CHARACTERIZING, MAPPING, AND INTERPRETING THIN LOESS DEPOSITS IN THE
WESTERN UPPER PENINSULA OF MICHIGAN
By
Michael David Luehmann

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTERS OF SCIENCE
Geography
2011

ABSTRACT
CHARACTERIZING, MAPPING, AND INTERPRETING THIN LOESS DEPOSITS IN THE
WESTERN UPPER PENINSULA OF MICHIGAN
By
Michael David Luehmann
This research examines the distribution, thickness and textural characteristics of thin,
patchy, loess deposits in the western Upper Peninsula of Michigan. Although local soil surveys
have documented loess in Baraga, Marquette and Iron Counties, at the northeastern margins of
the North American, midcontinent loess region, this thesis is the first detailed study of these
loess deposits. Within the study area, loess is ~ 30-60 cm thick and is usually underlain by sandy
glacial deposits. Where the loess deposits are thin, it appears that pedoturbation processes have
likely mixed some of the lower, sandy materials into the loess. For this reason, most of the loess
has a bimodal continuous textural curve, with a primary modal particle-size within the 25-75 µm
fraction (the loess) and a secondary mode in the 250-500 µm fraction (the in-mixed sand). These
loess deposits were likely sourced from both distant and local areas, including ground moraines,
outwash plains, and the floodplains of small meltwater streams. Within the Peshekee Loess
region, I have identified four loess sections, each of which has unique characteristics that set it
apart from the whole: the Amasa, Covington, Republic, and Champion sections. This research
(1) recognizes multiple source areas for loess, (2) develops and interprets textural data on the
thin Peshekee Loess, and (3) documents the effects of mixing within these loess deposits. This
research is the first to document both the extent and textural characteristics of loess deposits
situated at the extreme margins of much larger and thicker, regional-scale, loess region.

Copyright by
MICHAEL DAVID LUEHMANN
2011

ACKNOWLEDGEMENTS

I owe my deepest gratitude to my advisor Dr. Randy Schaetzl, for always having his door
open, listening, and truly leading by example. His guidance and confidence has allowed me to
become a better student, writer, and scientist. I look forward to our future research collaborations
as I work towards my PhD here at Michigan State University.
I am also greatly indebted to my committee members: Dr. Alan Arobgast and Dr. David
Lusch for their constructive suggestions during the formulation of this thesis. I am grateful of the
various off-the-record conversations with Al, which provided the courage to accomplish this
research. I am also grateful to have Dr. Lusch as a committee member; he helped me to learn the
vocabulary needed to describe geomorphology and could knowledgeably answer any question I
had about the evolution of Michigan’s landscape.
A special word of thanks is extended to Dr. Anderton for introducing me to my passion in
soils and advising me at Northern Michigan University. I also wish to recognize and thank Larry
Carey, William Anzalone, Ken Wikgren, and Mark Farina for taking me on as an Earth Team
Member and allowing me to help construct a “once-over” soil survey for Isle Royale.
This thesis has also benefitted from the contributions of many colleagues, friends, and
family members. In particular I gratefully acknowledge Kristine Stanley, who I have never met
in person but in many ways “paved the road” for this project, Claire Forgacs for proof reading
my research proposal, Mike Bigsby for allowing me to bounce ridiculous ideas about loess off
him, Brad Miller for always having the answers (specifically in dealing with Excel and ArcGIS),
and the Johnson family for mentoring me throughout my academic career and allowing me to
stay at their house during fieldwork. Sherri Johnson is an outstanding editor and wonderful cook;
her cooking often provided the fuel I needed for collecting the thesis data. I would also like to
iv

thank Dwight Johnson (a.k.a., DJ) for keeping the house stocked with food, providing me with a
car while mine was in the shop, and for helping me carry soil samples to the attic, so they could
be dried. I am truly grateful of friend and partner, Natalie Johnson; she continually supported and
assured me that I could overcome my disability, and thus, deserves to be co-author of this
project. Lastly, I am thankful for the invaluable support from my parents’, Christina and David
Luehmann for never asking me “why” but instead have pillared my life decisions and provided
help along the way.
Funding for fieldwork was provided by the National Science Foundation, Geography and
Spatial Sciences Program, under grant 0851108 made to Dr. Randall Schaetzl and colleagues, a
graduate fellowship from the Department of Geography at Michigan State University, a
scholarship from the Soil Classifiers Association of Michigan, and the Huron Mountain Wildlife
Foundation.

v

TABLE OF CONTENTS

List of Tables ............................................................................................................................... viii
List of Figures ................................................................................................................................. x
1. Introduction ................................................................................................................................. 1
2. Literature Review........................................................................................................................ 4
2.1 Background and Significance................................................................................................ 4
2.2 Loess Production ................................................................................................................... 5
2.3 Loess Transportation and Deposition .................................................................................... 5
2.4 Preservation ........................................................................................................................... 6
2.5 Spatial Characteristics of Loess Deposits ............................................................................. 7
2.6 Loess Chronology in North America .................................................................................... 9
2.7 Loess in Michigan and nearby areas ................................................................................... 11
2.8 Summary ............................................................................................................................. 14
3. Study Area ................................................................................................................................ 15
3.1 Extent of the Broad Study area ........................................................................................... 15
3.2 Bedrock Geology................................................................................................................. 17
3.3 Glacial History .................................................................................................................... 20
3.4 Physiography and relief ....................................................................................................... 25
3.5 Climate and vegetation ........................................................................................................ 34
3.6 Soils ..................................................................................................................................... 36
3.6.1 Spatial distribution of surface textures and loess cap thicknesses................................ 36
3.6.2 Major loessal soils and comparing their diagnostic horizons ....................................... 43
3.7 Summary of Broad Study Area ........................................................................................... 50
4. Methods..................................................................................................................................... 50
4.1 Field Methods...................................................................................................................... 50
4.2 Laboratory Methods ............................................................................................................ 53
4.3 Post Processing.................................................................................................................... 56
5. Results and Discussion ............................................................................................................. 56
5.1 Data Set “Clean-up” ............................................................................................................ 57
5.2 Characteristics of Loess within the Broad Study Area ....................................................... 64
5.2.1 Eolian Textural Curve Type Categories ....................................................................... 64
5.2.1.1 Type 1: Unimodal Silt Curves ................................................................................ 64
5.2.1.2 Type 2: Bimodal Silt Curves................................................................................... 67
vi

5.2.1.3 Type 3: Bimodal Sand Curves ................................................................................ 71
5.2.1.4 Type 4: Silt Shoulder Curves .................................................................................. 73
5.2.1.5 Type 5: Sand Shoulder Curves ............................................................................... 76
5.2.1.6 Distribution of Sites with Different Textural Curves ............................................. 79
5.3 Spatial Characteristics of Loess within the Study Area ...................................................... 86
5.3.1 Loess Thickness Data ................................................................................................... 86
5.3.2 Particle Size Analysis of the Peshekee Loess ............................................................... 88
5.3.2.1 Spatial Trend in Silt Contents ................................................................................. 88
5.3.2.2 Spatial Trends in Sand Contents ............................................................................. 94
5.3.2.3 Spatial Trends in Modal Particle Size ................................................................... 101
5.4 Peshekee Loess Individual Sections and Potential Source Areas ..................................... 107
5.4.1 Amasa Section ............................................................................................................ 112
5.4.2 Covington Section ...................................................................................................... 120
5.4.3 Republic Section ......................................................................................................... 126
5.4.4 Champion Section....................................................................................................... 129
5.5 Summary ........................................................................................................................... 134
6. Conclusions ............................................................................................................................. 135
References ................................................................................................................................... 138

vii

LIST OF TABLES

Table 3.1:

Description of the Yellow Dog Plains and Baraga Plains physiographic regions
(after Schaetzl et al. (in prep)). ............................................................................. 28

Table 3.2:

Description of the Peshekee Highlands physiographic region (after Schaetzl et al.
(in prep)). .............................................................................................................. 29

Table 3.3:

Description of the Michigamme Bedrock Terrain, Gwinn Sandy Terrain, and Iron
Mountain Bedrock Uplands physiographic regions (after Schaetzl et al. (in prep)).
............................................................................................................................... 32

Table 3.4:

Characteristics of loess with cobbly silt loam surface textures
(http://soildatamart.nrcs.usda.gov)........................................................................ 38

Table 3.5:

Characteristics of loess with Cobbly fine sandy loam surface textures
(http://soildatamart.nrcs.usda.gov)........................................................................ 39

Table 3.6:

Characteristics of loess with silt loam surface textures
(http://soildatamart.nrcs.usda.gov)........................................................................ 39

Table 3.7:

Comparing the diagnostic character between the Champion and Keewaydin soil
series (http://soildatamart.nrcs.usda.gov). ............................................................ 46

Table 3.8:

Comparing the diagnostic character between the Champion, Michigamme and
Dishno soil series (http://soildatamart.nrcs.usda.gov). ......................................... 47

Table 3.9:

Comparing the diagnostic character between the Champion, Petticoat, Goodman,
and Wabeno soil series (http://soildatamart.nrcs.usda.gov).................................. 48

Table 5.1:

Characteristics of type 0 eolian textural curves. ................................................... 61

Table 5.2:

Characteristics of type 1 eolian textural curves. ................................................... 66

Table 5.3:

Characteristics of type 2 samples. ......................................................................... 70

Table 5.4:

Characteristics of type 3 samples. ......................................................................... 72

Table 5.5:

Characteristics of type 4 samples. ......................................................................... 74

Table 5.6:

Characteristics of type 5 samples. ......................................................................... 77

Table 5.7:

Characteristics of the four major loess sections within the study area. .............. 109

viii

Table 5.8:

A table ranking the four loess sections, within the study area, by calculating each
of the sections mean loess thickness and the mean of several particle-size
fractions............................................................................................................... 113

ix

LIST OF FIGURES

Figure 3.1:

Generalized map of soils with loess parent materials in western Upper Michigan
(after Berndt 1988; Linsemier 1989; Schwenner 2007). Red represents soils series
mappped with loess thicknesses of 51 - 100 cm. Pink represents soil series
mapped with loess thicknesses of 25 - 50 cm. Soils with loamy eolian sediment
parent materials are colored purple. The broad study area outline includes parts of
two loess regions named by Scull and Schaetzl (2011). For interpretation of the
references to color in this and all other figures, the reader is referred to the
electronic version of this thesis. ............................................................................ 16

Figure 3.2:

Generalized bedrock map of Michigan’s western Upper Peninsula and an outline
of the broad study area’s common rock episodes, along with examples of formal
rock unite names and dominant rock types (after Sims 1993; Cannon 2007;
Bornhorst and Brandt 2009).................................................................................. 18

Figure 3.3:

Extent of the last four major ice lobes that sculpted the glacial landscape of the
broad study area (Attig et al. 1985; Peterson 1986).............................................. 21

Figure 3.4:

A generalized map of the moraines or ice-margin positions in the broad study area
(after Attig et al. 1985).......................................................................................... 22

Figure 3.5:

Major physiographic regions within the broad study area (Schaetzl et al. (in
prep.)). ................................................................................................................... 26

Figure 3.6:

Photographs of the Mulligan Plains (A) and Yellow Dog Plains (B), both lowrelief, excessively well drained, sandy outwash plains, interspersed with poorly
drained soils. Photograph (A) by M. Luehmann and (B) by R. Schaetzl. ............ 28

Figure 3.7:

