistivity survey equipment, our research group conducted geophysical surveys and
seepage measurements at Mirror Lake, New Hampshire.

12
CHAPTER 2
STUDY SITE
2.1 Study Site Overview
The site of the study was Mirror Lake, a small (15 hectare), glacially-formed lake in
the White Mountains of central New Hampshire (Figure 2-1). The lake has been a National
Science Foundation-designated Long Term Ecological Research site (LTER) since the
1980’s, and is part of the Hubbard Brook Experimental Forest of the United States
Department of Agriculture (Likens, 1984). Of particular importance to this project was that
seepage rates up to 250 ml per minute at discrete points were reported (Rosenberry, 2005),
implying that the site would offer measurable seepage rates against which to compare the
results of geophysical surveys. The availability of other geologic and hydrologic data (e.g.,
Winter, 1984; Paillet, 1988; Winter, 1989; Harte, 1997; Tiedeman et al., 1997; Johnson,
1999; Ellefsen et al., 2002) also made Mirror Lake appealing as a site for study; it meant we
would have multiple measures of ground truth against which to compare the results of
geophysical surveys.
2.2 Study Site Hydrology and Geology
Mirror Lake has three inlet streams and one outlet (Figure 2-1). It is 11 m deep at its
maximum. The sides of the lake are generally steep along the northern, eastern and southeast
shores, with water depths often plunging to 6 or 7 meters within 5 or 6 meters from shore.
Bedrock beneath Mirror Lake is fractured quartz monzonite, schist, and a variety of igneous
intrusives (Winter, 1984; Johnson, 1999). The rock is Early Devonian in age and comprises
the igneous and metamorphic Littleton and Kinsman Formations. Schist and slate are the
primary constituents of the Littleton; microcline, plagioclase and quartz, in approximately
equal parts, are the main constituents of the Kinsman monzonite (Winter, 1984). Bedrock is
immediately overlain by a layer of poorly-sorted glacial drift, which ranges in thickness from

13
nearly non-existent at its minimum (the center of the lake) to over 50 meters at its maximum
Figure 2-1

14
(lake’s edge). The drift is clay-poor and ranges in composition from silt to silty sand, with
cobbles and boulders distributed throughout (Winter, 1984; Winter, 2000). A thick (3 m+)
layer of organic sediment (“gyttja”) blankets all depths greater than approximately 7 m.
Rosenberry et al. (1999) calculate that groundwater seeps into Mirror Lake along
approximately 80% of the shoreline, and that that seepage provides 11% of annual influx to
the lake (Figure 2-2). Lake water seeps out of Mirror Lake into the groundwater system
along the southern shore and accounts for approximately 40 to 47% of lake out-flux
(Rosenberry et al., 1999). Rosenberry and Winter (1993) comment that seepage rates at
Mirror Lake are in fact lower than might be expected if the lake lay directly atop bedrock
rather than atop glacial drift. They attribute the lower-than-expected seepage rates to the low
hydraulic conductivity of the drift (range of K ≈ 10-7
to10-5
m/sec), and to the presence of the
thick layer of gyttja, which has an even lower hydraulic conductivity (Rosenberry and
Winter, 1999).
The highest seepage rates at Mirror Lake were generally observed within 1-2 m of
shore, with seepage rates decreasing rapidly with distance from shoreline (Asbury, 1990;
Rosenberry, 2005). This observation is in keeping with numerical models of seepage, which
suggest an exponential decline in-seepage with distance from a lake’s shore (McBride and
Pfannkuch, 1975; Winter, 1976; Genereux and Badopadhay, 2001; Mikochik, 2008).
Seepage rates are highest along the southwest shore of Mirror Lake: rates in excess of 200 cm
per day have been reported within a zone approximately 40 meters long (Asbury, 1990;
Rosenberry, 2005). On-shore core samples from regions ~10 – 100 m from the locations of
high seepage rates record the presence of relatively well-sorted, and therefore relatively
porous, sand and gravel within the upper 50 meters of sediment (Wilson, 1991; Tiedeman et

16
al., 1997; Scheutz 2002; Harte, 1997). Because sediment at Mirror Lake is clay poor (Harte,
1997), it is reasonable to consider porosity as a proxy for hydraulic conductivity: porous (i.e.
well-sorted) material will conduct water well, and less porous material (poorly-sorted) will
conduct it poorly. Where sediment at Mirror Lake is poorly sorted, as it is around much of the
lake, seepage rates will be low; where it is better sorted, seepage rates will be higher. The
results of slug tests of on-shore wells at Mirror Lake are consistent with this hypothesis; the
hydraulic conductivity of well-sorted sand and gravel deposits along the southwest shore of
Mirror Lake is approximately 10 to 100 times greater (K ~ 10-4
m/sec) than that measured in
more poorly sorted sediment elsewhere around the lake (Wilson, 1991; Harte, 1997;
Tiedeman et al., 1997; Scheutz, 2002). Research suggests that this sand and gravel deposit is
responsible for the very high out-seepage rates observed along the southwest shore of Mirror
Lake, and for the fact that out-seepage represents such a high portion (~47%) of all out-flux
for the lake (Tiedeman et al., 1997; Rosenberry et al., 1999; Scheutz, 2002).
The presence of comparatively well-sorted sediment along the southwest shore is
geologically plausible. Several lines of evidence suggest that the southwest shore of Mirror
Lake represents an outwash lens for the receding glacier that formed Mirror Lake. First,
while the topography around much of the lake is steep, the ground adjacent to the southern
end of the lake is relatively flat (Figure 2-1). Such comparatively flat terrain is consistent
with a zone of high deposition, as would be expected with an outwash lens. Second, as
described above, cores along the southwest shore show deposits of sand and gravel up to 50
m thick (Harte, 1997; Tiedeman, 1997). The presence of discrete deposits of sand and gravel
is indicative of sorting associated with fluvial processes rather than of simple ablation. A
similar model of outwash deposition has been used to explain variations in hydraulic
conductivity and seepage rates in Trout Lake, Wisconsin (Krabbenhoeft and Anderson,
1986).

17
2.3 Road salt Contamination at Mirror Lake
The eastern portion of the Mirror Lake drainage basin includes Interstate Highway 93
(Figure 2-3). Since construction of the highway in the 1970’s, the concentrations of sodium
and chloride in Mirror Lake have risen steadily. Researchers have attributed this rise to the
flow of water contaminated with road salt from Highway 93 into Mirror Lake (Rosenberry et
al., 1999). The majority of road salt contaminated water entered Mirror Lake through the
Northeast Inlet stream, which directly abutted the highway. A diversion berm was
constructed to capture contaminated runoff in an artificial wetland and thereby prevent it
from entering the Northeast Inlet. Though this berm appears to have slowed overland flow of
road salt contaminated water to the Northeast Inlet, the berm leaks in several places and so
road salt continues to enter Mirror Lake via the Northeast Inlet stream. Measurement of
chloride concentrations in bedrock wells between Highway 93 and Mirror Lake indicate that
chloride has entered the groundwater system as well, though not at all measurement points.
Piezometer measurements of hydraulic gradient suggest that groundwater seeps into the
Northeast Inlet stream and within the cove to which the Northeast Inlet discharges. Seepage
therefore also represents a likely mode of delivery for road salt contaminated water to Mirror
Lake (Rosenberry et al., 1999). Because dissolved salts like chloride decrease water
resistivity, it was possible that resistivity surveys would respond to road salt contamination of
pore-fluid at Mirror Lake.

18
Figure 2-3 – Road Salt Contamination

19
CHAPTER 3
METHODS
3.1 Electrical Resistivity Survey Overview
Geologic heterogeneity is one factor that controls seepage. Because geologic
heterogeneity can cause changes in any of the factors which control electrical resistivity -
porosity, pore fluid conductivity, saturation and mineralogy – electrical resistivity profiling
offers the potential to image these geologic heterogeneities. To profile the resistivity structure
of the subsurface environment, we injected current into the subsurface and measured resultant
electric potential across dipoles at known distances from the point of injection. Drop in
electric potential at measuring electrodes was a function of the resistivity of the material
through which current passed and of the distance between current and measuring electrodes.
Electrical resistivity of geologic materials is a function of porosity, pore fluid
conductivity, volume fraction of pores filled with water, and the electrical properties of the
minerals that comprise the material and that cement individual grains (Archie, 1942). The
relationship between these factors and electrical resistivity is quantified by Archie’s law
(3.1):
= a -m
s-n
w 3.1
where is the bulk (overall) resistivity of a substrate, w is pore fluid resistivity, is the
porosity, s is saturation and a, m and n are constants, characteristic of the electric properties
of constituent minerals or of the cement which binds grains. Variation of any of these factors
leads to variation in bulk resistivity.
Voltage potential drop is proportional to the length of the current path and inversely
proportional to the cross-sectional area through which current flows (Figure 3-1). This
constant of proportionality defines a substrate’s resistivity and it is typically measured in

20
Ohm-meters (Ω-m).
Figure 3-1. Electrical Resistivity. The resistivity of a geologic
material is defined by the ratio of the drop in Voltage per unit
length divided by the cross-sectional area. Adapted from Reynolds,
1997.
In a homogeneous half-space of known resistivity ρ, potential drop from an injection of
current at a point source on the surface varies as a function of distance from the source of the
current:
3.2
where r is the distance from current source, and voltage drop is negative overall because
potential decreases with distance from the current source (Reynolds, 1997). Greater distance
between current and measuring electrodes (r) implies that the measuring electrodes will sense
current that has penetrated more deeply in the earth. Where the earth is not homogeneous,
measured resistivity will reflect the resistivities of all media through which the current has
passed in its path through a given section of earth. Such a measurement is known as apparent
resistivity (ρa). If multiple measurements of apparent resistivity are made within a study area,

21
and different electrode spacings used for each measurement, inferences about the distribution
of resistive materials can be made. For instance, because wider electrode spacings allow
greater depth of penetration, a resistivity survey wherein a wider electrode spacing yielded a
lower apparent resistivity value than a closer spacing implies the existence of a low resistivity
body at depth. Conversely, a resistivity survey wherein a wider spacing yielded a higher
apparent resistivity implies the existence of a high resistivity body at depth.
By comparing and weaving together apparent resistivity measurements from multiple
locations and made using multiple electrode spacings, a 2-D image of the subsurface
environment can be created. Modern resistivity survey equipment facilitates and automates
this process; most, including the SuperSting R8 Marine Resistivity system used in this project
(Figure 3-2), have computers which fire electrodes in pre-determined patterns, and multiple
channels for measurement of potential drop at multiple locations for each injection of current.
Also available is modeling software which solves for the subsurface structure implied by a
given set of apparent resistivity measurements.
The SuperSting® resistivity survey system, which uses all the features described
above, uses watertight electrode cables for current injection and potential measurement.
Electrode cables with variable numbers of and spacing between electrodes may be freely
interchanged. The Sting is equipped with a variety of pre-set firing patterns, each optimized
for a particular survey condition; it also gives the user the option of creating a custom pattern.
Electrode number and geometry, along with desired firing/measuring pattern is input to the
SuperSting prior to a survey.
We conducted two types of surveys in this project: towed-cable surveys and stationary
(lake-bottom) cable surveys. The cable used in the towed survey has a total of 11 electrodes.
The first two of these eleven electrodes were dedicated current electrodes and injected current
at regular intervals as the cable was towed around the lake. The other 9 electrodes worked in