Photographs at the base of a resistant, high-relief, Precambrian bedrock knob
within the Peshekee Highlands. Photograph (A) by L. Safford, and (B) by R.
Schaetzl. ................................................................................................................ 29

Figure 3.8:

Local relief variation within the broad study area, and the physiographic region
boundaries. ............................................................................................................ 30

Figure 3.9:

Upper solum parent materials within the broad study area (after Berndt 1988;
Linsemier 1989; Schwenner 2007). ...................................................................... 33

x

Figure 3.10: Presettlement vegetation of the broad study area (after Comer et al. 1995). ........ 35
Figure 3.11: A map of the surface textures of the seven dominant loessal soil series in the
broad study area (Berndt 1988; Linsemier 1989; Schwenner 2007). ................... 40
Figure 3.12: Map of loess thickness, based on the seven dominant loessal soil series in the
broad study area (Berndt 1988; Linsemier 1989; Schwenner 2007). ................... 42
Figure 3.13: A comparison of the similarities and differences between the Champion series, the
most abundant mapped loessal soil series within the broad study area, and the
Michigamme, Dishno, and Keewaydin series. (http://soildatamart.nrcs.usda.gov).
............................................................................................................................... 44
Figure 3.14: A comparison of the similarities and differences between the Champion series, the
most abundant mapped loessal soil series within the broad study area, and the
Petticoat, Goodman, and Wabeno series. (http://soildatamart.nrcs.usda.gov)...... 45
Figure 4.1:

(A) Map illustrating the distribution of loessal soils on uplands within the broad
study area, as indicated by the NRCS. Green dots are locations where loess
samples were taken. (B) Sampling loess at a target location using a standard handheld auger. Photograph by R. Schaetzl. ................................................................ 51

Figure 4.2:

The distribution of loess samples across the broad study area, including a few
points taken north and south of the study area. ..................................................... 55

Figure 5.1:

Samples curves that were eliminated from the kriging data set, as a result of not
having a type 1-5 ETC. See Table 5.1 for more details. ...................................... 60

Figure 5.2:

Differences between type 0 and type 5 eolian textural curves. Mean values are
shown. ................................................................................................................... 61

Figure 5.3:

The spatial distribution of ETC types 0-5 throughout the broad study area. ........ 62

Figure 5.4:

The continuous textural curves of the 30 samples assigned a type 1 ETC, with the
blue line representing the mean textural curve. See Table 5.2 for more details. .. 66

Figure 5.5:

The continuous textural curves of two samples taken from site 870; the yellow
line showing the continuous textural curve of the loess and blue showing the
sample’s continuous texture at depth. ................................................................... 69

xi

Figure 5.6:

The continuous textural curves of the samples assigned a type 2 ETC, with the
blue line representing the mean textural curve. See Table 5.3 for more details. .. 70

Figure 5.7:

The continuous textural curves of the samples assigned a type 3 ETC, with the
blue line representing the mean textural curve. See Table 5.4 for more details. .. 72

Figure 5.8:

The continuous textural curves of the samples assigned a type 4 ETC, with the
blue line representing the mean textural curve. See Table 5.5 for more details. .. 74

Figure 5.9:

The continuous textural curves of two samples taken from site 1086; the yellow
line is the continuous textural curve of the loess and blue is the continuous
textural curve for the sediment at depth. ............................................................... 75

Figure 5.10: The continuous textural curves of the samples assigned a type 5 ETC, with the
blue line representing the mean textural curve. See Table 5.6 for more details. .. 77
Figure 5.11: Summary diagram of the five ETC types and their mean continuous textual
curves. ................................................................................................................... 78
Figure 5.12: The distribution of soil (loess) samples taken within the broad study area,
symbolized by their eolian textural curve type. .................................................... 80
Figure 5.13: The distribution of ETC types plotted against a background of local relief, within
the broad study area. ............................................................................................. 81
Figure 5.14: The distribution of eolian textural type curves with generally high silt contents
and low sand contents (labeled in red), and high sand contents and low silt
contents (labeled in green). ................................................................................... 82
Figure 5.15: The distribution of eolian textural type curves with high silt and high sand
contents, plotted against relief within the broad study area. ................................. 84
Figure 5.16: Interpolated loess thickness map for the study area.............................................. 87
Figure 5.17: Interpolated map of clay-free silt contents in the study area. ............................... 89
Figure 5.18: Interpolated map of clay-free fine silt contents in the study area. ........................ 90
Figure 5.19: Interpolated map of clay-free medium silt contents in the study area. ................. 91

xii

Figure 5.20: Interpolated map showing the distribution of clay-free coarse silt contents. ....... 92
Figure 5.21: Interpolated map showing the distribution of clay-free fine, very fine sand
contents. ................................................................................................................ 95
Figure 5.22: Interpolated map showing the distribution of clay-free coarse, very fine sand
contents. ................................................................................................................ 96
Figure 5.23: Interpolated map showing the distribution of clay-free fine, fine sand contents. . 98
Figure 5.24: Interpolated map showing the distribution of clay-free coarse, fine sand contents.
............................................................................................................................... 99
Figure 5.25: Interpolated map showing the distribution of clay-free medium sand contents. 100
Figure 5.26: As an example, this type 2 sample taken from the study area was assigned a 48
µm silt mode and a 300 µm sand mode, for the two modal maps. ..................... 102
Figure 5.27: As an example, this type 3 sample taken from the study area was assigned a 48
µm silt mode and a 415 µm sand mode, for the two modal maps. ..................... 103
Figure 5.28: A map illustrating modal particle size of eolian sediment in the study area, using
sample modes within the 26-99 µm fraction. Samples without a “true” mode
within the 26-99 µm fraction, such as type 5 ETCs, were not used in this kriging
exercise. .............................................................................................................. 104
Figure 5.29: A map illustrating modal particle size of coarser sediment in loess samples, within
the study area. Samples without a “true” mode in the 200-780 fraction, such as
types 1 and 4 ETCs, were not used in this kriging exercise. .............................. 105
Figure 5.30: Map of interpolated clay-free silt contents and a comparison of loess samples’
textural curves within four loess sections, in the study area. .............................. 110
Figure 5.31: Map of interpolated clay-free fine, very fine sand (50-75 um) contents and a
comparison of loess samples’ textural curves within four loess sections, in the
study area. ........................................................................................................... 111
Figure 5.32: An example of four continuous textural curves from the Amasa section. Note that
the textural curves are very similar and all have a mode within the coarse silt
fraction (35-50 um). ............................................................................................ 114
xiii

Figure 5.33: Ice-margin positions in the western Upper Peninsula of Michigan and loess
sections within the study area. The ice-margins were determined using the work
of Attig et al. (1985), Peterson (1985, 1986) and various NRCS county-level soil
surveys. ............................................................................................................... 117
Figure 5.34: Map of interpolated loess thickness, overlain on the NRCS soil surface texture
map. The four smaller loess sections are shown in relation to the potential source
areas. ................................................................................................................... 118
Figure 5.35: Map of interpolated distribution of clay-free silt contents, overlain onto the NRCS
soil surface texture map. The four smaller loess sections are shown in relation to
the potential source areas. ................................................................................... 119
Figure 5.36: An example of four continuous textural curves from the Covington section. Note
that the textural curves are very similar and all have a mode within the coarse silt
fraction (35-50 µm). ............................................................................................ 121
Figure 5.37: The unstable, deeply incised Ontonagon Clay Plains, showing V-shaped valleys
and silt loam and clay loam upper soil textures. ................................................. 123
Figure 5.38: Map showing loess thicknesses and the anomalously thick loess deposits that are
located within the Covington section, just southeast of the Baraga Plains. ........ 124
Figure 5.39: The Baraga Plains, a low-relief outwash plain with scattered sand dunes.
Photograph by R. Schaetzl. ................................................................................. 125
Figure 5.40: Examples of continuous textural curves from the Republic section. Note that the
textural curves are very similar and all have a mode within the fine, very fine sand
fraction (50-75 um). ............................................................................................ 127
Figure 5.41: Examples of continuous textural curves from the Champion section. Note that the
textural curves of loess deposits within this section generally vary considerably
across short distances, and that the modal particle-size of these loess deposits is
generally within the fine, very fine sand fraction (50-75 µm). ........................... 131
Figure 5.42: Map showing the anomalously thick loess deposits that are located near the
Peshekee River, within the Champion section. ................................................... 132
Figure 5.43: Peshekee River terrace and terrace sediments. Photograph by R. Schaetzl........ 133
Figure 5.44: The Peshekee River - an underfit stream. Photograph by R. Schaetzl. .............. 133
xiv

1. Introduction
Loess, wind-blown silt, is found across vast parts of China, Central Asia, Europe, New
Zealand, Alaska, and on both the Great Plains and Central Lowlands of North America (Smalley
1975; Mason et al. 1999; Bettis et al. 2003; Roberts et al. 2003). In North America, there are five
distinct areas of thick, extensive loess deposits: (1) Alaska, (2) eastern Washington and the
adjacent uplands in Oregon (3) Idaho, (4) the Great Plains, and (5) the Central lowlands. At
small scales, loesses of the Great Plains (Nebraska, Kansas, and Colorado) and Central Lowlands
(Missouri, Iowa, Illinois, Indiana, and areas to the north and south) appear to be a continuous
sheet, however at large scales it is apparent that these loess bodies have very different thickness
trends and origins, and are interrupted by areas where loess is absent (Muhs and Bettis 2003).
Loess in these areas tends to be associated with glacial episodes and provides a critical record of
environmental change during and after the last glacial phase in North America (Muhs and Bettis
2003). Interpretation of this record of environmental change requires accurately linking
stratigraphic sequences (and deposits) of loess to their source areas, along with understanding the
spatial characteristics of the loess itself (Mason et al. 1999; Muhs and Bettis 2000; 2003;
Schaetzl and Hook 2008; Scull and Schaetzl 2011).
Loess deposits, regardless of origin, require three conditions in order to form: (1) a
persistent sediment source, (mainly silt size and fine sand particles, that are loose, usually on an
unvegetated surface, and able to be deflated), (2) a sustained/suitable wind direction and wind
velocity, and (3) a site for sediment accumulation (Pye 1995). Loess deposits tend to be thickest
near their source and become progressively thinner downwind, providing evidence of paleowind
directions (Wascher et al. 1947; Frazee et al. 1970; Rutledge et al. 1985; Fehrenbacher et al.
1986; Leigh 1994; Pye 1995 Schaetzl and Hook 2008, Stanley and Schaetzl 2011). In the
Midwestern United States, loess can exceed 30 meters in thickness (Smith, 1942; Olson and
1