22
pairs to measure electric potential (Figure 3-3A). The stationary-cable has 28 electrodes and
was left in place while its 28 electrodes cycled through a pre-set pattern of firing and
Figure 3-2

23
measuring (Figure 3-3B). This variable firing pattern allows the stationary-cable survey to
gather resistivity measurements from multiple different angles over a given area without
physically moving the cable.
Each survey type has advantages. With a towed cable the surveyor can rapidly
characterize a large distance; the boat towing the cable generally travels at a rate of 2-3
km/hour, enabling characterization of the subsurface at that same rate. The stationary-cable
survey, on the other hand, requires 15-20 minutes to complete a single survey, the coverage
of which is never larger than the length of the cable used in the survey (ours was 27 meters).
Where it would take a matter of minutes to characterize a 100 meter stretch of shoreline with
a towed-cable resistivity survey, it would take several hours to cover the same ground with a
series of stationary-cable surveys. What the stationary-cable survey lacks in rapidity,
however, it makes up for in resolution: with a stationary-cable survey the current does not
have to penetrate a thick water column before reaching the target of the survey, so there is
less attenuation of the resistivity signal. The density of resistivity measurements per unit
length surveyed is also far greater with a stationary-cable survey than it is for towed surveys,
and the spacing between electrodes in our stationary-cable was 1 meter rather than the 2- or 4
meter spacing of our towed-cable survey. The collective impact of these differences between
the towed- and stationary-cable surveys is that resolution in our stationary-cable surveys was
4 to 5 times that of our towed cable surveys. The strengths and limitation of the two
techniques make them complimentary; with a rapidly acquired towed resistivity profile,
regions of interest can be identified for detailed survey via a stationary-cable survey.

25
3.2 Towed-Cable Surveys at Mirror Lake
We conducted 4 towed-cable resistivity surveys at Mirror Lake: two surveys using a
2-m electrode spacing and two using a 4-m spacing. Because previous studies of seepage
showed that the highest seepage rates were within 2 m of shoreline (McBride and Pfannkuch,
1975; Winter, 1976; Asbury, 1990; Rosenberry, 2005), that region was of great interest to us.
It was, however, generally not possible to approach the shoreline with the towed survey
closer than 3 meters; shallow water and an abundance of boulders and fallen trees precluded
closer passes. Surveys at two distances from shore, one nearest possible (2-5 m) and one
slightly further out (5-8 m) were conducted (Figure 3-4). In some cases, shallow water or
irregular shoreline, as exists in the southern and eastern portions of the lake, precluded any
approach to the shore closer than 5, 10 or even 20 m. Where the water was shallow or there
were peninsulas, it was necessary to detour from the desired path to avoid grounding the boat
or snagging the cable on land. Where there were coves, the electrodes at the tail-end of the
cable traveled a different path than those closest to the boat, which reduced the accuracy of
the data.
The procedure for conducting a towed resistivity survey was as follows: the
SuperSting control unit was placed in the boat. The cable was attached to the SuperSting, and
then hooked to the boat by means of a large metal loop embedded in the cable. The cable,
which was fed out from the back of the boat as the boat moved forward so that it was
ultimately in a straight line, was kept afloat by small (~15cm) sections of pool floats attached,
via plastic wire ties, to the section of cable between each of the electrodes (Figure 3-5a).
Another port in the SuperSting was attached to a Lowrance GPS and sonar unit which
recorded position and water depth information used later in the data inversion process. The
GPS receiver was mounted to the top of the trolling motor, where reception of the GPS signal

28
was best. The sonar transducer was attached, by means of a metal collar, to the bulb which
housed the propeller motor (Figure 3-5b).
Survey parameters, including electrode geometry, GPS and sonar connection
information, transmitter voltage, firing and measuring patterns, were entered into the
SuperSting by means of a keypad prior to the start of a survey. Once the cable had been fed
out behind the boat and the boat was in position to begin the survey, the “Start” key was
pressed and the Sting launched its survey, requiring, if all went well, no further input from the
user. Current was injected and potential measured every four seconds as the cable was towed
along the perimeter of Mirror Lake. At typical speeds (2-3 km/hour), this produced a set of
resistivity measurements on average every 3.5 m. A constant towing velocity of 2-3 km per
hour was maintained until the survey was finished. Survey data were stored by the
SuperSting’s internal computer and downloaded once the survey was complete. Apparent
resistivity measurements were integrated during the processing stage of the survey using the
GPS information recorded with each resistivity measurement: because electrode array
geometry was constant (except when making sharp turns), it was possible to interpolate all
electrode positions from a single GPS reading. Resistivity values could in this way be
assigned to particular places along a survey track and easily plotted on a map
3.3 Stationary-Cable (Lake-Bottom) Survey
For the lake-bottom survey we used a 28-electrode cable, where each electrode was
capable of functioning as either a current or measuring electrode (Figure 3-3B). This ability
to toggle between modes allowed the stationary-cable to accomplish from a single position
what the towed-cable did as it was physically moved about the lake; that is, by injecting
current and measuring potential drop from multiple different positions, we gathered
resistivity data about a region from multiple angles.

29
Completing all possible permutations of electrode pairs is not only impractical but
also unnecessary because certain combinations of electrodes duplicate others. Studies have
been conducted (e.g., Peake, 2005) to determine optimum survey patterns in terms of both
time and data quality, survey patterns. The dipole-dipole array, in which a pattern of 56
different pairs of current electrodes is used and potential measured at between 3 and 8
different “dipoles” (pairs of measuring electrodes) for each injection of current, is particularly
effective for mapping shallow lateral variation (Peake, 2005). It was employed for all
stationary-cable surveys in this project.
Stationary-cable surveys were conducted with the survey cable approximately
perpendicular to or parallel with the shoreline: such a regular measurement pattern made
comparing profiles easier in the analysis phases of the investigation. We conducted a total of
18 perpendicular-to-shore resistivity surveys and 25 parallel-to-shore surveys (Figure 3-4).
The majority of these (33 of 43) were conducted along a 130-meter stretch of the SW shore of
Mirror Lake. We placed flags every meter along this stretch of shoreline to facilitate
measurement reference. Locations of resistivity and radar surveys, seepage measurements,
and sediment samples were all referenced against this “Long-Line.” The other stationary-
cable surveys were conducted along the eastern shore of Mirror Lake and within the
Northeast Inlet area.
The procedure for conducting a stationary-cable survey was as follows: one person
stood on shore with the SuperSting control unit and one end of the electrode cable while the
other backed away from the shoreline in the boat and fed the remaining cable from the boat
into the lake. The cable was pulled taut once the end was reached, and then released to sink to
the bottom of the lake, with a float attached to the lake ward end of the cable to facilitate
retrieval. Electrodes were metallic and usually visible through the water column. Depth to
each electrode was measured using a measuring tape with a weight attached to it. The

30
shoreward, connection end of the cable was then attached to the SuperSting and the survey
begun. As with the towed-cable survey, once a survey had been launched no further input was
required from the user until the preset measuring pattern was completed.
3.4 Data Inversion
Once an electrical resistivity survey was completed, the resistivity data was processed
(“inverted”) into a form interpretable by the surveyor. We used EarthImager 2D for this
purpose. Preparatory steps were required before data were passed to EarthImager for
inversion. These steps varied depending on the type of survey conducted. In the case of
towed-cable resistivity surveys, GPS and depth data gathered from the Lowrance unit needed
to be synchronized with the resistivity data. A second program, Marine Log Manager®,
facilitated this process by tying resistivity data to the appropriate coordinate location by time
stamp, extracting sections of long surveys, and creating depth profiles from the sonar data
which would be used to constrain the inversion process. Marine Log Manager also allowed us
to correct minor user errors made during the survey. It was possible, for instance, to correct
for the input of a 4-m electrode spacing when a 2-m spacing was actually employed. In the
case of the stationary-cable survey, it was necessary to create a file with water depths to
constrain the inversion process.
The inversion process, by which images of the subsurface are generated from a set of
resistivity data, is straightforward in theory. The inversion software creates a model of the
subsurface environment and then conducts an “imagined” resistivity survey over that model
(Figure 3-6). It then compares resistivity data from that imagined survey with the resistivity
data actually gathered from the field. Regions of the resistivity profile where the match
between actual resistivity data and modeled resistivity values is good are kept; in regions
where the fit between observed and modeled values is poor, model resistivity values are
modified. The entire process is then repeated, with iterations continuing until the rate of

32
decrease in variance between modeled and actual resistivity values (where variance is
quantified by Root Mean Square error, described in Section 3.5) approaches zero or falls
below some other specified tolerance. The final image of the subsurface environment is a
graphic representation of the program’s model of a subsurface environment that generates the
best fit to the observed field resistivity values.
3.5 Inversion Error Analysis
Although the theory behind inversion of resistivity data is relatively straightforward,
the process is mathematically complex. A fundamental difficulty is that there may exist
multiple solutions for a single inversion problem; that is, there are potentially multiple
different subsurface environments that could generate a particular set of observed field
resistivity values. There are many different ways of modeling and sorting through data to help
mitigate this issue of non-unique solutions. Inversion software such as EarthImager offers
multiple different processing settings which affect starting models and algorithms for
inversion, thresholds for the discarding of bad data, iteration cutoff limits, and so on. Broadly
speaking, the analyst can be most confident in her data set and inversion settings when
inversion of the same data using multiple different settings yields similar final images, and
least confident when it yields very different ones. There exist a number of mathematical
measures of model accuracy, including assessments of average variation between modeled
and measured resistivity values, data misfit plots, and analysis of the sensitivity of models to
depth.
EarthImager offers several built-in tools for such error analysis. With its “Data Misfit”
tool, for example, relative model accuracy is assessed by subtracting measured apparent
resistivity from calculated apparent resistivity, and dividing that quotient by measured
apparent resistivity:

33
3.3
Results of this data misfit analysis may be displayed graphically as a “pseudosection”, in
which data misfit values are assigned a color and displayed as a function of their location
along the survey line, or as a cross-plot (Figure 3-7) (Advanced Geosciences Incorporated,
2007).
Variance between observed and modeled resistivity values can also be quantified by
averaging the square roots of the squares of the difference between the apparent resistivity
values at each equivalent point within the two models (“Root Mean Square error”) (Advanced
Geosciences Incorporated, 2007).
3.4
= predicted apparent resistivity value at each point;
= measured apparent resistivity value at each point;
N = total number of measurements
The final assessment of the root mean square error (RMS) of an inverted resistivity dataset
gives the surveyor a quick sense of the accuracy of that inverted section. This measure is
useful not only for assessing the accuracy of given models but also for determining at what
point the iteration process ought to be halted. Where a successive iteration of the model
generation/pseudo-survey process resulted in, for instance, a 20% improvement in fit between
modeled and actual resistivity values, the new model could reasonably be kept. Where
successive iteration resulted in a 0.5% improvement in model fit, on the other hand, the
iteration process is halted; such small improvements in model fit are not supported by actual
data but instead represent a fitting of the model to noise.

35
A sensitivity analysis may be used to quantify how strongly supported a model is at
different depths and at different regions. We used the method proposed by Oldenburg and Li
(1999) and further modified by Marescot and Loke (2003). This method quantifies the degree
to which the use of different starting models or inversion parameters yields a common final
model of the earth. In this technique, the resistivity data is inverted twice; once using a very
high starting resistivity value and once using a much lower one. The resistivity values of each
block of the model generated from the lower starting value of these two inversions are
subtracted from the values of each block of the model generated by inversion of the larger
one, and that value is divided by the difference between the two starting average resistivity
values.
3.5
Where inversions are well constrained by data points, RAB will approach 0 because there
exists enough apparent resistivity data that the inversion routine can calculate true resistivity.
Where the model is not well constrained by data, RAB will be large because there is not
enough data to calculate true resistivity, so the starting “guess” about the resistivity of the
subsurface strongly influences the final solution. In this way the accuracy of a given model
can be assessed on a cell-by-cell basis.
3.6 Resistivity Test Box
We used the resistivity test box to assess directly the resistivity of small (~330 cm3
)
sediment samples. The resistivity test box is a 3 x 5 x 33 cm plastic box with 2 current and 2
potential electrodes (Figure 3-8). It connects with an earlier version of the SuperSting
resistivity system, the Sting R1 IP resistivity system. This single-channel resistivity system
measure potential drop across measuring electrodes M and N for each injection of current

36
through electrodes A and B. An
Figure 3-8

37
on-board computer calculates and reports resistivity from that potential drop.
3.7 Sediment Sampling
We collected sediment samples using clear plastic collection tubes, approximately 5
cm in diameter (Figure 3-9), to assess the resistivity of saturated sediment samples using the
resistivity testbox. We collected cores from locations where it was possible to insert the
collection tube (approx. 5 cm in diameter) at least approximately 10 cm directly into
sediment. This was generally only possible where sediment was very soft, as it tended to be
in regions where there were thick deposits of organic-rich material. We also collected
sediment samples from regions where it was not possible to insert the tube directly into
sediment (i.e., in regions where sediment was hard-packed) by scraping samples from the
upper few cm of sediment into collection tubes. We assessed the resistivity of saturated
sediment samples using the resistivity test box.

39
3.8 Ground Penetrating Radar
Ground penetrating radar was employed in this project as an adjunct to electrical
resistivity profiling. Whereas electrical resistivity profiling introduces an electric current into
the earth and measures how well the earth conducts that current, ground penetrating radar
(GPR) introduces electromagnetic waves to the subsurface and measures how those waves
reflect. The technique is useful to geologists because radar waves are reflected by bedding
planes and other major changes in sedimentation, and can reveal a detailed image of
subsurface structure.
The behavior of radar waves in a given medium is controlled by the dielectric
permittivity (ε) of that medium, its electrical conductivity (σ) and its magnetic permeability
(μ), properties that are primarily functions of freshwater content of the substrate and of more
difficult to quantify structural properties (Neal, 2004; Reynolds, 1997).The relationship of
these properties to wave velocity in a substrate (ν) is described by:
3.6
where c0 is the velocity of electromagnetic waves in a vacuum, ω = 2πf and is the angular
frequency of the signal in radians/sec, ε is the dielectric permittivity of free space, εr is the
host medium’s dielectric permittivity relative to free space, σ is its bulk conductivity at the
given frequency, and μr is its relative magnetic permeability, and describes attenuation of
the signal as it passes through host media (Neal, 2004). A wave’s amplitude (A) also changes
as it passes through a medium, that amplitude declining exponentially from its initial value
and described by the equation

40
3.7
where α is an attenuation constant caused by loss of energy via conversion of the radar signal
to heat and is dependent upon the dielectric and conductive properties of the host medium, A0
is a wave’s initial amplitude, and z is distance traveled by the wave.
Where ε, σ, ω and μ remain constant, radar waves will be transmitted through the
medium without reflection. Where ε, σ, and ω vary through the course of a geologic section
(μ generally remains constant) as the result of changes in porosity, grain composition,
orientation and packing, or amount and type of pore fluid, reflections will be generated at the
points where those properties change over a zone greater than or equal in length to
approximately three times the dominant radar wavelength (Neal, 2004). Objects such as
cobbles and boulders represent sudden, marked variations in ε, σ, and ω; such objects reflect
electromagnetic energy strongly, and produce characteristic diffraction hyperbolae in
radargrams (Figure 3-10). The amplitude of reflected waves is quantified by R, the
coefficient of reflection,
3.8
where V1 and V2 are the velocities of radar waves in adjacent layers 1and 2. It is these
reflections which are collected and plotted in a GPR survey and which collectively comprise
the images of the subsurface. If average radar velocities for a given medium are known,
reflections can be plotted as a function of depth by measuring time between the introduction
of an electromagnetic wave to a substrate and the return of reflections to a receiving antenna.
Shading the positive (or negative) portions of the plots of reflected waves, as is done to create
radargrams (Figure 3-10) can yield an image of the sub-surface environment. Since radar
surveys are sensitive to changes in grain size, porosity, packing, orientation and other

41
sedimentological factors that tend to be associated with bedding planes, such plots can image
bedding planes, lithologic, and other structural features.
Figure 3-10

42
Selection of antenna frequency for a GPR survey depends upon the aims of the survey.
It is generally possible to resolve features as small as approximately one quarter the
wavelength of the dominant radar frequency, so with higher frequency comes improved
resolution (Reynolds, 1997). Higher frequency signals, however, do not penetrate as deeply
into the subsurface. A good first approximation of the depth of penetration for any GPR
survey is given by the so-called skin depth, defined as the depth at which wave amplitude has
decreased to (approximately 37%) of its original amplitude (Reynolds, 1997). Skin depth
is defined mathematically as
3.9
where δ is skin depth (in meters), ω is angular frequency, μ is magnetic permeability and σ is
conductivity (Reynolds, 1997). Depth of penetration, therefore, is not only dependent upon
frequency but also upon the nature of the material being studied. Highly conductive materials
such as clay and sea water are virtually opaque to radar waves. Highly resistive materials, on
the other hand, such as unfractured granite, can transmit radar waves for many tens of meters.
We used a MALA geosciences 250 MHz antenna in this project, which represents a
frequency middle ground in terms of vertical resolution vs. depth of penetration.
Our 250 MHz GPR system included a transmitting and a receiving antenna housed in
a plastic case 0.73 meters apart and connected to a computer recording system. To this
computer recording system we attached a Trimble GPS unit, which recorded position
information for the radar data. We loaded the entire suite of into the bow of our polyurethane
boat, with the bottom of the antennae nearly flush with the floor of the boat (Figure 3-11).
We conducted a total of 40 GPR surveys: 20 perpendicular to shore at regular

43
intervals (every 50 meters) around the perimeter of the lake to develop a sense of the basic
Figure 3-11

44
lithology, and 20 perpendicular to shore at closer intervals (approximately every 10 meters)
within our southwest shore study area (Figure 3-12). We also conducted a parallel-to-shore
survey of the lake’s perimeter, keeping a distance of 3-5 meters from the shore so as to
traverse the same area covered by the outer of our towed-cable resistivity surveys.
Several data processing steps facilitated interpretation of GPR data. We processed
GPR data using both Reflex 2-D Quick® and MatGPR software. First, reflector depths were
calculated and assigned to radargrams, using a wave velocity of 0.033 m/ns, the speed with
which radar travels through water. Next, linear coordinates were extrapolated from GPS data
and assigned to the radargrams. Third, an automatic gain control (AGC) function was applied
to amplify radar reflections as a function of depth. Without such amplification, all but the
nearest-surface reflections would be invisible because signal strength decreases exponentially
with distance from the source. Finally, the appearance of noise in radar data, primarily caused
by reverberation and by scattering, was reduced using a variety of filtering functions. The
same processing steps and setting were applied to all radargrams.

46
3.9 Seepage Meters
In-situ measurements of the exchange between groundwater and the lake was
measured with seepage meters of the design suggested by Lee (1977) as modified by
Rosenberry et al. (2005) (Figure 3-13). A seepage meter works by sealing a selected region of
sediment from the water column around it. A measuring bag attached to an outlet in this
sealed zone, filled with a small quantity of water, will respond to local flow patterns.
We used one half of a 55-gallon (0.208 m3
) plastic drum for these purposes. A hole
(approximately 1.8 cm3
, the size of a common garden hose) was drilled in the top of each
half-drum. A threaded brass fitting was placed in the hole and. sealed against the sides of the
hole by means of caulk and plumber’s tape. A length of garden hose and a measuring bag was
attached to the metal fitting. The drum was pressed into the sediment above the region of
interest. The entire unit was then allowed to sit undisturbed for at least one hour to allow re-
equilibration of seepage in the region. At the end of this re-equilibration a period an
approximately 1.5 liter plastic bag, filled approximately half-way with water and weighed,
was attached to the free end of the hose by means of a brass fitting. The bag was placed in a
protective plastic case to prevent waves, wind and other disturbances from affecting seepage
from or to the collection bag. The shut-off valve of the brass-fitting was opened to allow
exchange between the local hydrologic system and the water in the measuring bag. All
seepage from the area beneath the drum changed the quantity of water in the bag. Where
seepage was from the lake to the groundwater system (“out-seepage”), the contents of the
plastic bag was drawn down. Where seepage is from the groundwater system to the lake (“in-
seepage”), the measuring bag filled. After an appropriate length of time (hours to days if the
seepage rates are low, minutes if rates are high) the shutoff valve of the measuring bag was
closed and the bag removed from the seepage meter and weighed. From the known cross-
sectional area of the seepage meter, the weight of the collection bag before and after the

47
measurement, and the length of the measurement period, seepage rates are calculated,
Figure 3-13

48
generally expressed in centimeters per day.
We made seepage measurements at a total of 74 points in Mirror Lake, the vast
majority of which (72 out of 74) were within our intense study area of the southwest shore of
that lake. Seepage measurements were made every 10 meters along the 130 meter stretch of
our southwest shore “Long Line” (all of which were at a distance of approximately 2 meters
from the shoreline) and at several intermediate points as well. We made additional
measurements further from shore, including five seepage “transects” in which seepage meters
were placed at 2-3 m intervals along a straight line out into the lake to a maximum of 16
meters from shore. At least three repeat measurements were made at each seepage meter site
to ensure reproducibility of results.
3.10 Bathymetry
We gathered detailed bathymetric information for the southwest shore of Mirror Lake
by traversing the area with our boat while we logged Sonar depth and GPS data (both given
by the Lowrance navigation unit we used for SuperSting towed-cable resistivity surveys) with
a laptop computer. By “zigzagging” over the southwest shore area and collecting depth and
position information every second, we collected a dataset from which we could create a
contour map of the area. We did so using the Kriging interpolation method via the Surfer®
(Golden Software) gridding and plotting program. The resulting bathymetry map is included
in Appendix A.
3.11 Data Coordinates
We collected GPS data for nearly all measurements conducted at Mirror Lake using a
Trimble Geo XT GPS unit and a Lowrance GPS & sonar unit. Much of the data collected
with the Trimble unit were logged directly by that unit. Data logged directly into the Trimble
could be differentially corrected during post-processing, a step which improved precision to

49
approximately ± 0.5 m. Where signal quality was poor, the Trimble would not allow logging
directly into the unit; in those cases, we recorded approximate coordinate locations by hand
as they could be read from the Trimble display screen. Precision in those cases was
approximately ±5 m. The Lowrance unit was employed only during towed-cable resistivity
surveys; that unit connected and communicated directly with the SuperSting resistivity survey
system. No differential correction of the Lowrance GPS data was possible, and position data
collected with the Lowrance unit were accurate ± 5m. ArcGIS mapping software was used to
plot the position data.