Ruhe 1979; Fehrenbacher et al. 1986), but generally become thinner to the east and northeast,
from the Mississippi River valley, until, by Michigan they may only be as thin as 30 cm, and
often are thinner or absent (Scull and Schaetzl 2011). Relatively thin loess deposits blanketing
much of the Great Lakes region have, until recently, largely been ignored (e.g. Schaetzl 2008;
Schaetzl and Hook 2008, Schaetzl and loope 2008; Hobbs et al. 2011; Scull and Schaetzl 2011;
Stanley and Schaetzl 2011). Often, these thinner and smaller loess deposits do not have an
obvious source because they do not lie adjacent to broad meltwater river valleys (e.g., the
Mississippi and Missouri River valleys) that, in the Midwest, have traditionally been assumed to
be the main sources for loess. On the contrary, loess deposits in Michigan have mainly been
linked to nontraditional loess source areas, like outwash plains, moraines, lacustrine plains, and
mid-size glacial meltwater valleys (Schaetzl 2008; Schaetzl and Hook 2008; Schaetzl and Loope
2008; Hobbs et al. 2011; Stanley and Schaetzl 2011; Scull and Schaetzl 2011).
The Natural Resource Conservation Service (NRCS) county-level soil surveys identify a
number of soils within Michigan’s western Upper Peninsula that have formed in “modified
eolian material and underlying loamy and sandy glacial drift” (Berndt 1988; Linsemier 1989;
Schwenner 2007). Most researchers that have worked in this area have concluded that this thin,
“mantling” material is, indeed, loess (Flint 1971; Berndt 1988; Linsemier 1989; Schwenner
2007). To that end, Scull and Schaetzl (2011) identified (at small-scale) multiple loess areas or
“sheets” throughout Wisconsin and Michigan’s western Upper Peninsula (WUP). Within the
WUP, Scull and Schaetzl (2011) identified, from east to west, the Peshekee, Iron County,
Marenisco-Winegar, and Keweenaw loess sheets.
The focus of this thesis research mainly centers on the Peshekee Loess, located in Baraga,
Iron, and Marquette Counties. This landscape is extremely heterogeneous, with hummocky,

2

high-relief (averaging > ~30 m), Precambrian bedrock-controlled uplands, and interspersed
glacial meltwater streams and valleys. Until ~11 500 cal. yrs BP, the northern parts of the area
were occupied by the Laurentide Ice Sheet (LIS). Outwash plains, end moraines, ground
moraines, and midsize glacial meltwater valleys formed at this time, and are common and often
in close proximity to one another. Glacial drift is relatively thin (< 30 m) and patchy in most
areas – and nonexistent on many bedrock uplands. On this recently deglaciated landscape, the
NRCS has mapped many hectares of two-storied soils that have formed in thin (avg. ~ 40-60
cm), silty, eolian material and underlying loamy and sandy glacial till on high-relief bedrock
uplands. The research goals of this thesis are to characterize the thickness and textural attributes
of the Peshekee Loess, and to determine the likely source area(s) for this loess.
Mapping the textural and thickness character of the Peshekee Loess is vital to a
determination of the region’s loess source(s). These data will also provide evidence on the postglacial environment (e.g. paleowind directions and wind velocities) during loess transport in the
western Upper Peninsula of Michigan. Moreover, the loess here is at the far eastern edge of the
vast North American loess, and thus, data from it will provide a unique perspective on thin loess
at the margins large loess areas. Information about the margins of a deposit like the Peshekee
Loess can provide valuable insights about its formation, perhaps more than data from the core, or
central, parts of the deposit. Therefore, this research will contribute to (1) a complete mapping of
North America’s loess and (2) advancing our knowledge of the paleoenvironmental conditions
while the LIS was retreating from the Great Lakes region.

3

2. Literature Review
Smalley and Smalley (1983) defined loess in terms of process and considered four
mechanisms related to its formation: (1) silty sediments between 20-60 µm forms, (2) the
material is transported by wind, (3) the sediment is deposited, and (4) the deposit experiences
post-depositional changes. In broad terms, Pye (1995) defined loess as sediment composed
predominantly of silt-size particles (20-60 µm), formed by the accumulation of wind-blown dust.
Loess deposits are common in the high latitudes of the world where continental and alpine
glaciers have historically developed (Flint 1971). Glaciers are known for grinding clastic
sediment into sand and silt sized particles. Loess maps worldwide show that wind-blown silt
covers a significant amount of the Earth’s land surface, perhaps as much as 10 percent (Pecsi
1963; Muhs and Bettis 2003). Eolian deposits, specifically loess, provide evidence of
environmental change during the Quaternary period, and thus, have been the focus of much
sedimentological and paleoenvironmental study (Mason et al. 1999; Muhs and Bettis 2000). In
this literature review, I will (1) discuss the background and significance of loess, (2) describe the
mechanical and physical characteristics of loess, and (3) describe research on the loess deposits
in Michigan, USA.
2.1 Background and Significance
North America has some of the thickest deposits of last-glacial loess in the world
(Roberts et al. 2003). In the Midwestern United States, loess deposits are mapped adjacent to
broad glacial meltwater valleys and are thickest near their source meltwater valley (e.g.,
Mississippi, Missouri, Illinois and Wabash Rivers) where loess exceeds ~ 30 meters in thickness
(Smith 1942; Olson and Ruhe 1979; Fehrenbacher et al. 1986; Muhs and Bettis 2000; Mason
2001; Bettis et al. 2003). Here, loess deposits not only tend to be thickest near major meltwater
valleys, but also texturally coarsest (Smith 1942; Frazee 1970; Bettis et al. 2003; Roberts et al.
4

2003). The systematic thinning and fining of loess, downwind from its source area, provide
evidence of both source area and paleowind directions (Wascher et al. 1947; Frazee et al. 1970;
Rutledge et al. 1985; Fehrenbacher et al. 1986; Leigh 1994; Pye 1995). These data can later be
used to test atmospheric general circulation models (Muhs and Bettis 2000). Moreover, loess can
record periods of landscape stability (soil development) and instability (loess accumulation) in
periglacial environments (Follmer 1996; Bettis et al. 2003; Hobbs et al. 2011; Schaetzl 2008)
2.2 Loess Production
Mechanisms for the production of naturally occurring silt-size material include: (1)
release of existing silt-size material from parent rocks, (2) glacial grinding, (3) frost weathering,
(4) eolian abrasion, (5) salt weathering, (6) chemical weathering, (7) clay pellet aggregation and
(8) biological processes (Pye 1995). In the mid-latitudes, silt is commonly produced by
continental glaciers in a process that Smalley (1990) defines as glacial grinding. Loess research
in central North America has been dominated by a conceptual model linking loess deposits to
large rivers that carried meltwater and glacially ground sediment, often from the LIS. The model
describes glacial meltwater valleys choked with silt-rich discharge in the summer, which later
become dry (low flow) and unprotected in the winter (and during low-flow periods in summer);
strong prevailing winds are then capable of blowing dry silt out of the river valley and onto the
adjoining uplands (Frye et al. 1962; Smalley 1966, 1975; Ruhe 1983; Johnson and Follmer 1989;
Follmer 1996).
2.3 Loess Transportation and Deposition
Transport of eolian sediment can occur by (1) suspension, (2) saltation and/or (3) creep in all cases transport depends on the size of the particle and wind characteristics. Eolian sediment
is not only transported as single grains, but also as aggregates of silt and clay (Mason et al.
2003). Suspension within the atmosphere can take place with coarse silts or grains < 50 µm and
5

as long as the particle’s minimum threshold velocity and/or turbulence is obtained (Tsoar and
Pye 1987). Saltation usually takes place with fine and medium sand (125-500 µm) grains (Tsoar
and Pye 1987). Grains between these particle ranges (50-125 µm) can be transported by shortterm suspension or modified saltation, depending on the wind velocity or turbulence (Tsoar and
Pye 1987).
Loess deposition occurs as a result of (1) a decrease in wind velocity or (2) particles
being washed out of the atmosphere (via precipitation). In any case, deposition is facilitated by
surface roughness and/or vegetation, as a result of decreases in wind velocity or turbulence.
(Tsoar and Pye 1987; Mason et al. 1999). Loess is initially deposited rather uniformly across an
area, covering the landscape evenly, with the thickest depths and coarsest sediment recorded near
the source (Ruhe 1954; Frazee et al. 1970; Fehrenbacher et al. 1986). However, the spatial
relationships between the loess source and associated deposits depend largely on the nature of
the relief, climatic and vegetation gradients in an area (Tsoar and Pye 1987; Pye 1995; Mason et
al. 1999). For example, loess deposits adjacent to and downwind from broad river valleys exhibit
a classic wedge-shape, where particles traveling close to the ground are eventually trapped by
vegetation and/or a topographic barrier proximal to the source, while the finer particles traveling
higher in the atmosphere are deposited further downwind (Pye 1995).
2.4 Preservation
Loess is initially deposited uniformly across a landscape, regardless of position along a
catena (Schaetzl and Anderson 2005). However loess preservation is greatly affected by the
topography of the underlying surface and the geography of the source area. Ruhe (1954) showed
that, due to post depositional processes, loess is least sorted when it overlies summit landscape
positions, and is better sorted on slopes where slopewash processes have sorted the loess with the
regolith. Furthermore, where the land surface topography follows an irregular relief pattern, post6

depositional processes can cause loess landscapes to be highly dissected. Loess deposits in a
dissected landscape tend to be best preserved and thickest on uplands (Ruhe 1954), whereas
loess on side slopes tends to be thinner, better sorted, and coarser, due to erosion and sorting of
fine grained sediments (Ruhe 1954). Furthermore, foot and toeslope positions tend to have
thicker loess deposits, in comparison to uplands that have loess, and are composed of even finer
sediment (Ruhe 1954). As a result of post depositional processes, colluvial additions can easily
be incorporated into loess deposits and lead to an overestimation of (1) loess thicknesses and (2)
percentage of fine-earth fraction, especially on foot and toeslopes (Smalley 1972; Pye 1995). In
sum, flat uplands are ideal sampling site locations when characterizing a loess deposit because
they tend to be the most stable landscape position, and thus would be the most representative
location for a loess deposit (Ruhe 1954).
2.5 Spatial Characteristics of Loess Deposits
Eolian particles fall out of suspension according to Stokes Law (Pye 1995), thus, larger
particles fall out of suspension more quickly than small particles. Because of this sorting, loess
deposits often exhibit a systematic fining of particle size with increasing distance from the source
(Smith 1942; Ruhe 1954; Frazee et al. 1970; Olson and Ruhe 1979; Muhs and Bettis 2000). Very
fine sand, as well as coarse and very coarse silt contents often decrease with increased distance
from a source, whereas the amount of fine silt and clay increases with distance downwind (Ruhe
1954; Frazee et al. 1970; Olson and Ruhe 1979). As a result, the textural character of a loess
deposit can, and often is, mathematically predictable downwind from its source area(s) (Smith
1942; Ruhe 1954; Fehrenbacher et al. 1965; Ruhe 1969; Frazee et al. 1970; Muhs and Bettis
2000).
Particle size distributions and the thickness of loess deposits vary greatly, based on the
characteristics of the region’s vegetation and topography and the distance from the source, along
7

the dominant wind direction. Tsoar and Pye (1987) showed how the wedge-shape of a loess
deposit is greatly affected by the landscape surface roughness and the type of vegetation present
during deposition. Forests are more efficient in trapping dust than steppe or tundra vegetation
due to its greater roughness (Tsoar and Pye 1987). A large change in surface roughness leads to a
rapid increase in net deposition closer to the roughness boundary. For example, at the edge of a
forest where a large change in surface roughness is present, loess is easily trapped by the tall
vegetation (and washed to the surface) creating a steep gradient loess wedge (Tsoar and Pye
1987; Pye 1995). Conversely, where the height of the vegetation is lower (e.g., grass land or
tundra), suspended particles are less likely to be trapped, resulting in a more gradually sloping
loess wedge (Tsoar and Pye 1987).
Mason et al. (1999) expanded on the surface roughness model developed by Tsoar and
Pye (1987) and suggested that not only do vegetation and topographic barriers facilitate loess
deposition, but these factors also provide a barrier between thick and thin loess deposits. Mason
et al. (1999) argued that loess distribution in the Central Lowlands of North America often
reflects regional topographic controls on the transport of sand from eolian sediment sources, and
in turn controls the distribution of thick loess. In thier model, long-term accumulation of loess
occurs mainly downwind of topographic obstacles that limit eolian transport of sand, and
associated re-entrainment dust. This process begins as sand from the source area saltates along
the source surface and, in the process, stirs up silt and clay sized particles. The deflated, fine
particles are then re-entrained and carried further downwind. This process continues until the
sand grains encounter a barrier, e.g., a steep valley, an escarpment, or an increase in local relief.
Because of the barrier, saltating sands become trapped and the windward surface of the barrier
eventually becomes stabile, while the surface downwind of the barrier continues to accumulate