50
CHAPTER 4
RESULTS AND DISCUSSION
4.1 Introduction
Our work at Mirror Lake aimed to test the hypothesis that geophysical techniques can
image geologic heterogeneities which control seepage, or image seepage itself. To test this
hypothesis we made seepage measurements at regular intervals over areas where we
conducted electrical resistivity and ground-penetrating radar (GPR) surveys. We conducted a
total of 4 around-the-lake towed-cable resistivity surveys, 43 stationary-cable resistivity
surveys, 2 parallel-to-shore GPR surveys (including one round-the-lake survey), and 57
perpendicular-to-shore GPR surveys. We measured seepage at a total of 74 locations around
Mirror Lake, 72 of which were along its southwest shore. Finally, we collected sediment
samples at 22 locations, 20 of which were along the southwest shore.
4.2 Resistivity Survey Results Overview
We employed both 22-m and 44-m cables in the towed surveys, with electrode
spacings of 2-m and 4-m respectively. We conducted two surveys with each of these cables,
one attempting to stay within 2-3 m of shore and one slightly further away (4-5 m). Total
depth of penetration, including the water column, was approximately 5 m with the 22-m cable
and approximately 10 m with 44-m cable. Where the slope of the lake was gentle and where
there were near-shore debris and man-made obstacles such as rafts, it was not possible to
approach the shoreline by closer than 5-20 m. It was also not possible to maintain the desired
distances where the shoreline was irregular, as where there were coves and peninsulas. Such
difficulties were more frequently encountered and were more pronounced with the longer
cable. The 22-m cable survey therefore generally returned the longest uninterrupted data
streams. In particular, the 22-m cable survey conducted 2-3 m from shore returned the most
useful data; water depths at the furthest distance were 3-4 m, so depth of penetration into

51
sediment was only 1-2 m. All towed-cable resistivity profiles presented in the body of this
thesis are therefore from the 22-m cable survey at a target distance of 2-3 m from shore.
Portions of the survey omitted are places where irregular shoreline or near-shore debris
precluded towing the cable in a straight line and in sufficiently shallow water.
Results from the 22-m cable survey are presented in Figure 4-1A-G. Resistivity
values are contoured by color; a constant color scale is used throughout this thesis. The
highest resistivity values (~3000 Ω-m) are in red. Intermediate resistivity (~1500 Ω-m) values
are in yellow. The lowest resistivity values (<200 Ω–m) are in blue. The white line indicates
lake-bottom, inferred from sonar data collected with the towed-cable surveys. Depth scale is
provided on the left hand side of the inverted images. Inversions were constrained using lake
depth. Lake water conductivity, a second inversion constraint, was established via a
conductivity-meter as 330 Ω-m. That value was consistent with the results of inversion of a
section of resistivity survey conducted over open water. Root mean square error (RMS) for
the towed-cable surveys ranged from approximately 3 -10%. Sensitivity analyses (described
in Section 3.5) suggested that RAB (Equation 3.5) was ≤10% in the upper 1-3 m, with a
gradual transition to greater than 30% below 4m. The surveys have therefore been truncated
at a depth of approximately 4 meters.
Towed-cable resistivity surveys indicated that the resistivity of most sediment at
Mirror Lake was 1000 – 3000 Ω-m, with discrete zones of low (≤ 200 Ω-m) resistivity (Figure
4-1A-G). In some regions, sediment resistivity was relatively homogeneous (e.g., Figure 4-
1A). In others, it was heterogeneous (e.g. 4-1B). Some features observed in resistivity
profiles were easily related to geologic features. Along the eastern peninsula bedrock was
exposed; along the northern shore, boulders were exposed. In both cases, sediment resistivity
was high (> 3000 Ω-m) (Figure 4-1C and G). In the region of inlet and outlet streams, and in
a few other locations, thick (> 0.5 m) deposits of organic matter were apparent (discussed in

52
more detail in Section 4.4). Where such deposits were observed, resistivity was low (≤ 200 Ω-
Figure 4-1A

55
m) (Figure 4-1B, C, D, F, and G).Where a curved path was taken to navigate a cove,
electrodes at the tail end of the cable were pulled across open water; such paths produced
resistivity profiles wherein the upper few meters of “sediment” were apparently equal in
resistivity to lake water. Such resistivity values do not accurately reflect the resistivity of
sediment traversed by the boat and the electrodes nearest to it but rather the resistivity of the
open water traversed by the tail-end electrodes.
The results of stationary-cable resistivity surveys are presented in Figure 4-2A-K. One
line (Perpendicular Line 3) is omitted because proximity of that survey was influenced by
proximity an electromagnetic flow meter. Two others, perpendicular resistivity surveys of the
eastern shore, have been omitted because the very steep terrain along the shoreline there
prevented inversion of those data. The black dots in stationary-cable profiles represent the
location and depth of electrodes. The same color and depth scale used for presentation of the
towed-cable results is used for the results of the stationary-cable surveys. Because the cable
used in stationary-surveys is only 27 m long, any inversion images which show an inverted
section longer than that represent the merged results of two or more overlapping stationary-
cable surveys. RMS was less than 5% for inversion of single stationary-cable surveys, and 7-
15% for merged datasets. RAB values for inversions of single stationary-cable surveys were
similar to those associated with towed-cable surveys; RAB was ≤10% in the upper 1-3 m, with
a gradual transition to greater than 30% below 4 m. Like the towed-cable profiles, stationary-
cable surveys have therefore been truncated at a depth of approximately 4 meters. Even with
this truncation, however, zones of lower resistivity (highlighted in 4-2A) appeared at the
bottom of all stationary-cable profiles. Zones of low resistivity at depth also appeared in the
inversion of synthetic, uniform earth resistivity data. Low resistivity at depth is therefore an
artifact of inversion and does not necessarily reflect actual low resistivity at depth. For
inversion of merged datasets, RAB was up to 50%, with an average of approximately 30%.

59
Stationary-cable surveys allowed the approach of terrain inaccessible to towed-cable
resistivity surveys, such as coves, peninsulas, and shallow water. The stationary-cable
surveys also offered resolution which was 4-5 times that of the towed-cable surveys. The
higher resolution of stationary-cable surveys was a product of several factors: first the
electrodes in a stationary-cable survey lay directly upon lake-bottom, so electrical current did
not have to penetrate a thick water column before reaching the target of the survey. Second,
many more measurements per unit length were made in the stationary-cable surveys than in
towed surveys. Finally, the electrode spacing of our lake-bottom cable was 1-m rather than
the 2-m and 4-m spacing we used in the towed-cable surveys. This higher resolution and
greater versatility, in terms of where the cable could be deployed, however, came at the
expense of slower data acquisition.
Like the towed-cable resistivity surveys, stationary-cable surveys showed that
sediment resistivity ranged from less than 200 Ω–m to approximately 3000 Ω–m. Where
towed- and stationary-cable surveys were conducted on overlapping terrain, however, the
towed-cable survey consistently returned lower measurements of resistivity highs than did the
stationary-cable surveys. A comparison of the results of towed- and stationary-cable surveys
is presented in Figure 4-3. Along the eastern edge of Mirror Lake, towed-cable surveys
returned maximum average resistivity values of approximately 1500 Ω-m, while the
stationary-cable surveys returned values of approximately 3000 Ω-m (Figure 4-3A). Towed-
cable surveys of the northeast inlet area of Mirror Lake returned maximum average resistivity
values of approximately 1500 Ω-m, whereas highest average resistivity values from the
stationary-cable survey were in the 3000 Ω-m range (Figure 4-3B). Along the southwest
shore, towed-cable surveys conducted at a distance of approximately 10 m from shore
showed resistivity values which average around 1500 Ω-m (Figure 4-3C). Parallel-to-shore

60
stationary-cable surveys conducted at a distance of 8 m from shore, showed average
Figure 4-3

61
resistivity values to be about 3000 Ω-m. In all three cases, un-inverted apparent resistivity
values as well as the inverted data were lower in the towed-cable surveys, suggesting that the
discrepancy observed in inverted sections was the product of differences in raw data rather
than a product of inversion.
I believe that the discrepancy is likely the product of differences in current travel path
between towed- and stationary-cable surveys. Towed-cable surveys were conducted where
water depths were approximately 2 m, which corresponded to being generally over or just
beyond the edge of a narrow (~ 2 m) underwater plateau which rings the lake; beyond that
plateau the sides of the lake are much steeper, plunging steeply towards the maximum lake
depth of 7m +. Stationary-cable surveys, however, were generally conducted atop these
plateaus: that is, regions where we conducted stationary-cable surveys had to be flat enough
that the towed-cable would not roll down the banks of the lake. They were therefore
necessarily along the plateaus. A greater portion of each resistivity “signal” was therefore
likely influenced by lake water in the towed-cable surveys. Towed-cable resistivity surveys
were subject to more sideswipe – survey of terrain adjacent to the survey cable (in this case
lake water) rather than immediately below it – than were stationary-cable surveys. The lower
resistivity values returned by towed-cable surveys therefore probably reflect the fact that
proportionally more current flowed through the water. Because more of the current in a
stationary-cable survey will interact with the geology, and because the density of resistivity
measurements per unit area with the stationary-cable survey is approximately 6 times that of
the towed-cable survey (thereby providing significantly greater constraint on the models
generated by the inversion), I believe the resistivity values returned by the stationary-cable
surveys more accurately reflect the resistivity of sediment. Towed-cable resistivity values
therefore ought to be considered underestimates of true sediment resistivity. The majority of
interpretation about resistivity survey results in relation to seepage is limited, in this thesis, to

62
the results of the stationary-cable surveys.
4.3 Southwest Shore Resistivity And Seepage Results
We focused study along the southwest shore of Mirror Lake where high seepage rates
were previously observed (Asbury, 1990). We placed flags every meter along a 130 m stretch
of that shore at a distance of 2 m from land. We conducted seepage measurements every 10 m
along that “Long Line", and at several intermediate points as well. We also conducted 5
seepage “transects”, measuring seepage every 2-3 m extending out perpendicular to shore, to
a maximum distance of 16 m from shore. We conducted parallel-to-shore stationary-cable
resistivity surveys over the entire length of the Long Line at a distance of 2 m from shore,
overlapping adjacent surveys by approximately 50% so that we could generate a single
resistivity profile for the Long Line. We also conducted parallel-to-shore surveys at 4, 6 and 8
m from shore within a zone where seepage rates were highest. Finally, we conducted
perpendicular-to-shore surveys along each of the seepage transects. The results of
Perpendicular Survey 3, the location of which was shown in Figure 4-2, are ignored in the
interpretation because proximity to an electromagnetic flow meter influenced those data.
4.3.1 Resistivity Values 1500 to 3000 Ω–m and Seepage Rates
The results of southwest shore seepage measurements are shown in Figure 4-4. The
highest seepage rates were observed within 2 m of shore. These highest seepage rates
(average = -80 cm/day, with a high of -277 cm/day) were clustered within a zone
approximately 40 m long. Two of the five seepage transects were within the zone of very
high seepage. Seepage rates declined to close to 0 cm/day by 12 m from shore in-seepage
transects 1 through 4. Transect 5 represented an exception to the pattern of decreasing
seepage rates with distance from shore: slightly positive values were observed at 12 and 16 m
from shore.
Parallel-to-shore resistivity surveys indicated that average sediment resistivity

64
increased with distance from shore (Figure 4-2E-H). At 2 m from shore, average sediment
resistivity was approximately 1500 Ω–m (Figure 4-2E). At 8 m from shore, it was
approximately 3000 Ω–m (Figure 4-2H). There is some evidence in the perpendicular-to-
shore surveys for increasing resistivity with distance from shore, but the variation is not clear-
cut. No decrease in resistivity with proximity to shore appears in the Survey 1 (Figure 4-2A),
the near-shore end of which overlapped with the southern end of the parallel-to-shore surveys
(Figure 4-2E-G). Sediment resistivity in Survey 2 (Figure 4-2B), which also overlapped with
the parallel surveys, appears to decrease within 4 m of shore; this apparent decrease,
however, does not necessarily reflect a decrease in sediment resistivity. Data are sparse at the
ends of stationary-cable surveys and so inversions are not well constrained there.
Perpendicular Surveys 4 and 5 (Figure 4-2C & D), indicate that average sediment resistivity
is about 1500 Ω–m until approximately 10 and 12 m from shore respectively, several meters
further than is indicated by the parallel-to-shore surveys. Because the increase in resistivity
from 1500 to 3000 Ω–m indicated by the parallel-to-shore profiles occurs in the middle of the
profiles where inversion models are well-constrained, I believe it is fair to assert that
sediment resistivity does indeed increase in such a manner. The fact that perpendicular-to-
shore surveys are inconclusive in this regard and in some cases differ from the parallel lines
is likely a product of the fact that data are sparse at the ends of the surveys. In short, the
results of parallel-to-shore resistivity provide more information about the near-shore
environment than do the perpendicular-to-shore surveys.
As stated previously, parallel-to-shore resistivity surveys at 2-, 4-, 6-, and 8-m from
shore (Figure 4-2E-H) within the zone of highest seepage rates showed that average sediment
resistivity was approximately 1500 Ω–m at a distance of 2 m from shore and 3000 Ω–m at a
distance of 8 m from shore. Seepage measurements showed that seepage rates decreased with
distance from shore. That is, lower seepage rates were coincident with higher resistivity. A

65
similar coincidence between lower resistivity (~1500 Ω–m) values and higher seepage rates
was observed at a distance of 2 m from shore, where a transition in sediment resistivity was
observed approximately midway along the Long Line survey (Figure 4-5A and B): average
sediment resistivity is approximately 1500 Ω–m until meter 70 of the Long Line, and
approximately 3000 Ω–m after meter 70. Seepage rates were highest along the Long Line
(average = -80 cm/day, with a high of -277 cm/day) where resistivity values were
intermediate (≈ 1500 Ω-m). Seepage rates were lowest (average = -20 cm/day, with a high of
-90 cm/day) where resistivity values were ≥3000 Ω-m or ≤100 Ω-m.
I have interpreted the relationship between intermediate (~1500 Ω–m) resistivity and
high seepage rates based upon Archie’s law. Bulk resistivity is, according to Archie’s law,
dependent upon porosity, pore fluid conductivity, saturation, and mineralogic and cementing
characteristics inherent to a given lithology (Equation 3.1). Saturation was constant in the
case of our study because sub-bottom sediment represents a fully-saturated environment. We
also expected pore fluid conductivity to remain constant because in all but a few cases all
seepage was out-seepage, so pore fluid would have had the same conductivity as lake water.
The source of variation in resistivity was therefore either a change in porosity, mineralogy, or
cementing factors. Major shifts in mineralogy or a change in clay content, reflected in
variables a, m, and n, are necessary to cause significant changes in resistivity. Clay content at
Mirror Lake is uniformly low, and mineralogy relatively consistent (Harte, 1997). Variation
in porosity is, by process of elimination, a likely source of the variations in resistivity
observed along the southwest shore of Mirror Lake.
The results of testbox analysis of sediment resistivity are consistent with this
interpretation; testbox values of the resistivity of siliceous sediment were 300-1000 Ω–m.
These values were lower than the field values of 1500-3000 Ω–m. Testbox values for the

66
resistivity of organic-rich sediment resistivity were 150 - 300 Ω–m, where field values were
Figure 4-5A

68
< 200 Ω–m. I believe this discrepancy arises because the sediment samples were disturbed
during the process of transferring samples from collection tubes to the testbox: this
disturbance almost certainly increased porosity and affected the resistivity values. In the case
of siliceous sediment, increased porosity led to lower resistivity. In the case of organic-rich
sediment, an increase in porosity increased bulk resistivity because there was a lower
concentration of high-conductivity material.
To test the hypothesis that changes in porosity were responsible variations in
resistivity along the southwest shore of Mirror Lake, I used Archie’s Law to model the effects
of a change in porosity on resistivity values. Because sediment is unconsolidated at Mirror
Lake and because the lake water was resistive (330 Ω –m), only the lowest published
(Reynolds, 1997) values of variables a and m (0.5 and 1.3 respectively) yielded resistivity
values consistent with those observed at Mirror Lake and within a reasonable range of
sediment porosity values (12-40%) (Harte, 1997); higher values of a and m yielded resistivity
values higher than any observed at Mirror Lake. Holding those variables constant at the
lowest published values, models showed that an approximately 12% change in porosity can
cause a 1500 Ω–m change in resistivity (Figure 4-6).
It is reasonable to assume that porosity varies along the southwest shore because that
shore likely contains an outwash lens from the receding glacier which formed Mirror Lake
(Section 2.2). While sediment underlying most of a glacially derived lake tends to be poorly
sorted, sediment within its outwash lens is sorted by stream action. Degree of sorting within
these systems will vary depending upon the position of channels, with discrete zones of more
poorly-sorted material and also discrete zones of more well-sorted material (Ritter et al.,
1995). Because porosity is affected by degree of sorting, porosity will vary with sorting
within outwash fields. In a clay-poor environment such as Mirror Lake, variations in porosity

69
are related to variations in hydraulic conductivity. The results of previous core, hydrologic
Figure 4-6

70
and modeling studies at Mirror Lake provide evidence of variable sorting along the southwest
shore of Mirror Lake, and associated variations in hydraulic conductivity. In particular, there
is evidence for a deposit of well-sorted sand and gravel along the southwest shore (Figure 2-
2). Archie’s law suggests that such a deposit would have a lower bulk resistivity than more
poorly sorted sediment because it has a higher porosity. The bounds of the southwest shore
sand and gravel deposit coincide approximately with the transition from intermediate (1500
Ω–m) to higher (3000 Ω–m) average resistivity within our Long Line survey area (Figure 4-
7). There are, therefore, multiple lines of evidence to suggest that the transition from
intermediate (1500 Ω–m) to higher (3000 Ω–m) average resistivity within the Long Line
survey area reflects variations in sorting, porosity and hydraulic conductivity.
Where there is an exception to the pattern of low seepage rates coincident with high
(3000 Ω–m) resistivity, (i.e., at meter 90 of our Long Line, where a -62 cm/day seepage rate
was measured within a region where average sediment resistivity is 3000 Ω–m) I believe it is
likely that a zone of lower resistivity is present but that it was small or slightly offset from the
resistivity line and could not be resolved with the equipment used in this study. The fact that
a considerably lower seepage rate (-8 cm/day) was observed at meter 92 of the Long Line,
just 2 m from the higher rate, is consistent with the interpretation of small-scale geologic
heterogeneity. Further study with a resistivity survey with a smaller electrode spacing,
sensitive to features on the sub-meter scale, would provide more information.

72
4.3.2 Low (200 Ω-m) Resistivity and Low Seepage Rates
Zones where resistivity was less than or equal to 200 Ω-m coincided with deposits of
organic-rich matter at least 0.5 m thick. Organic matter was readily distinguishable from
siliceous deposits in the field by color (organics were dark brown-black, siliceous sediment
was grey and tan), texture (organics were fine and smooth where siliceous sediment was
gritty) and smell (organic deposits had a distinct sulfurous odor where siliceous sediment was
odorless). The low resistivity (<200 Ω–m) of organic material was confirmed in the lab with
the resistivity test box; where organic matter content was high, as judged from color, texture,
and odor, resistivity was approximately 150-300 Ω–m; where organic matter content was
lower, resistivity was approximately 300-1000 Ω–m. Organic content was confirmed in the
lab via drying and sieving the sediment; pieces of sticks, bark, and leaves were observed in
the sieved, dried samples along with siliceous sediment. Permeameter tests of the hydraulic
conductivity of this sediment suggested K was lower than that of organic poor sediment: K
values of 3 x 10-3
and 5 x 10-4
cm/sec were measured in organic-rich sediment, whereas K in
more organic-poor samples was greater than 5x 10-3
cm/sec, with an average of 1x 10-2
cm/sec (Dr. Laura Toran, personal communication, 6/15/08). The low seepage rates observed
where resistivity surveys indicate sediment resistivity values of ≤ 200 Ω–m are attributed to
the low hydraulic conductivity of organic-rich sediment.
4.4 Eastern Shore Seepage and Resistivity
The results of parallel-to-shore resistivity surveys, conducted approximately 2 m from
shore along the steep eastern shore of Mirror Lake, were presented in Figure 4-2J and K.
Those sites were chosen to test the hypothesis that geologic heterogeneities observed with the
towed-cable survey would relate to seepage rates. The results of the 22-m towed-cable
resistivity survey indicated average sediment resistivity transitioned from approximately 1000