8

fine sediments and is, therefore, unstable. In this model, topographic barriers often act as a
boundary between thick and thin loess deposits.
Additionally, Putman et al. (1988; 1989) suggested that loess is both thicker and coarser
near wider valley sections, and ultimately correlates to the amount of sediment available and the
dominant wind direction.
2.6 Loess Chronology in North America
The loess and intercalated paleosols in the mid-continental United States have been
extensively studied during the past century and serve as a key stratigraphic foundation in
deciphering the Pleistocene history of North America (Ruhe 1983; Follmer 1996; Muhs and
Bettis 2000; Forman and Pierson 2002). The ages of loess deposits found within North America
are usually based on the dates obtained from (1) radiocarbon dating of soil organic material
(either underlying and/or overlying the deposit), and/or (2) optically stimulated luminescence (or
thermoluminescence) dating of the loess itself. The loess deposits found on the Great Plains and
Central Lowlands of North America are generally correlated based on the age of the deposit. The
following section will include a brief discussion on the nomenclature and chronology of the
major loess units found in central North America.
There are up to seven major loess deposits in the Mississippi and Missouri River basins
(Follmer 1996). Three of the most recent, and most extensive, of these loess deposits are the
Loveland Loess, the Roxana Silt and the Peoria Loess (Willman and Frye 1970; Ruhe 1983). The
Loveland Loess is the oldest unit known in North America and is identified throughout the upper
Midwest in the Missouri, Arkansas, Mississippi, and Ohio River basins (Ruhe 1969; Willman
and Frye 1970; Ruhe and Olson 1980; Rutledge et al. 1996; Follmer 1996). Based on
thermoluminescence ages, Loveland Loess was first deposited ~ 165,000 cal. yr BP. Deposition
ended ~ 110,000 cal. yr BP, confirming its Illinoian age (Forman et al. 1992). The Sangamon
9

Geosol subsequently formed in the Loveland Loess (Follmer 1996; Ruhe 1956) during slow or
nondepositional periods, between ~ 130,000 and ~ 60,000 cal. yr BP, which represents the last
interglaciation (Leigh and Knox 1993, 1994; Markewich et al. 1998). Within the middle
Mississippi Valley, usually overlying the Loveland Formation (i.e., the Sangamon Geosol
developed in Loveland Loess) is Roxana Silt (loess) - a middle-Wisconsin aged deposit (Johnson
and Follmer 1989; Leigh and Knox 1993). Thermoluminescence ages on the Roxana Silt range
from ~ 30,000 to ~ 53,000 cal. yr BP for sites in the middle Mississippi Valley (Forman et al.
1992; Rodbell et al. 1997; Markewich et al. 1998). Ages of the Roxana Silt vary based on
location along the Mississippi River Valley and by different dating methods. The Farmdale
Geosol developed in the upper part of the Roxana Silt, at sites in and near the Mississippi River
Valley (Follmer 1983). Thermoluminescence ages in the Farmdale Geosol range from ~30,000 to
~ 23,000 cal. yr BP (Forman et al. 1992). Loess in a similar stratigraphic position as the Roxana
Formation (i.e., the Formdale Geosol developed in Roxana Silt) is recognized in Iowa as the
Pisgah Formation and in eastern Nebraska and Kansas as the Gilman Canyon Formation
(Forman et al. 1992; Mason et al. 2007). However, these correlative units generally have younger
ages; ranging between ~ 30,000 and ~ 45,000 cal. yr BP (Leigh and Knox 1993; Follmer 1996;
Maat and Johnson 1996).
The youngest and thickest loess unit east of the Missouri River is the Peoria Loess (Bettis
et al. 2003). In many localities within the Midwest and Central Lowlands Peoria Loess
accumulated on top of the Roxana Silt and the Farmdale Geosol. At Loveland, Missouri, Peoria
Loess is as much as 41 m thick (Forman et al. 1992). In the upper and middle Mississippi Valley,
deposition of Peoria Loess began ~ 30,000 cal. yr BP and ended ~ 12,900 cal. yr BP (Leigh and
Knox 1993; Markewich et al. 1998; Grimley 2000; Forman and Pierson 2002; Bettis et al. 2003).

10

The irregular loess thickness of Peoria Loess suggests that loess accumulation during this
time was likely variable. For example, recently, Stanley (2008) reported on thin (between 35 and
70 cm, respectively) loess in north-central Wisconsin, and found that these loess deposits range
between ~ 15,200 cal yr. BP and ~ 12,000 cal yr. BP. In north-central Lower Michigan Hobbs et
al. (2011) provided evidence of episodic loess deposition during the early-Holocene.
Radiocarbon ages on bulk charcoal from nine paleosols within kettle-bottom silt deposits were
found to span the entire Holocene. One date is late Pleistocene in age (10,930 14 C years ago) and
another is from the last 1000 years (920 14 C years ago). The remaining seven dates span from ~
9,500 to ~ 5,400 14 C years ago, i.e., generally between ~ 10,700 and ~ 6000 cal. years BP. Thus,
Hobbs et al. (2011) and Stanley (2008) provided evidence for loess deposition in the Great Lakes
region well after the LGM and into the Late Pleistocene and even well into the Early Holocene.
2.7 Loess in Michigan and nearby areas
Until recently, loess in Michigan has been largely overlooked. The state was not
recognized as having major loess source areas, due to the absence of large glacial meltwater river
valleys. However, lately, thin loess deposits have been recognized in areas such as northwestern
Lower Michigan (Schaetzl and Hook 2008), eastern Upper Peninsula (Schaetzl and Loope 2008),
north-central Lower Michigan (Schaetzl 2008), and central Lower Michigan (Hobbs et al. 2011).
The loess deposits noted in Michigan are thin compared to the loess deposits studied in the Great
Plains and Central Lowlands of North America. Thin loess deposits in Michigan could be
attributed to (1) the recent, and in some cases rapid, retreat of the LIS from this region and (2)
the mechanics of the source area(s) from which the loess deposit’s originated.
Schaetzl and Hook (2008) reported on soils with a silt-rich mantle, generally 35-45 cm
thick, on a section of the Outer Port Huron outwash plain in northwest Lower Michigan, known
locally as the Buckley Flats. The silty cap thins progressively from south to north, away from the
11

Manistee River valley, and also becomes finer in surface texture toward the north. The authors
concluded that the silt-textured mantle on the Buckley Flats is loess, thereby documenting loess
in Lower Michigan. The presumed source for this loess is the Manistee River valley, which is
thought to have carried glacial meltwater from the Port Huron Advance. This model suggests
that Port Huron meltwaters, choked with silty sediment, later deposited these materials on the
Manistee River floodplain, from which, these fluvial deposits were deflated and transported by
wind and, over time, deposited on the nearby uplands.
Schaetzl and Loope (2008) examined the distribution of a thin, silty deposit on uplands in
eastern Upper Peninsula of Michigan. The authors explored the distribution, likely source, and
general nature of this 20-60 cm-thick silt cap, using the loess formation model of Mason et al.
(1999). Textural, geochemical, and mineralogical characteristics of the silty cap provided
evidence that the uplands are mantled by loess which deflated from dried-up lake floors after
nearby post-glacial lakes had drained.
Stanley and Schaetzl (2011) concluded that a thin loess deposit in north-central
Wisconsin had two loess sources, neither of which fell within the traditional mode of “glacial
meltwater valley.” The eastern margins of the north-central Wisconsin loess deposit has a thin (<
35 cm thick) silt loam surface texture, and this eolian cap thickens (to > 70 cm) and coarsens
towards the west and northwest, such that, on its western margins, the loess mantle is dominated
by very fine and fine sands. Collectively, data on loess particle size distributions, thickness, and
silt mineralogy suggest that this loess had sources in two distinct and disjunct landscapes: (1) the
late Wisconsin terminal moraine to the northwest, and (2) the sandstone-dominated landscapes to
the west and southwest. Stanley and Schaetzl (2011) suggested that post-glacial thawing of the
permafrost-affected landscape probably led to draining of ice-walled lake plains on and behind

12

the moraine, located just northwest of the loess, as well as destabilization of slopes in the
sandstone landscapes to the south and southwest of the loess; in both cases, large quantities of
sediment were exposed for deflation. The study made two significant contributions to eolian
research: (1) it suggested that loess deposits can have two distinctly different and disjunct loess
sources, and (2) landscapes other than the traditional “glacial meltwater valley” can serve as
loess sources.
Schaetzl (2008) focused on a silty cap overlying till and outwash in the Graying Fingers
region of northern Lower Michigan. Data there suggested that the loess was derived from two
local-source areas, the Port Huron outwash plain and (to a lesser extent) the Manistee River
valley. Schaetzl (2008) found that the silty cap is geochemically and texturally unlike the glacial
sediments that underlie it. Also, loess is only located on the flattest parts of the Finger uplands
and within the bottoms of upland dry kettles. Because of post depositional erosion, loess was
only preserved on flat upland surfaces or within closed depressions. The Graying Fingers work
suggests that outwash plains and medium-size meltwater streams can also be significant sources
for loess (Schaetzl, 2008).
Hobbs et al. (2011) reported on small bodies of silty sediment, frequently occurring in the
bottom centers of dry kettles. The anomalous silty sediment is found in an otherwise dry and
sandy interlobate region of central Lower Michigan, known as the Evart Upland. The small silty
deposits are dominated by fine silts, and often have one or more charcoal-rich paleosols within
their sedimentary sequence. Secondary particle size peaks of medium sand in these deposits
match the sands found on the kettle backslopes. Moreover, the backslope sediments contain little
silt, and what silt they have was typically coarse silt, suggesting that the silt in the kettles was not
derived solely from the kettle backslopes. The authors suggest that the silt is loess that was

13

intermittently blown onto the landscape from the nearby Muskegon River floodplain and/or from
other, silt-rich surfaces recently disturbed by fire. Charcoal fragments within the paleosols
provided evidence for Holocene-age loess deposition in central Lower Michigan.
Michigan’s loess deposits, although rare, have mainly been linked to last-glacial outwash
plains, mid-size meltwater river valleys and abandoned lake floors. Additional loess research in
Michigan could potently provide more evidence in support of glacigenic landforms as loess
source areas, e.g., glacial outwash plains, mid-size glacial meltwater stream valleys, dried-up
lake plains, ground moraines, and end moraines. Additionally, expanded loess research in
Michigan could contribute to mapping the total extent of North America’s loess deposits.
2.8 Summary
Mapping the spatial characteristics of loess deposits is an important step to understanding
the paleoenvironmental conditions that existed during loess deposition. After a loess provenance
is identified, paleoenvironmental conditions during loess deposition can often be inferred, e.g.,
temperature, wind direction and/or wind velocity.
Loess deposits in mid-continental North America are common and are often reported
adjacent to, and downwind of, large glacial meltwater river valleys. The thickest and coarsest
loess deposits are found near the source whereas thinner loess deposits and finer-textured
sediment is reported farther from the source. Thus, loess deposits have spatial trends with respect
to thickness and particle size that enable the determination of source areas.
Recently, loess research in Michigan has suggested that glacigenic landforms rich in silt
content (e.g. outwash plains, mid-size stream valleys and dried up lake floors) can also serve as
significant source areas for loess. Although Michigan loess deposits are relatively thin (i.e.,
between ~ 20 – 60 cm) and particle size distributions are more variable, these deposits do
illustrate the traditional “wedge-shaped” character of a loess deposit.
14