73
Ω-m to approximately 2500 Ω-m within the span of approximately 30 m along the
northeastern shore. Stationary-cable surveys conducted over that terrain confirmed a
transition in sediment resistivity, though with slightly higher values than indicated by the
towed-cable surveys; average sediment resistivity, as assessed with the stationary-cable
surveys, was 1500 -2000 Ω-m in the more northern of those surveys and approximately 3000
Ω-m in the more southern. A layer of organic-rich material, which varied in thickness from
approximately 0.10-0.5 m thick, was also indicated in the stationary-cable surveys, and was
confirmed through examination of hand-samples of sediment there. Seepage rates were
measured coincident with those resistivity surveys, at distances of approximately 2 and 6 m
from shore and approximately in line with the perpendicular-to-shore resistivity surveys there
(shown in Figure 4-1A)*
. Seepage rates at both locations were slightly positive (<0.06
cm/day) but below detection limits. These low seepage rates were likely influenced by the
presence of the organic matter deposits. That is, though a geologic heterogeneity was
identified with the towed-cable survey which might have related to seepage (i.e., a transition
from intermediate to higher average resistivity), finer-scale features observed with the follow-
up, stationary-cable surveys (i.e., organic matter) were likely a more powerful control on
seepage.
4.5 Northeast Inlet: Resistivity and Road salt Contamination
Towed-cable resistivity surveys indicated a zone of low (≤ 200 Ω-m) resistivity in the
Northeast Inlet area (Figure 4-8A). This zone extended to a depth of approximately 4 m in
sediment. We followed up with stationary-cable resistivity surveys there, which also imaged
a zone of ≤200 Ω-m resistivity extending from the surface of the lake-bottom to a maximum
depth of 4 m (4-8B and C). In other regions of the lake, zones of low resistivity were related
*
Tree cover along the Eastern Shore prevented the collection of firm coordinate data for
both seepage and resistivity measurements made there. Map locations of all data are
therefore approximate.

74
to deposits of organic matter. Sediment samples from the northeast inlet region indicated that
there were significant organic matter deposits there. The deposits of organic matter, however,
ranged in thickness from approximately 0.1 to 0.5 m. The zone of low (≤200 Ω-m) resistivity
in the northeast inlet area extends to a depth of approximately 3-4 meters, more than 2 meters
past the organic matter deposits into the siliceous drift.
I hypothesize that the source of this zone of very low resistivity within the siliceous
sediment of the northeast inlet area is a decrease in pore-fluid resistivity caused by the
presence of road salt. Since the construction of a Highway 93 along the eastern edge of
Mirror Lake in the 1970s, sodium and chloride concentrations in the lake have risen steadily
(Rosenberry et al., 1999). Research on the delivery mechanism of this road salt contamination
suggests that both groundwater and surface water transport sodium chloride to the lake.
Rosenberry et al. (1999) estimate that, on average, 2.2% of total chloride contamination is
delivered to Mirror Lake by groundwater, and 49% via the Northeast Inlet stream. A pore
water sample from the shallow (≈ 40 cm) subsurface revealed chloride content three times the
level observed in the lake (9 mg/L Cl-
in northeast-inlet area pore water versus 2.5 mg/L Cl-
in the lake). We also found that the resistivity of this pore water was 3.3 times lower that the
background level (approximately 200 Ω-m rather than 330 Ω-m).
To test whether the presence of chloride could cause the very low resistivity observed
beneath the organic matter deposits of the northeast inlet area, I used Archie’s law to model
the impact of an increase in pore fluid conductivity on bulk resistivity. The results of these
calculations show that a 3.3x increase in pore fluid conductivity ( w) will lead to a 3.3x
decrease in resistivity. The range of resistivity we measured for the drift which underlies
Mirror Lake was approximately 1500 to 3000 Ω-m. A 3.3x decrease in pore-fluid
conductivity would lower that range to 450 - 900 Ω-m. Our resistivity surveys of the
northeast inlet indicated minimum resistivities of less than 100 Ω-m at 3-4 m depth. This

75
suggests that chloride concentrations were likely even greater at depths of 3-4 meters within
Figure 4-8

76
sediment than that measured at 0.4 m. Such a conclusion is also suggested by the nature of
resistivity data inversion, which tends to smooth transitions between adjacent survey areas,
and to thereby underestimate resistivity contrasts. It is therefore likely that the hypothesized
zone of chloride contamination is even more conductive than is indicated by the inversion
image.
These results collectively suggest that resistivity surveys were able to confirm the
presence of road salt contamination and to partially map its extent. The results suggested
highly localized, concentrated deposits of chloride within the northeast inlet area, consistent
with a contaminant plume. It is likely, in fact, given that resistivity surveys underestimate and
smooth resistivity contrasts in adjacent regions, that the zone of chloride contamination is
both smaller and more concentrated than is indicated by the zone of low resistivity observed
in the profiles from the contaminated region. It is not possible to tell from the resistivity
surveys alone whether road salt is delivered to northeast inlet via groundwater or surface
water. Given, however, that previous research has measured chloride contamination in
groundwater wells between the northeast inlet area of Mirror Lake and the highway which is
the source of the chloride, and hydraulic gradient measurements suggest groundwater
discharges to Mirror Lake within the Northeast inlet area (Rosenberry, 1999), it is likely that
seepage of contaminated groundwater to the northeast inlet area contributes to the chloride
concentrations in the lake.
4.6 Ground Penetrating Radar Surveys
We conducted a total of 54 radar surveys at Mirror Lake: 34 perpendicular-to-shore at
50-m intervals along the lake’s perimeter, 19 perpendicular-to-shore surveys at approximately
10 m intervals along the southwest shore, 1 parallel-to-shore survey about the entire
perimeter of the lake and 1 parallel-to-shore survey within the Long Line survey area of the
Southwest Shore (shown in Figure 3-12). Representative samples of the perpendicular-to-

77
shore surveys are presented in Figures 4-9A-D. Resolution in the parallel-to-shore surveys
was poor because the shallow water in which we conducted them caused significant
reverberation of the radar signal (highlighted in Figure 4-9A); the results of those surveys are
omitted from the body of this thesis.
Several features appeared consistently in the radargrams we collected at Mirror Lake:
diffraction hyperbolae (produced by objects such as cobbles and boulders which scatter
electromagnetic energy), bedding planes, and the organic sediment (“gyttja”) which covers
the bottom of Mirror Lake. The distance scale along the top of each radargram represents
distance from shoreline. The depth (in meters) on the right y axis represents the inferred
depth of features that reflected a radar pulse. The wavy lines within the water column are
noise amplified by the application of automatic gain control to raw data. Upside-down U-
shaped features with very straight edges are interpreted as diffraction hyperbolae and are
highlighted in red (4-9A’). Roughly linear features which extend for at least several meters
and that are not parts of diffraction hyperbolae are interpreted as bedding planes. Bedding
planes are highlighted in yellow and in green (4-9A’). Where bedding planes existed within
the smooth, flat sediment of the lake bottom they were interpreted as bedding within gyttja
(yellow). Where bedding was observed within sediment at the side of the lake it was
interpreted as stratification of sediment (green).
Stratification within the sediment at the sides of the lake was observed in only 3
radargrams at Mirror Lake: radargrams B and C (the map locations of which are shown in
Figure 3-12), radargrams for which are shown in Figure 4-9B and C) and the line in between
them. That stratification was observed at distances of approximately 15 to 45 m from shore at
depths of 2-3 m in sediment. It corresponded with water depths of approximately 2-7 m. All
other radargrams from Mirror Lake showed bedding only within the organic sediment (gyttja)

82
which covers the bottom of the lake (e.g. as shown in Figure 4-9D).
Depth of penetration into the stratified bedding shown in radargrams B and C was
greater than into glacial till elsewhere around the lake (e.g., that shown in D): it was
approximately 2-4 m into sediment rather than 1-2 m. The glacial sediment elsewhere around
the lake strongly scattered the electromagnetic signal, and showed few coherent reflectors;
diffraction hyperbolae were ubiquitous. The fact that few coherent reflectors were observed
within most sediment at Mirror Lake, along with the ubiquity of diffraction hyperbolae in all
radargrams confirmed the results of earlier research that the glacial sediment under Mirror
Lake is boulder and cobble rich (Likens, 1984). The stratification observed in radargrams B
and C, was, however, unexpected. Those radargrams were conducted over terrain near the
mouth of a canal dug to Mirror Lake during the 19th
century (Likens, 1984). It is possible that
the stratification observed in them reflects deposition of sediment from recent, anthropogenic
activity.
The contact between the smooth, flat gyttja blanket and the diffractor-rich till was
readily identifiable in all radargrams (highlighted in 4-9A). I created a map of the boundaries
of the gyttja by locating the interface between the gyttja and the sides of the lake on each of
the perpendicular-to-shore radargrams, measuring the distance of that interface from shore,
and then plotting a point to represent the location of that interface on a pre-existing map of
Mirror Lake (Figure 4-10). I then drew a line to connect all the points, thereby estimating the
bounds of the gyttja layer. The map I constructed coincided approximately with the map
created by an earlier research group (Moeller, 1975), though there were some discrepancies
(Figure 4-11). Discrepancies are likely the result of both geologic ambiguity and because the
earlier gyttja map was created without GPS data. In cases where the earlier map shows gyttja
blanket approaching the shoreline more closely than the map created in this project, it is

83
likely that thin deposits of organic material on the sides of the lake led the divers who created
the first map to conclude those deposits were gyttja when in fact they were simply thin
organic “skins” or isolated deposits. Similarly, gyttja mapped by radar closer to the shore
may be covered by thin sediment deposits. The sharpness of the contact of the edge of the
gyttja blanket with the sides of the lake as illustrated in the radar surveys eliminated such
ambiguity. Registration of our GPS data against the results of previous location data was
another source of error. The bounds of the lake shown in the 1975 map of sedimentation there
did not line up perfectly with the shape of the lake as it was indicated in 1992 and 2003
orthophotographs of it. I estimate that the error associated with matching our data to previous,
non-GPS derived location data was up to 10 m. It is likely, therefore, that in registering the
1975 map of the gyttja blanket against GPS data we collected there was misalignment which
contributed to the apparent disagreement between the two maps.
The ability to map the location of gyttja has implications for seepage: while head
measurements reveal an upward hydraulic gradient beneath much of the gyttja at Mirror
Lake, seepage likely is at a minimum within zones of gyttja. Rosenberry and Winter (1993)
believe that a pattern of out-seepage followed immediately by in-seepage may be
characteristic of the edge of the gyttja (Figure 4-10inset). This conceptual model has not been
tested thoroughly with seepage meter measurements, but the map provided through radar
surveys would facilitate such testing.