3. Study Area
The focus of this study is Peshekee Loess and the very northeastern margins of the Iron
1

County Loess, located in Michigan’s western Upper Peninsula (Scull and Schaetzl 2011).
Combined, both regions of loess span parts of Baraga, Dickenson, Marquette, and Iron Counties
(Figure 3.1). The NRCS has recognized loess deposits in the soils in all of these counties, where
it usually overlies glacial till, glacial outwash and/or bedrock (Berndt 1988; Linsemier 1989;
Schwenner 2007). This chapter provides background information on the extent of the broad
study area, in addition to its bedrock geology, glacial history, physiography and soils.
3.1 Extent of the Broad Study area
Loess in the western Upper Peninsula, although extremely patchy, extends across
Menominee, Marquette, Dickinson, Baraga, Iron, Houghton, Keweenaw, Ontonagon, and
Gogebic Counties (NRCS SSURGO Data). Due to the vast extent of the loess and the lack of
road access and time, this research was limited to a ~ 560,000 hectare area that extends east-west
for 80 km and north-south for 70 km (Figure 3.1). The broad study area, lies within Baraga,
Dickinson, Marquette, and Iron Counties and includes the Peshekee Loess and the very
northeastern parts of the Iron County Loess (Scull and Schaetzl 2011). Later in the thesis the
study area is defined by the extent of the loess sample site locations.

1

Although this is the first study to focus on the characterization of the silt mantle on soils of
Baraga, Dickinson, Iron, and Marquette Counties, all of the county soil surveys for this region
have identified soils formed in silty or loamy eolian deposits over sandy and gravelly outwash,
till, and bedrock (Berndt 1988; Linsemier 1989; Schwenner 2007). Because, by definition,
sediment with this description has been widely accepted as loess (Schaetzl and Loope 2008;
Schaetzl and Hook 2008; Schaetzl 2008; Scull and Schaetzl 2010), I will refer to the silt loam
mantle as loess for the remainder of this paper.
15

Figure 3.1: Generalized map of soils with loess parent materials in western
Upper Michigan (after Berndt 1988; Linsemier 1989; Schwenner 2007). Red
represents soils series mappped with loess thicknesses of 51 - 100 cm. Pink
represents soil series mapped with loess thicknesses of 25 - 50 cm. Soils with
loamy eolian sediment parent materials are colored purple. The broad study
area outline includes parts of two loess regions named by Scull and Schaetzl
(2011). For interpretation of the references to color in this and all other
figures, the reader is referred to the electronic version of this thesis.

16

The broad study area is situated in the Superior Upland physiographic province.
Fenneman (1938) describes the Superior Upland as part of the Canadian Shield, dominated by
underlying Precambrian bedrock, mainly crystalline in structure, with some rock types
completely worn down, while others are more resistant and stand up as ridges. Repeated
Pleistocene glacial advances have deposited glacial drift in places, removed the mantle of many
rock types in other locations, destroyed historical drainage networks, and left behind what now
looks like a virgin landscape – minus the dense forests and abundant inland lakes and wetlands
(Leverett 1911; Fenneman 1938). The landscape’s hydrology is young, and thus streams,
swamps, and lakes are abundant, forming a deranged drainage network.
3.2 Bedrock Geology
Because of only thin deposits of glacial drift in many places, the land surface topography
generally mimics the underlying bedrock surface. Precambrian crystalline bedrock underlies the
southern and central sections of the broad study area, whereas Paleozoic sandstone benches are
exposed on the northern edge along Lake Superior (Card 1990; Wikgren 2007). The bedrock
geology of this region is extremely complex, and knowledge of the geology is not at all complete
(Bornhorst and Brandt 2009). Sedimentary, igneous and metamorphic rocks exposed at the
surface can vary greatly over short distances, and most are obscured by glacial deposits and/or
vegetation and water.
The Precambrian rocks of the broad study area are over 2.5 Ga old, and include the
southern extent of the Canadian Shield (Card 1990). Some of the more extensive Precambrian
rock units within the broad study area include Archean granites and gneisses, as well as
Paleoproterozoic and Mesoproterozoic sedimentary rocks (Figure 3.2). The Archean granites and
gneisses are not only some of the oldest rocks in Michigan, but also on Earth (Sims 1993). The
Republic (granite) and Palmer (gneiss) Formations are common in the north-central part of the
17

broad study area and are Eoarchean (4.0-3.6 Ga) to Mesoarchean (3.6-2.8 Ga) in age (Schulz and
Cannon 2007; Figure 3.2). These rock types form granitic bedrock glades with steep hills, short
cliffs, and talus slopes at the base of bedrock exposures. Glacial drift is generally < 6 meters
thick on the uplands, with many bare bedrock outcrops; thicker glacial deposits occupy the
valleys (Schwenner 2007; Wikgren 2007).

Figure 3.2: Generalized bedrock map of Michigan’s western Upper Peninsula and an
outline of the broad study area’s common rock episodes, along with examples of
formal rock unite names and dominant rock types (after Sims 1993; Cannon 2007;
Bornhorst and Brandt 2009).
18

An abrupt transition from granites and gneisses to high-grade metamorphic sedimentary
rock types is evident in the central and southwestern regions of the broad study area (Figure 3.2).
South of the region dominated by granites and gneisses, outcrops are less concentrated and form
large, rounded ridges. Here, glacial drift is thicker (< ~ 20 meters) and parallel drainage networks
are more common (as opposed to the dendritic systems common to the north). This landscape is
mostly underlain by metamorphic sedimentary rocks that are Paleoproterozoic (2.3-1.76 Ga) in
age; in Michigan known as the Marquette Range Supergroup. They formed as fine-grain
sedimentary rock types, which later formed into shale and in some cases slate (metamorphosed
shale). Rocks of the Paleoproterozoic era are particularly complex in Michigan, and are lumped
into the Marquette Range Supergroup which is subdivided into the lower Chocolay Group (the
oldest), the middle Menominee Group, and the upper Baraga Group. Each is separated by
unconformities that represent long periods of erosion (Bornhorst and Brandt 2009; Figure 3.2).
The northwestern corner of the broad study area is underlain by a thick succession of
volcanic and sedimentary rocks of Mesoproterozoic (1.76 to 1.1 Ga) age rocks that formed
within the Midcontinent rift system (Davis and Green 1997) (Figure 3.2). A common
Mesoproterozoic sedimentary rock found in the northeastern parts of the broad study area and
along the Lake Superior shoreline is the Jacobsville Sandstone (Sims 1992; Figure 3.2). The
Jacobsville Sandstone is identified by its mottled reddish-pinkish-whitish colors; it is almost pure
quartzose sandstone with some minor siltstone, shale and conglomerate lithologies.
The bedrock underlying the broad study area is dominantly Precambrian igneous and
high-grade metamorphic rock type. Given the resistant nature of these lithologies, it is unlikely
that any of the local bedrock formations have weathered enough to form silt-rich residual soils,
as described by the NRCS.

19

3.3 Glacial History
During the Pleistocene Epoch, the broad study area (BSA) was repeatedly covered by
glacial ice (Attig et al. 1985; Lowell et al. 1999). However, overall, glacial deposits in this region
are relatively thin (average < ~ 30 meters) and vary greatly in thickness (Vanlier 1963; Derouin
et al. 2007). In addition to glacial till and outwash, alluvium, lacustrine and eolian deposits
typically are the main forms of regolith in the region (Berndt 1988; Linsemier 1989; Schwenner
2007). In places, ice advance positions are well-marked by end moraines, but others are poorly
defined due to later ice readvances, wave action, and/or collapse topography (e.g., Larson and
Schaetzl 2001). Despite this lack of temporal accuracy, the glacial landforms and deposits of this
region have been correlated to any of four glacial lobes, which last advanced into the area
2

between approximately 15,000 and 11,500 cal. yrs ago , during the Wisconsin Glaciation (Attig
et. al. 1985; Lowell et al. 1999; Larson and Schaetzl 2001; Derouin et al. 2007). The four major
ice lobes that contributed to the construction of the glacial landscape, from west to east, are the
Ontonagon, Keweenaw Bay, Michigamme, and Green Bay Lobes (Attig et al. 1985; Peterson
1986; Figure 3.3). This section discusses the major moraines and other glacial deposits within the
broad study area and their significance to Peshekee Loess.

2

In this thesis, all 14C age estimates taken from the literature have been converted into calendar
years BP, using the calibration curve of Fairbanks et al. (2005):
http://www.radiocarbon.ldeo.columbia.edu/research/radcarbcal.htm (Fairbanks et al. 2005).
20

Figure 3.3: Extent of the last four major ice lobes that sculpted the glacial landscape of
the broad study area (Attig et al. 1985; Peterson 1986).

21

The southern-most moraine within the BSA is the Winegar/Watersmeet Moraine,
deposited by the Ontonagon and Keweenaw Bay Lobes (Figure 3.4). The Winegar Moraine is a
discrete feature that extends across northeastern Wisconsin and into Iron County, Michigan
where it is referred to as the Watersmeet Moraine (Attig et al. 1985). The Watersmeet Moraine
is a disintegration moraine built during the Winegar-Early Athelstane Phase

3

(~15,000 cal. yr

BP) that is correlated with the Port Huron Advance (Attig 1984).

Figure 3.4: A generalized map of the moraines or ice-margin positions in the broad
study area (after Attig et al. 1985).

3

During wastage from a glacial lobe’s maximum extent, the ice margin paused or advanced a
number of times, from here on out “Phase” refers to these pause and advance cycles (after Attig
et al. 1985).
22

South of the Watersmeet Moraine the landscape is primarily composed of outwash/fluvial
surfaces, with many kettles, and small areas of drumlins, whereas to the north of the moraine, the
landscape is marked by hummocky bedrock controlled topography with many closed depressions
and few lakes (Attig et al. 1985). Moreover, the moraine is composed of gravelly sandy brown
till (Peterson 1986). Attig et al. (1985) mapped the Watersmeet Moraine as the Republic Moraine
in Marquette County and as the Sagola Moraine in Dickinson County (Figure 3.4). Peterson
(1986) suggested that the Sagola Moraine was built by the Green Bay Lobe and consists of
calcareous red drift composed of ice contact, stratified sand and gravel; the moraine is marked by
hummocky topography with many kettles to the south. The red color of the drift suggests that its
ultimate source was the red Precambrian sandstones and shale from the Lake Superior basin
(Attig et al. 1985; Peterson 1986).
After the Winegar-Early Athelstane Phase, the margins of the four lobes eventually
retreated north, possibly well into the Lake Superior basin, but advanced again to build the
Marenisco, Watton, and Marquette/Sixmile moraines (Figure 3.4). The Marenisco Moraine was
built by the Green Bay Lobe during the Marenisco-Late Athelstane Phase (approximately
between 13,700 and 13,400 cal. yr BP). Unlike the moraines of the Winegar-early Athelstane
Phase, the Marenisco Moraine is hard to trace because there is little evidence of sediment
collapse. During the Winegar-early Athelstane Phase, the ice advanced over sediment and ice
from earlier events, which later resulted in widespread collapse of till and supraglacial sediment,
making the ice-margins traceable. However, the Marenisco Moraine forms a subdued and
dissected ridge just north of the Winegar/Watersmeet Moraine (Peterson 1982; Attig et al. 1985;
Figure 3.4).

23

Following the Marenisco-Late Athelstane Phase were the Porcupine and Marquette
Phases (Attig et al. 1985; Derouin et al. 2007). The Watton moraine, built by the Keweenaw Bay
Lobe, formed at the Porcupine ice-margin position which extends from southwestern Baraga
County to northeastern Marquette County (Figure 3.4). The western and southern parts of the
Lake Superior basin are known to have been ice free approximately between 13,000 and 10,000
cal. yr BP, based on evidence of Lake Agassiz draining eastward through Lakes Superior and
Huron at this time (Clayton 1983). Therefore, the Porcupine Phase is speculated to have occurred
~13,000 cal. yr BP (Clayton 1984; Attig et al. 1985).
The last documented glacial phase to have advanced beyond the northern rim of the BSA
and southern rim of the Lake Superior basin is known as the Marquette Phase (Attig et al. 1985),
previously known as the Marquette readvance (Hughes and Merry 1978; Farrand and Drexler
1985; Lowell et al. 1999). The Keweenaw and Michigamme Lobes built what is known in
Baraga County as the Sixmile Moraine, and in Marquette County as the Marquette Moraine; both
are composed of red clayey till (Attig et al. 1985). Wood buried in glaciofluvial sediment from
the Marquette Phase has been dated at ~ 11,500 cal. yr BP (Hughes and Merry 1978; Lowell et
al. 1999; Pregitzer et al. 2000).
The glacial history, outlined above, suggests a maximum of ~ 11,500 years (Marquette
Phase) available for soil development in the northern regions of the broad study area ~ 15,000
years (Winegar-Early Athelstane Phase) in the southern regions for pedogenesis.

24

3.4 Physiography and relief
The landscape encapsulating the broad study area is bedrock-controlled, with uplands
ranging from 0 - 205 m in local relief, interlaced with valleys that often contain sandy and
gravelly glacial and fluvial deposits. For discussion purposes, the broad study area is separated
into six physiographic regions; each region is delineated based on its uniform
physical/environmental attributes, such as physiography/relief, soils, hydrology and vegetation
(Schaetzl et al. (in prep)). The six sizable physiographic regions (Figure 3.5) that make up ≥ ~ 2
percent of the BSA include, from north to south, the Yellow Dog Plains (Black 1969), Peshekee
Highlands (Peterson 1986), Baraga Plains (Doonan and Byerlay 1973), Michigamme Highlands
(Brubacker 1975), Gwinn Sandy Terrain (Schwenner 2007), and the Iron Mountain Bedrock
Uplands (Schaetzl et al. (in prep).
Local relief for these regions was measured using a DEM raster, at 10 m resolution
(USGS, 2009) and using map algebra in ArcMap 10 software (ESRI 2011) to calculate the
elevation range in a 250 m circle for each 30-m output cell. The use of similar local relief
measures has been successfully applied to other glacial landscapes, as a means of delineating
regions (e.g. Stoelting 1989).
Taken together, the Yellow Dog Plains and the Baraga Plains combined make up ~ 2
percent of the BSA. Both are low relief outwash plains, with vast expanses of excessively
drained sandy soils, interspersed with poorly drained, organic rich soils (Figures 3.5 and 3.6;
Table 3.1). Jack pine forest occupies the dry and sandy soils, whereas black spruce thrives in the
wetter areas.

25

Figure 3.5: Major physiographic regions within the broad study area
(Schaetzl et al. (in prep.)).

26

South of the Yellow Dog Plains and east of the Baraga Plains are the Peshekee
Highlands. The Peshekee Highlands total 28 percent of the BSA and have a maximum relief of
205 m (per 250 m radius circle) and an average relief of 30 m. This region is a bedrockcontrolled landscape, with poorly drained soils dominating the intervening lowlands, and well
drained and moderately well drained sandy soils developing on the uplands (Figures 3.5 and 3.7;
Table 3.2). There is a large change in relief – often present as a ragged bedrock escarpment between the surrounding regions and the Peshekee Highlands; Figure 3.8 illustrates the
variability in relief across the region.

27

Figure 3.6: Photographs of the Mulligan Plains (A) and Yellow Dog Plains (B), both low-relief,
excessively well drained, sandy outwash plains, interspersed with poorly drained soils. Photograph
(A) by M. Luehmann and (B) by R. Schaetzl.
Table 3.1: Description of the Yellow Dog Plains and Baraga Plains physiographic regions (after Schaetzl et al. (in prep)).
Region
Yellow Dog
Plains

Physiography/relief
Low relief (~ 7 m
average), sandy, outwash
or glaciolacustrine plain.

Soils/sediments
Excessively drained,
sandy soils dominate,
with wet soils on the
southern margin.

Hydrology
Yellow Dog River
flows across the
southern margin.

Presettlement vegetation
Jack and red pine on dry
expanses, northern
hardwoods at edges, and bogs
on wettest, southern margins.

Baraga Plains

Low relief (~ 4 m
average), sandy, dry,
outwash or
glaciolacustrine plain.

Excessively drained,
sandy soils dominate,
with very poorly
drained on the SE
margin.

Very little surface
water.

Jack and red pine forests and
barrens. Conifer swamps on
SE margin.

28

Figure 3.7: Photographs at the base of a resistant, high-relief, Precambrian bedrock knob within
the Peshekee Highlands. Photograph (A) by L. Safford, and (B) by R. Schaetzl.
Table 3.2: Description of the Peshekee Highlands physiographic region (after Schaetzl et al. (in prep)).
Region

Physiography/relief

Peshekee Bedrock-controlled
Highlands landscape of high relief (~ 30
m average) and irregular
topography. Glacially eroded
bedrock knobs and driftfilled valleys.

Soils/sediments

Hydrology

Presettlement vegetation

Well drained, shallow
soils on uplands, poorly
drained soils and
Histosols in lowlands.
Many upland soils have
a thin loess cap.

Major upland of the
western UP, forming a
regional drainage divide.
Abundant lakes, few
through-flowing streams.

Sugar maple-hemlock
forests on uplands, conifer
swamps in lowlands.

29

Figure 3.8: Local relief variation within the broad study
area, and the physiographic region boundaries.

30

The Michigamme Bedrock Terrain is the largest, homogeneous physiographic region in
the western Upper Peninsula. In addition, it is the largest of the six physiographic regions within
the broad study area, occupying 45 percent of the region. This region has a highly deranged
drainage pattern, with moderate relief (averaging ~ 13 m), and many small, linear bedrock knobs
(Figures 3.5 and 3.8; Table 3.3). The linear bedrock knobs are mantled with thin regolith
composed of a silt-rich cap overlying sandy loam till or outwash. East of the Michigamme
Bedrock Terrain is the Gwinn Sandy Terrain, a moderate relief landscape, with variable
topography and sandy parent materials (Figure 3.9). Occupying the southeastern corner of the
BSA is the Iron Mountain Bedrock Uplands. This region is dominantly underlain by Archean
granites and gneisses (as opposed to slate directly to the west and north in the Michigamme
Bedrock Terrain). The Iron Mountain Bedrock Uplands consist of rolling terrain with isolated
hills and well drained sandy and loamy soils.
A heterogeneous mixture of parent materials and drift types occupies the Peshekee
Highlands and Michigamme Bedrock Terrain - the two largest physiographic regions within the
BSA. However a commonality between these two regions is that the glacial drift here is sandy,
and commonly overlain by a thin, silt-rich and very fine sand loess cap (Figure 3.9).

31

Table 3.3: Description of the Michigamme Bedrock Terrain, Gwinn Sandy Terrain, and Iron Mountain Bedrock Uplands
physiographic regions (after Schaetzl et al. (in prep)).
Region

Presettlement
vegetation
Deranged drainage, many Sugar maple-yellow
swamps and lakes.
birch-hemlock
General, broad upland
forests on uplands,
that is also a regional
conifer swamps in
drainage divide; low
lowlands.
stream gradients. Many
wetlands are expansive.

Physiography/relief

Soils/sediments

Michigamme
Bedrock
Terrain

Hummocky and occasionally
incised plains with moderate
relief; many bedrock knobs.
Drift is sandy, thin in many
places but can be, locally,
thick. Many areas could
simply be bedrock-influenced
ground moraine.

Moderately well drained,
or drier, soils on uplands.
Many poorly drained
soils, and Histosols, in
lowland swamps. Many
soils have a thin mantle of
loess.

Gwinn Sandy
Terrain

Heterogeneous area of
moderate relief but widely
variable topography and
sandy soils. Topography
includes hummocky areas,
incised stream valleys, small
bedrock outcrops and
bedrock-influenced terrain,
and low, rolling terrain on
thick drift.
Rolling, moderate relief
terrain with numerous
bedrock-cored knobs and
isolated hills, usually
sandstone. Heterogeneous
mix of parent materials and
drift types, but most areas are
sandy. Strong west-to-east,
regional, topographic slope.

Generally sandy soils that
span a wide range of
drainage classes. Parent
materials are typically
sandy glaciofluvial
materials. Histosols
common in lowlands,
especially in the west.

Complex hydrology, with
swamps in lowlands;
uplands are often very
dry and sandy.

Heterogeneous mix
of pine and
hardwood forests, as
well as sprucecedar-fir stands in
certain areas.

Well drained, sandy loam
soils formed in till and
glaciofluvial materials, on
uplands. Small areas of
Histosols in lowlands,
between bedrock knobs.

Eastwardly draining
streams, some swamps.
Generally a well-drained,
upland landscape.

Mix of northern
hardwoods (sugar
maple-basswood,hemlock-yellow
birch) on uplands;
conifer swamps in
lowlands. Some
areas of mostly
pine.

Iron Mountain
Bedrock
Uplands

32

Hydrology

Figure 3.9: Upper solum parent materials within the broad study area
(after Berndt 1988; Linsemier 1989; Schwenner 2007).

33

3.5 Climate and vegetation
The mean annual temperature (1971-2000) for Champion, in central Marquette County, is
~ 4° C, with a mean January temperature of ~ -12°C and mean July temperature of ~ 18°C
(National Climatic Data Center 2002). Total mean annual precipitation is ~ 84 cm, of which ~
54 cm falls between May and October (National Climatic Data Center 2002). Snowfall generally
occurs from September through April, and becomes more prevalent further from the lake (higher
elevations average annually > 500 cm). The mean annual snowfall for the period 1971-2000 in
Champion is 350 cm (http://www.weatherreports.com). Although current winds are
multidirectional due to seasonal variations, the dominant wind is from the northwest (Eichenlaub
et al. 1990; Arbogast and Packman 2004).
The vegetation of this region varies based on soils, elevation and landscape position
(Barrett et al. 1995). The uplands are dominated by sugar maple (Acer saccharum), aspen
(Populus), red maple (Acer Rubrum), and yellow birch (Betula alleghaniensi), with admixtures
of some coniferous tree species, while the lowlands are generally occupied by mixed conifer
swamp species such as black spruce (Picea mariana), northern white cedar (Thuja occidentalis),
and balsam fir (Abies balsamea) (Comer et al. 1995; Figure 3.10).