86
CHAPTER 5
CONCLUSIONS
5.1 Electrical Resistivity
Electrical resistivity surveys at Mirror Lake imaged subsurface heterogeneities. In
some cases, as with deposits of organic matter, it was a relatively straightforward matter to
link resistivity variations to geologic variations. In other cases, as where a zone of lower
(1500 Ω –m) average resistivity along the southwest shore of Mirror Lake did not correspond
to any obvious surficial geologic heterogeneities, interpreting resistivity results was more
difficult. There it was possible to use knowledge of Archie’s law to facilitate interpretation:
we were able to eliminate pore fluid conductivity and saturation as variables, thereby pointing
toward variations in porosity as a factor which influenced resistivity along that shore.
Knowledge of the process by which resistivity data are inverted was also useful for
interpreting results. In the case of the northeast inlet area, for instance, knowing that inversion
smoothed transitions between adjacent features and also tended to underestimate resistivity
contrasts allowed us to conclude that a zone of chloride contamination was probably both
smaller and more intense than indicated by the resistivity surveys.
Features the size of or smaller than electrode spacing were unlikely to be imaged in
resistivity surveys. Resolution was poorer for merged datasets, where overlap data did not
always agree and a large number data points were therefore discarded. In the case of the
southwest shore, for instance, where high seepage rates were observed in a region where
resistivity surveys suggested little seepage would occur, we speculated that a heterogeneity
existed which was too small to be resolved by the merged dataset from which that profile was
created.
5.3 Electrical Resistivity Field Techniques

87
Towed- and stationary-cable electrical resistivity surveys had different strengths. The
rapidity of the towed-cable technique made it ideal for generating a preliminary sketch of the
subsurface environment. This first sketch could be used to guide placement of the more time-
consuming stationary-cable surveys, as it was in the case of the northeast inlet area of Mirror
Lake. The 22-m towed-cable was generally more useful than the 44-m cable; the small size of
the lake (0.5 km maximum diameter) made the 44-m cable particularly prone to snagging on
peninsulas and to cross-cutting inlets. At a larger lake or one with longer stretches of straight
shoreline, use of the longer towed-cable would be more feasible.
The stationary-cable surveys were useful because of their higher resolution (4-5 times
that of the towed-cable surveys) and because they could be deployed over terrain not
approachable with the towed-cable survey. In particular, stationary-cables could be used in
the near-shore (< 3m depth) environment, where a majority of seepage occurs, and in coves
and peninsulas, where use of the towed-cable was limited.
Parallel-to-shore stationary-cable resistivity surveys were useful for providing cross-
sectional images of the near-shore environment. Variation with distance from shoreline could
be observed in the series of 4 parallel-to-shore surveys at 2-, 4-, 6-, and 8-m from shore we
conducted along the southwest shore. Perpendicular-to-shore resistivity surveys were useful
for imaging variations with distance from shoreline in more detail. The near-shore
environment, however, was more difficult to image with perpendicular-to-shore surveys
because measurement density is low at the ends of stationary-cable surveys. Steep terrain was
difficult to image with both the parallel- and perpendicular-to-shore cables; because of side-
swipe in the case of the parallel-to-shore surveys and because inversion software could not
accommodate very steep terrain changes sometimes encountered with the perpendicular-to-
shore cables. Further software development could eliminate the latter problem.
In many cases we wished to survey a distance greater than that permitted by a single

88
stretch of the stationary-cable. In such cases we found a 50% overlap between survey lines
was desirable. Precise duplication of electrode placement in these cases minimized error
associated with merging datasets. Flagging of the survey area facilitated such placement.
5.3 Ground Penetrating Radar
GPR surveys at Mirror Lake provided information about the texture and layering of
sediment beneath the lake. The smooth, layered appearance in radargrams of the organic
sediment (“gyttja”) which blankets the bottom of the lake was unique; it allowed us to easily
identify the boundaries of that blanket. Boulders and cobbles had characteristic diffraction
hyperbolae and were also readily mapped with GPR, though depth of penetration into heavily
cobbled areas was only 1-2 meters maximum. The fact that GPR surveys imaged boulders
mapped by previous research (Moeller, 1975) boosted our confidence in the accuracy of both.
Stratification within non-organic sediment was also identifiable in a few radargrams.
It is not clear whether this stratification, observed clearly only along portions of the eastern
shore, is related to outwash processes or to the construction of the canal. The fact that
stratification was observed, however, made us confident that GPR could image it, even in the
scatterer-rich environment of Mirror Lake. No such stratification was observed in radargrams
from within the zone where previous research had indicated a transition from glacial till to
comparatively well-sorted outwash. A small change in degree of sorting, however, would not
create horizontal reflectors and would therefore not be visible with GPR.
5.4 Ground Penetrating Radar Field Techniques
The folding, polyurethane boat we used in this survey was ideal for conducting GPR
surveys of the lake: its nearly flat bottom allowed the GPR antennae unit to be placed directly
on the floor of the boat, and allowed us to approach the shoreline very closely. It also allowed
us to conduct surveys rapidly. Resolution was best in water depths greater than ¼ the
wavelength of the dominant frequency of our antenna (i.e. > 1.5 m). It was worst in water

89
depths closest to that wavelength, and improved slightly in shallower water.
5.5 Using Electrical Resistivity and Radar In Tandem
Electrical resistivity and Ground Penetrating Radar were complementary techniques.
Heterogeneities were imaged in one technique that were not in the other; a property contrast
within the Long Line survey area of the southwest shore was apparent in resistivity surveys
but not in GPR surveys. Stratification within sediment, on the other hand, was apparent in
GPR surveys along the eastern shore where no property contrast was evidenced by resistivity
surveys.
5.6 Position Data
The GPS spatial data we collected at Mirror Lake to tie resistivity, radar and seepage
data to measurement locations were self-consistent. It was difficult, however, to align this
data with that collected by previous research groups: many of those measurements (e.g.
seepage and core analysis) were made prior to the development of or without the use of GPS
location.
The resolution of our own spatial data was also sometimes limited. Occasionally
satellite distribution was such that our GPS unit could not gather enough signals to fix the
coordinates of our location; this condition was perhaps worsened by the low mountains
around the lake. Signal reception was poorest within approximately 10 m of shoreline: the
presence of trees there often inhibited the collection of GPS data. In such cases we
interpolated the missing data during processing, generally by extending a line from data
points collected where signal strength was good towards where it was lost. It is likely that the
ability to gather position data will improve in coming years as GPS technology continues to
advance.
5.7 Suggestions for Further Study
There are several features which we would be interested to investigate in future work

90
at Mirror Lake. First, we would be interested to conduct resistivity surveys using smaller
electrode spacings. Seepage rates varied on a finer scale than was imaged by the 1-m
electrode spacing we used in this project. Modeling experiments suggest that variations in
the upper 0.5 m of sediment exert the greatest control on seepage rates (Mikochik, 2008). An
electrode-cable with 0.5 m spacing or smaller could potentially image the small-scale features
which control seepage rates. Work with such a cable is planned for the summer of 2008, as
part of the next phase of the project of which this thesis was a part.
A radar antenna which could effectively be used in the very near-shore (< 3m water
depth) could potentially yield useful results since the near-shore environment tends to have
the highest seepage rates. A waterproof, shielded GPR unit, with which the radar antennae
could be coupled directly with the lake bottom, would serve this purpose well. A small raft,
with which the antenna could be pulled in the shallowest water, might also serve the purpose;
reverberation was worst in water depth approximately equal to dominant wavelength, and
was slightly better in water shallower than that. It would be particularly interesting to gather a
parallel-to-shore GPR survey; bedding features not apparent in perpendicular-to-shore
surveys might be obvious in parallel-to-shore surveys.
Finally, it would be useful if GIS data for Mirror Lake were compiled and posted to a
central location. There exists vast quantity of data about the region, collected by diverse
research groups and across many disciplines. A central database would facilitate comparison
of data from these different studies.
5.8 Implications for understanding groundwater-lake interaction
Our work at Mirror Lake highlighted the versatility of correlating electrical resistivity
surveying to seepage. Variations in resistivity along the southwest shore of Mirror Lake
probably resulted from changes in porosity, which at that site appear to relate to the hydraulic
conductivity of sediment. Variations in resistivity within the northeast inlet were the product

91
of changes in pore-fluid conductivity and were probably indicative of road salt
contamination. Both have implications for characterizing underwater seepage: the former in
terms of its ability to suggest the geologic variations which control seepage, the latter in
terms of evidence of flow paths within a lake. Radar was able to image location of the gyttja
boundary, which previous work has shown to be related to seepage rates. Radar also showed
stratification within sediment, which might in some cases have implications for seepage,
though not in this case. Radar surveys can highlight features which resistivity cannot, and
vice versa. Together or separately these techniques can highlight geologic heterogeneities and
guide point measurement of seepage.
Our work at Mirror Lake confirms the efficacy of electrical resistivity as a tool which
can image geologic heterogeneity which controls seepage and can guide point measurement
of seepage. The controls on electrical resistivity and seepage rates were different at Mirror
Lake than they were at Lake Michigan where the first work relating resistivity profiles to
seepage rates was conducted (Taylor and Cherkauer, 1984): clay content in sediment at Lake
Michigan was high and represented a dominant control on resistivity and also on seepage
rates. Clay content at Mirror Lake was low; organic matter content, road salt contamination
and porosity represented primary controls on resistivity. The first and third of these (organic
matter and porosity) likely represented major controls on seepage at Mirror Lake. The
substantial technological advances which have evolved in the time which has passed since
Taylor and Cherkauer (1984) conducted their pioneering study have made the electrical
resistivity survey technique more powerful and more versatile. We hope that further study at
Mirror Lake, scheduled for the summer of 2008, will deepen our understanding of the
geophysical signatures of seepage.