34

Figure 3.10: Presettlement vegetation of the broad study area
(after Comer et al. 1995).
35

3.6 Soils
Soils within the broad study area mostly consist of well and moderately well drained
Spodosols on the high-relief bedrock (or drift) uplands. Upland soils here have largely formed in
silt loam-textured loess, generally ranging between 40 and 60 cm in thickness, overlying loamy
and sandy glacial drift, or bedrock (Berndt 1988; Linsemier 1989; Schwenner 2007). Valleys and
depressions between the uplands are commonly occupied by organic rich, poorly and very poorly
drained Histosols, with Histic epipedons that range from 120 and to 130 cm in thickness. For the
sake of discussion and relevance to the project, only the seven dominant upland loess-derived
(loessal) soil series will be discussed in this thesis. “Dominant,” in this case, refers to loessal soil
series mapped on uplands within the BSA that cover > ~1.5 percent of the broad study area.
The seven upland loessal soils are quite similar morphologically, with only subtle
differences in their loess cap thickness, surface texture, presence/absence of a fragipan and/or
presence/absence of an argillic horizon. In this section I will (1) discuss the spatial distribution of
the dominant loessal soils within the broad study area and their surface textures and loess cap
thicknesses, and (2) compare the major morphologic features, including diagnostic horizons, of
these soils.
3.6.1 Spatial distribution of surface textures and loess cap thicknesses
The seven dominant loessal soil series in the broad study area are Champion,
Michigamme, Petticoat, Dishno, Keewaydin, Goodman, and Wabeno (see Tables 3.4, 3.5, and
3.6 for taxonomic subgroup names). Each of these soil series covers a minimum of ~ 9,000
hectares within the BSA, and combined, these soils cover ~ 176,000 hectares, or 31 percent, of
the broad study area.
For the sake of organization, I grouped the seven soil series based on their surface texture
(i.e., based on first mineral horizon), as described in their official soil descriptions (Tables 3.4,
36

3.5, and 3.6; http://soildatamart.nrcs.usda.gov). All of these series have either silt loam, cobbly
silt loam, or cobbly fine sandy loam surface textures (Berndt 1988; Linsemier 1989; Schwenner
2007). On its southern margins, the BSA has mainly been mapped with loessal soil series that
have silt loam surface textures (Goodman and Wabeno series), whereas cobbly silt loam loessal
soil series are mapped in the western and central regions (Champion, Michigamme, Petticoat,
and Dishno series) and one cobbly fine sandy loam loessal soil series (Keewaydin series) is
mapped in the northern and central regions (Berndt 1988; Linsemier 1989; Schwenner 2007;
Figure 3.11).

37

Table 3.4: Characteristics of loess with cobbly silt loam surface textures
(http://soildatamart.nrcs.usda.gov).
Characteristics
Counties
mapped
within

Taxonomic
family

Parent
materials

Landform
Surface
texture
Eolian cap
thickness
Texture at
the *LD
Coverage
within the
broad study
area

Champion
Baraga, Iron,
and Marquette

Soil series
Michigamme
Petticoat
Baraga and
Baraga and
Marquette
Iron

Dishno
Marquette

Coarse-loamy,
mixed,
superactive,
frigid Oxyaquic
Fragiorthods

Coarse-loamy,
mixed, superactive,
frigid Fragic
Haplorthods

Coarse-loamy
over sandy or
sandy-skeletal,
mixed, active,
frigid Alfic
Haplorthods

Coarse-loamy
over sandy or
sandy-skeletal,
mixed,
superactive,
frigid Oxyaquic
Haplorthod

Modified loamy
eolian material
and in the
underling
gravelly sandy
or loamy glacial
till

Silty or loamy
mantle over loamy
glacial till
underlain by
igneous or
metamorphic
bedrock

Modified silty
eolian material
and in the
underlying
sandy glacial
till

Ground
moraines and
end moraines

Ground moraines

Ground
moraines

Silty or loamy
eolian deposits
over sandy and
gravelly till
underlain by
igneous or
metamophic
bedrock
Bedrock
controlled
moraines

Cobbly silt loam Cobbly silt loam

~ 56 cm
Gravelly sandy
loam

~ 69,000
hectares

~ 64 cm
Gravelly fine sand

~ 22,000 hectares

*LD - Lithologic discontinuity

38

Cobbly silt
loam
~ 97 cm
Very gravelly
loamy sand

~ 15,000
hectares

Cobbly silt
loam
~ 56 cm
Very stony
loamy sand

~ 9,000
hectares

Table 3.5: Characteristics of loess with Cobbly fine sandy loam surface textures (http://soildatamart.nrcs.usda.gov).
Counties
mapped
Taxonomic
Soil Series
within
family
Keewaydin Marquette Coarse-loamy
over sandy or
sandy-skeletal,
mixed,
superactive,
frigid Typic
Haplorthods
*LD - Lithologic discontinuity

Parent
materials
Loamy and
silty eolian
deposits and
in the
underlying
till

Landform
Bedrock
controlled
moraines

Surface
texture
Cobbly
fine
sandy
loam

Eolian
cap
thickness
~ 51 cm

Texture at
the *LD
Gravelly
loamy sand

Coverage
within the
broad study
area
~ 35,000
hectares

Table 3.6: Characteristics of loess with silt loam surface textures (http://soildatamart.nrcs.usda.gov).
Counties
mapped
Soil Series
within
Goodman Marquette
and Iron

Wabeno

Iron

Taxonomic
family
Coarse-loamy,
mixed,
superactive,
frigid Alfic
Haplorthods

Parent
materials
Loess and in
the
underlying till

Coarse-loamy,
mixed,
superactive,
frigid Alfic
Oxyaquic
Fragiorthods

Loess and the
underlying
loamy and
sandy till or
glacial mudflow sediment

Eolian
Surface
cap
Landform Texture thickness
Drumlins Silt loam ~ 64 cm
and
moraines

Drumlins
and
moraines

*LD - Lithologic discontinuity
39

Silt loam

~ 61 cm

Texture at
the *LD
Sandy loam

Gravelly
sandy loam

Coverage
within the
broad study
area
~ 15,000
hectares

~ 12,000
hectares

Figure 3.11: A map of the surface textures of the seven dominant loessal soil series in the broad study area (Berndt
1988; Linsemier 1989; Schwenner 2007).

40

The surface texture map (and later, the loess cap thickness map) has several distinct
breaks at county boundaries. These breaks are due to the different publication ages between
county-level soil survey projects, within the broad study area. The Baraga County soil survey
was published in 1988, Dickinson County in 1989 and the Iron County soil survey was published
in 1997, whereas the most recent county soil survey, the Marquette survey, dates to 2007. During
the period 1997 and 2007, series nomenclature and definitions had changed. At the MarquetteBaraga border there is a disjunct between soil series; the Champion soil series was dominantly
being mapped in Baraga County, whereas the subsequently developed Keewaydin series was
mapped only in Marquette County. Wikgren (personal communication 2011) explained the
introduction of the Keewaydin series, during the Marquette survey project, as follows. Marquette
County, like Baraga County, has many hectares of soils that have developed in ~50-60 cm of
silty-fine sandy eolian material overlying sandy drift. In Marquette County, however, these soils
have slightly coarser eolian material overlying till, and they rarely have fragipans. Soils with the
latter morphology were mapped as Keewaydin. Champion soils, mapped in Baraga County, have
siltier surface textures, and a well-developed fragipan. This mapping “break” suggests that there
may be a difference in loess texture as one traverses from Baraga to Marquette County. The
change in fragipan expression, across the same traverse, is less important to this research.
Obviously, neither of these transitions is abrupt or located immediately at the county border, and
thus, the disjunct in series nomenclature is due to the age of soil survey projects.
Based solely on data from NRCS county soil surveys, it appears that the broad study area
has finer surface textures (silt loam) in the southern and south-central regions and coarser surface
textures (cobbly fine sandy loam) in the northern and north-east regions (Figure 3.11). However,
it is important to note that, where the NRCS has mapped loessal soils with a cobbly silt loam

41

surface texture, these soils largely have silt loam fine earth fractions, but by volume have 15-35
percent cobbles (Soil Survey Manual 1993). Here, we are only concerned with each soil’s finetextured sediment, i.e., the sediment between the cobbles and gravel which was potentially
deposited via eolian processes. Thus, for this study, both cobbly silt loam and silt loam surface
textures are considered equivalent, because the “cobbly” modifier may be a reflection of mixing
processes, which have mixed cobbles and gravels that were below the loess to the surface (or
nearer to it).
After examining the NRCS county soil surveys, it became obvious that, generally, soil
series with thicker loess caps are mapped in the southern and south-central regions, whereas soil
series with thinner loess caps are mapped in the northern and north-central regions of the BSA
(Figure 3.12).

Figure 3.12: Map of loess thickness, based on the seven dominant loessal soil series in the
broad study area (Berndt 1988; Linsemier 1989; Schwenner 2007).

42

Further observations of the NRCS maps indicate that (1) the thickest loessal soil series
(Petticoat) is dominantly mapped in the southern and very center region of the broad study area
and (2) thickness trends for the broad study area generally point to a southerly and southwesterly source area, because the loess has a southwest-northeast thinning trend (Figure 3.12).
3.6.2 Major loessal soils and comparing their diagnostic horizons
The seven dominant upland loessal soils within the broad study area are very similar
morphologically, with only subtle differences in horizonation. Taxonomic designations for the
seven loessal soils are largely dependent on loess cap thickness, surface texture,
presence/absence of a fragipan and/or argillic horizon, and profile thickness. Figures 3.13 and
3.14 were developed in order to better illustrate the subtle differences and similarities between
the seven dominant soils. These figures compare the Champion series, the most abundant soil
series mapped within of the BSA, to the remaining six upland loessal soil series within the broad
study area.

43

Figure 3.13: A comparison of the similarities and differences between the Champion
series, the most abundant mapped loessal soil series within the broad study area, and the
Michigamme, Dishno, and Keewaydin series. (http://soildatamart.nrcs.usda.gov).

44

Figure 3.14: A comparison of the similarities and differences between the Champion
series, the most abundant mapped loessal soil series within the broad study area, and the
Petticoat, Goodman, and Wabeno series. (http://soildatamart.nrcs.usda.gov).

45

The Champion series covers > ~ 69,000 hectares of the broad study area, and is the most
dominant loessal soil series mapped within the broad study area (Tables 3.4 and 3.5). Second in
dominance is the Keewaydin series which

Table 3.7: Comparing the diagnostic
character between the Champion and
Keewaydin soil series
(http://soildatamart.nrcs.usda.gov).

covers ~ 35, 000 hectares. The Champion and
Keewaydin series are morphologically similar

Soil Series

Diagnostic
Champion Keewaydin
Character
Ochric
epipedon

Yes

in that they both have a spodic horizon, an
ochric epipedon, and have developed in loess

Yes

and underlying loamy sand glacial drift

Subsurface
diagnostic
horizon

- Spodic
- Fragipan

Loess cap
thickness

~ 56 cm

Drainage
class

Moderately
well
drained

Well
drained

Cobbly silt
loam

Cobbly
fine sandy
loam

(Figure 3.13; Table 3.7). The Keewaydin

- Spodic

series has a cobbly fine sandy loam surface
texture, and is commonly mapped in the

~ 51 cm

northeastern, northern and north-central
regions of the BSA, whereas the Champion
series has a cobbly silt loam surface texture

Upper solum
texture

and is mapped in the northern, western and
central regions of the BSA (Figure 3.12).