92
REFERENCES
Advanced Geosciences Incorporated company website, SuperSting R8 Marine Logging
System instrument manual. www.agiusa.com/files/eicust/earthimageruser.shtml#EI2D.
PDF accessed 10/07.
Asbury, Clyde Ellis, 1990. The role of groundwater seepage in sediment chemistry and
nutrient budgets in Mirror Lake, New Hampshire. PhD dissertation, Cornell University,
Ithaca, NY, pp1 - 275.
Archie, G., 1942. The electrical resistivity log as an aid in determining some reservoir
characteristics: Journal of Petroleum Technology, 5, pp 1-8.
Ball, L. B., Wade, H. K, Cannia, J. C. 2006. Determination of canal leakage potential
using continuous resistivity profiling techniques in western Nebraska and eastern
Wyoming. Proceedings of the 2008 Symposium on the Application of Geophysics to
Engineering and Environmental Problems (SAGEEP), April 2-6, Seattle, Washington:
Environmental and Engineering Geophysical Society, pp 851-862.
Brunke, M. and Gonser, T. 1997. The ecological significance of exchange processes
between rivers and groundwater. Freshwater Biol., 37, pp 1-33.
Cherkauer, D.S., 1991. Geophysical mapping of pathways for entry of contaminated
groundwater to lakes and rivers: Application in the North American great lakes.
Hydrological Science and Technology American Institute of Hydrology, 7, no. 1-4, pp
35-44.
Cherkauer, D. S. and Carlson, D. A., .1997. Interaction of Lake Michigan with a Layered
Aquifer Stressed by Drainage. Ground Water, Vol. 35, Issue 6, pp. 981-989.
Complex Systems Research Center, University of New Hampshire, Durham, 1992.
Digital Orthophoto quadrangle, 1:12,000 series, Quad 73, Woodstock, New Hampshire.
From University of New Hampshire GIS Clearinghouse, www.granit.unh.edu accessed
11/06.
Day-Lewis, F.D., Singha, K., Binley, A., 2005. Applying petrophysical models to radar
travel time and electrical resistivity tomograms: Resolution-dependent limitations.
Journal of Geophysical Research, Vol. 110, B08206, pp 1-17.
Day-Lewis, F.D., White, E.A., Johnson, C.D., and Lane, J.W. Jr., 2006. Continuous
resistivity profiling to delineate submarine groundwater discharge – examples and
limitations. The Leading Edge, June, 2006, pp 724-728.
Ellefsen, K. J., Hsieh, P. A., Shapiro, A.M., 2002. Crosswell seismic investigation of
hydraulically conductive, fractured bedrock near Mirror Lake, New Hampshire. Journal
of Applied Geophysics, 50, pp 299-317.
Fogg, G. E., Noyes, C.D., and Carle, S. F., 1998. Geologically based model of
heterogeneous hydraulic conductivity in an alluvial setting. Hydrogeology Journal, 5, pp.
131-143.

93
Genereux, D., and I. Badopadhay, 2001. Numerical investigation of lake bed seepage
patterns: effects of porous medium and lake properties. Journal of Hydrology, 241, pp
286-303.
Harte, P. T., 1997, Preliminary assessment of the lithologic and hydraulic properties of
the glacial drift and shallow bedrock in the Mirror Lake area, Grafton County, New
Hampshire, Open-File Report, U.S. Geological Survey, Pembroke, NH.
Hayashi, M., and Rosenberry, D. R., 2002, Effects of groundwater exchange on the
hydrology and ecology of surface water, Groundwater, Vol. 40, Issue 3, pp 309-316.
Heaney, M. J., Nyquist, J.E., Toran, L.E., (2006). Marine Resistivity as a Tool for
Characterizing Zones of Seepage at Lake Lacawac, PA. In Abstracts with Programs -
Geological Society of America, October 2006, Vol. 38, Issue 7, pp 42.
Heaney, M. J., 2007. Marine resistivity as a tool for characterizing groundwater discharge
zones at Lake Lacawac, PA. M.S. Thesis; Temple University, Philadelphia, PA, pp 1 –
108.
Johnson, C. D., 1999. Effects of lithology and fracture characteristics on hydraulic
properties in crystalline rocks; Mirror lake research site, Grafton County, New
Hampshire. Water Resource investigations. U.S. Geol. Surv., 1999, pp 795-802
Kalbus, E., Reinstorf, F., and M. Schirmer, 2006. Measuring methods for groundwater-
surface water interactions: a review. Hyrdrol. Earth Syst. Sci., Vol. 10, pp 873-887.
Krabbenhoeft, D.P. and Anderson, M.P, 1986, Use of a numerical ground-water flow
model for hypothesis testing, Groundwater, Vol. 24, No. 1, pp 49-55.
Lee, D. R., 1977. A Device For Measuring Seepage Flux in lakes and Estuaries.
Limnology and Oceanography, Vol. 22, No. 1, pp140-147.
Likens, G. E. ed., 1984. An Ecosystem Approach to Aquatic Ecology: Mirror Lake and
its environment. Springer-Verlag, New York, pp 1- 324.
Manheim, F. T., Krantz, D.E., Bratton, J. F., 2004. Studying Groundwater Under
Delmarva Coastal Bays Using Electrical Resistivity. Groundwater, Vol. 42, No. 7, pp
1052-1068.
Marescot, L. and M.H. Loke. 2003. Using the depth of investigation index method in 2D
resistivity imaging for civil engineering surveys. Proceedings of the 2004 Symposium on
the Application of Geophysics to Engineering and Environmental Problems (SAGEEP),
February 10-14, Las Vegas, Nevada, Paper 13 GSL, Environmental and Engineering
Geophysical Society, Denver, CO, pp 540-547.
McBride, M. S. and H.O. Pfannkuch, 1975. The Distribution of Seepage Within
Lakebeds. Jour. Research U.S. Geol. Survey., Vol. 3, No. 5, Sept-Oct., 1975, pp 505-512.
Mikochik, J., 2008. Need to get rest of reference!

94
Moeller, R. E., 1975. Hydrophyte biomass and community structure in a small,
oligotrophic New Hampshire lake. Verh. Internat. Verein. Limnol., Vol. 19, pp1004–
1012.
National Agriculture Imagery Program, 2003. 15-Minute Quad series, Plymouth-Grafton
counties, New Hampshire. From University of New Hampshire GIS Clearinghouse,
www.granit.unh.edu accessed 6/20/08.
Neal, A., 2004. Ground-penetrating radar and its use in sedimentology: principles,
problem and progress. Earth-Science Reviews, Vol. 66, pp 261-330.
Oldenburg D.W. and Li Y. 1999. Estimating depth of investigation in DC resistivity and
IP surveys. Geophysics, Vol. 64, pp 403-416.
Paillet, F. L., 1988. Characterizing the circulation of groundwater through fractures at
Mirror Lake, New Hampshire. Abstracts with Programs, Geological Society of America,
1988, Vol. 20, Issue 7, pp. 311.
Panthulu, R.V., Krishnaiah, C., and Shirke, J.M., 2001. Detection of seepage paths in
earthen dams using self-potential and electrical resistivity methods. Engineering
Geology, 59, pp. 281-295.
Peake, J. A., 2005. Comparison of electrical resistivity techniques to characterize karst
geology, Easton, PA. M.S. Thesis. Temple University, Philadelphia, PA, pp 1 – 119.
Reeve, R.C., and M.C. Jensen. 1949. Piezometers for groundwater flow studies and
measurement of subsoil permeability. Agric. Eng., Vol. 30, pp 435-438.
Reynolds, J. M., 1997. An Introduction to Applied and Environmental Geophysics. John
Wiley & Sons, NY, 1997.
Ritter, D. F., Kochel, R. C., and J. R. Miller, 1995. Process Geomorphology. Wm. C.
Brown Publishers, Dubuque, IA, pp 334-352.
Rosenberry, D.O. and Winter, T.C., 1993. The significance of fracture flow to the water
balance of a lake in fractured crystalline rock terrain. Memoires of the 24th
congress of
International Association of Hydrologists, Oslo, pp 967 -977.
Rosenberry, D. O., Bukaveckas, P.A., Buso, D.C., Likens. G.E., Shapiro, A.M. and
Winter, T.C.,1999, Movement of road salt to a small New Hampshire lake, Water, Air
and Soil Pollution, Vol. 109, Issue 1-4, pp. 179-206.
Rosenberry, D.O., 2005. Integrating seepage heterogeneity with the use of ganged
seepage meters. Limnology and Oceanography: Methods. 3, 2005, 131-142.
Scheutz, J.W., 2002, Numerical modeling of seasonal variability in groundwater flow
near Mirror Lake, Grafton county, NH, M.A. Thesis, Department of Geology, SUNY
Buffalo., pp 1 - 120
Schneider, R. L., 2005. Factors influencing groundwater seepage in a large, mesotrophic
lake in New York. Journal of Hydrology, August 01, 2005, Vol. 310, Issue 1-4, pp.1-16.

95
Sebestyen, S. D., and Schneider, R. L., 2001. Dynamic temporal patterns of nearshore
seepage flux in a headwater Adirondack lake. Journal of Hydrology, Vol. 247, pp. 137-
150.
Shaban, A., Mohamad K., Chadi A., and G. Faour, 2005. Geologic controls of submarine
groundwater discharge: application of remote sensing to north Lebanon. Environmental
Geology, Vol. 47, pp 512-522.
Snyder, D. D. and Wightman, A., 2002. Application of Continuous Resistivity Profiling
to Aquifer Characterization. Proceedings of the 2002 Symposium on the Application of
Geophysics to Engineering and Environmental Problems (SAGEEP), February 10-14,
Las Vegas, Nevada, Paper 13 GSL. Environmental and Engineering Geophysical Society,
Denver, CO.
Sophocleous, M., 2002. Interactions between groundwater and surface water: the state of
the science. Hydrogeology Journal (2002), 10. pp 52-67.
Taylor, R. W., 1982. The application of combined seismic and underwater resistivity
measurements to hydrological studies in Lake Michigan. Eos, Transactions, American
Geophysical Union, May 04, 1982, Vol. 63, Issue 18, pp 321.
Taylor, R. W. and D. S. Cherkauer, 1984. The application of combined seismic and
electrical measurements to the determination of the hydraulic conductivity of a lake bed,
Groundwater Monitoring Review, Vol. 4, Issue 4, pp 78-85.
Tiedeman, C.R., Goode, D.J and Hsieh, P.A. 1997, Numerical Simulation of
Groundwater flow through glacial deposits and crystalline bedrock in the Mirror Lake
Area, Grafton County, NH, U.S. Geological Survey Professional Paper 1572, 50 pp.
Toran, L.E., Weeks Chair in Environmental Geology, Temple University, Philadelphia,
PA.
Wilson, A., 1991, Distribution of hydraulic conductivity in the glacial drift at Hubbard
Brook Experimental Forest, West Thornton, New Hampshire, Hanover, N.H., Dartmouth
College, Senior honors thesis, 56 pp.
Winter, T. C., 1989. Evaluation of inflow to Mirror Lake, New Hampshire. Water
Resources Bulletin, October 1989, Vol. 25, Issue 5, pp. 991-1008.
Winter, T. C., 2000. Geologic controls on the interaction of ground water and surface
water. In: Abstracts with Programs - Geological Society of America, 2000, Vol. 32, Issue
7, pp58.
Winter, T. C., 1984. Mirror Lake and Its Watershed: Physiographic Setting and Geologic
Origin of Mirror Lake. In: Likens G. E., (Ed), 1985. An Ecosystem Approach to Aquatic
Ecology: Mirror Lake and Its Environment. Springer-Verlag: New York, NY, pp 40-83.
Winter, T.C., 1976. Numerical simulation analysis of the interaction of lakes and
groundwater: U.S. Geol. Surv. Prof. Pap. 1001, pp 1-45.

96
APPENDIX A
Southwest Shore Bathymetry, Gathered From Sonar Data.
Acquisition described in Section 3.10.

NM_Thesis

More Related Content

Viewers also liked (14)

Similar to NM_Thesis (20)

NM_Thesis