Lower most
horizon
texture

Gravelly
loamy sand

Very
cobbly
loamy
sand

Unlike the Champion soil profile (Oe-E-BhsBs1-Bs2-2Bx-2C), the Keewaydin series has a
“simple” soil profile (Oa-E-Bs1-Bs2-2BC-2C)

Shallow to
bedrock

No

No

(Figure 3.13). Keewaydin soils do not have a
Bhs horizon (i.e., evidence of illuvial

accumulation of organic matter and humus and the accumulation of sesquioxides Fe and Al), or a
Bx horizon (firm, brittle and/or high bulk density horizon- fragipan)

46

(http://soildatamart.nrcs.usda.gov). However, in context to this project, both soils have developed
in loess and the underlying loamy sand glacial till. The main difference is in surface texture and
the development of a fragipan.
The Champion, Michigamme, and Dishno series all have a cobbly silt loam surface
textures, a spodic horizon and an ochric epipedon (Table 3.8). The Michigamme and Dishno
series, like the Keewaydin series, are mapped in the northern and north-central regions of the
BSA. However, the Michigamme and Dishno series differ from the Keewaydin series in both
having a Bhs horizon, and being shallow to bedrock (Figure 3.13; Table 3.8). Dishno soils lack a
Table 3.8: Comparing the diagnostic character between the
Champion, Michigamme and Dishno soil series
(http://soildatamart.nrcs.usda.gov).

Soil Series
Diagnostic
Character Champion Michigamme
Ochric
epipedon

Yes

Subsurface
diagnostic
horizon

- Spodic
- Fragipan

Loess cap
thickness

~ 56 cm

Drainage
class

Yes

Dishno

- Spodic
- Fragipan

- Spodic

~ 64 cm

~ 56 cm
Moderately
well
drained

Upper solum
texture

Cobbly silt
loam

Cobbly silt
loam

Cobbly silt
loam

Lower most
horizon
texture

Gravelly
loamy sand

Firm loam

Loamy
coarse
sand

Shallow to
bedrock

No

Yes

Yes

47

and Michigamme soils have
developed a fragipan at the

Yes

Moderately
Well drained
well
drained

fragipan, whereas Champion

lithologic discontinuity
(Figure 3.13; Table 3.8).

In the southern and south-central regions of the BSA, the Petticoat, Goodman, and
Wabeno series are commonly mapped (Figures 3.11 and 3.12). Like Champion soils, the
Petticoat, Goodman, and Wabeno series have either a silt loam or cobbly silt loam surface
texture, a spodic horizon and an ochric epipedon (Table 3.9). However, unlike Champion soils,
Petticoat, Goodman, and Wabeno soils all have a Bt horizon (argillic horizon – accumulation of
silicate clay) (Figure 3.14). Furthermore, the Petticoat, Goodman, and Wabeno series have a
slightly thicker loess cap (~ 60 cm to 90 cm) in comparison to Champion soils in which the loess
cap is roughly 55 cm thick (Figure 3.14).
Table 3.9: Comparing the diagnostic character between the Champion,
Petticoat, Goodman, and Wabeno soil series
(http://soildatamart.nrcs.usda.gov).

Diagnostic
Character

Soil Series
Champion

Petticoat

Goodman

Wabeno

Yes

Yes

Yes

Yes

- Spodic
- Fragipan

- Spodic
- Argillic

- Spodic
- Argillic

- Spodic
- Fragipan
- Argillic

Loess cap
thickness

~ 56 cm

~ 97 cm

~ 64 cm

~ 61 cm

Drainage
class

Moderately
well
drained

Well
drained

Well
drained

Moderately
well
drained

Upper solum
texture

Cobbly silt
loam

Cobbly
silt loam

Silt loam

Silt loam

Lower most
horizon
texture

Gravelly
loamy sand

Very
gravelly
loamy
sand

Loamy
sand

Gravelly
sandy loam

Shallow to
bedrock

No

No

No

No

Ochric
epipedon
Subsurface
diagnostic
horizon

48

After examining the above data, I conclude that the majority of upland loessal soils
mapped in Iron County have a thicker loess cap, in comparison to what was found in Baraga and
Marquette Counties, and thus, the NRCS mapped out different soil series for each county soil
survey. The Petticoat series was limited to Iron County with a loess cap ~ 97 cm thick, whereas
Champion was mapped in Baraga County with a loess cap ~ 56 cm thick, and in Marquette
County Keewaydin was mapped with a loess cap ~ 51 cm thick. Additionally, loessal soils in
Iron County have finer and more clay-rich upper solum textures, as suggested by the loessal soils
with argillic horizons (Petticoat, Goodman, and Wabeno), and silt loam (Goodman and Wabeno)
and cobbly silt loam surface textures (Petticoat). In comparison, in Baraga and Marquette
Counties, loessal soils mapped there do not have an argillic horizon (Champion, Dishno,
Michigamme, and Keewaydin) and have cobbly fine sandy loam (Keewaydin) and Cobbly silt
loam (Champion, Michigamme, and Dishno) surface textures.
Within the broad study area, the seven dominant upland loessal soil series are similar in
that they are all two storied soils, mapped on uplands, and have formed in a finer-textured eolian
cap (loess) and underlying loamy and sandy glacial till. In addition, they have spodic horizons
and ochric epipedons (Tables 3.4, 3.5, 3.6, 3.7, 3.8 and 3.9). The main differences among these
soils is the subsurface diagnostic horizon (e.g. having a fragipan and/or argillic horizon), upper
solum texture, and the thickness of the loess cap (Tables 3.4, 3.5, 3.6, 3.7, 3.8 and 3.9). However,
soil horizonation and taxonomic delineations are of lesser concern for this project. The main
focus here is the presence/absence of the loess cap, loess cap thickness, and the surface texture of
the loess. Generally, based on the NRCS soil surveys, loessal soil series with a thicker eolian
mantle and a finer surface texture are mapped in the southern regions of the BSA, whereas

49

loessal soils with a thinner eolian mantle and a coarser surface texture are mapped in the northern
regions.
3.7 Summary of Broad Study Area
The broad study area, set within the western Upper Peninsula of Michigan, is complex in
its morphology and varies in topography. Multiple glacial episodes are evident for this area,
based on characteristics such as soil color, texture, lithology, and the terrain of the till sheets left
by these glaciers. The most recent glacial advance not only influenced the topography in the
western Upper Peninsula through its grinding and plucking mechanics, but also by its meltwater
streams. In western and northwestern Marquette County, the terrain is dominated by high-relief,
bedrock knobs that are interspersed with under-fit meltwater streams, small outwash plains, and
steep, V-shaped valleys. Here, on bedrock uplands, soils with fine sandy loam mantles often
overly glacial outwash, till, and/or Precambrian bedrock. South of this region, the terrain
becomes more subdued, as the thicknesses of glacial deposits increases and the distances
between bedrock uplands become larger. Generally, this landscape is wetter than the northern
regions, and the upland soils are thicker and have a silt loam mantle, opposed to a fine sandy
loam mantle.
4. Methods
4.1 Field Methods
In preparation for entering the field, potential loess sample locations were identified and
coded in a geographic information system (GIS). Loess “target locations” were required to meet
the following three criteria:
(1) At the site, the NRCS had mapped soils with a loess cap
(2) The site consisted of a broad and flat upland

50

(3) The site was in a woodlot, thereby negating major anthropogenic disturbances due to
cultivation.
Theoretically, sites with these three criteria would have well preserved loess and would be
representative of the loess here. Target locations were determined by inspecting three data sets in
a GIS: (1) topographic maps of the Upper Peninsula of Michigan – USGS digital raster graphic
(DRG), (2) a 10-meter
digital elevation model
(DEM), and (3) countylevel SSURGO (digital
soils) data (Figure 4.1).

Figure 4.1: (A) Map illustrating the distribution of loessal soils
on uplands within the broad study area, as indicated by the
NRCS. Green dots are locations where loess samples were taken.
(B) Sampling loess at a target location using a standard handheld auger. Photograph by R. Schaetzl.

51

Once in the field, at each target location, the site was carefully explored using a short handled
landscaping shovel to “peel open” multiple soils and hand texture the uppermost mineral
horizon. This procedure was done in order to identify soils that had the following four criteria:
(1) At least one mineral horizon
(2) Low percentage of cobbles – increasing the likelihood of an eolian deposit
(3) An upper horizon that was silty and/or had a very fine sandy texture.
If the soil met criteria 1-3, then it was more closely inspected to determine if it met the last
criterion: a silty or very fine sand textured upper mineral horizon, with either bedrock directly
below or sediment that was coarser in texture, in relation to the material above. Soils that met
these criteria were deemed optimal for sampling, and had the highest likelihood of being
representative of the original loess that was deposited in the broad study area.
Target locations were deemed unsuitable for sampling, and therefore “passed up,” if they
exhibited any of these four criteria:
(1) Bedrock at the surface
(2) Organic material (Oa, Oi, Oe) overlying bedrock
(3) Percentage of cobbles and gravels > ~ 80 percent
(4) Sand, loamy sand or coarser textures in the upper profile.
During June, July, and August 2010, loess samples were collected in the broad study
area, using either a hand-held bucket auger or a short-handle landscaping shovel. A bucket auger
was preferred when sampling loess > ~ 50 cm (not very often). Once a site was located that met
the four sampling criteria (outline above), the silt mantle’s thickness was determined by digging
a small exploratory pit - roughly 50 cm in diameter and 30 cm deep (depth varied with loess cap
thickness). Hand texturing was then used to identify the total thickness of the loess cap - the

52

distance between the top of the first mineral horizon and the estimated lithologic discontinuity
(to either bedrock or glacial drift) below. At each site, approximately 1 kilogram of soil was
taken from the clean, freshly exposed loess profile provided by the exploratory pit. The sample
was collected throughout the entire loess cap; however the loess sample was discontinued at ~ 7
cm above the lithologic discontinuity, i.e., the boundary between the silt mantle and the coarser
substrate or bedrock. This procedure limited the amount of mixed material in the sample, which
commonly exists at the base of the loess column or weathered bedrock. The objective of this
sampling procedure was to sample the most representative loess for that area, and loess that had
not been too contaminated by in-mixing of sediment from below. Soil samples were sequentially
labeled. Additionally, at each site, the location of the sample was stored in a GIS file, using a
laptop computer equipped with ArcMap 10 software (ESRI 2011) and a built in GPS unit.
Additional attribute information, digitally recorded at each site included:
(1) Loess mantle thickness
(2) The landform type or land-use, e.g., bedrock upland or woodlot
(3) Loess quality
Loess quality is a subjective measure based on loess thickness and texture; a rank of 0 implied
that the sediment seemed too coarse for classical silty loess. A rank of 10 suggested that the loess
was very silty and the loess cap was thick. For safe keeping, these attributes were also noted in a
field notebook.
4.2 Laboratory Methods
Each loess sample was individually ground with a mortar and wooden pestle, after being
either air dried and/or oven-dried at 30°C. The ground sample was then passed through a 2 mm
sieve to eliminate sediment or organic material > 2 mm dia. The remaining fine earth material
was sent through a sample splitter three times in order to homogenize the sample. All soil
53

samples were then prepared for particle size analysis (psa), which was performed using a
Malvern Mastersizer 2000 laser particle size analyzer (Malvern Instruments Ltd.,
Worcestershire, UK). Sample preparation, for particle size analysis, included placing ~ 1 g of
soil in a 25 ml vial and adding 5 ml of dispersant solution and 10 ml of distilled water to the vial.
The dispersing solution was 35.70 g (NaPO3)6 and 7.94 g Na2CO3 mixed into one liter of water
(Kilmer and Alexander 1949). Each vial with the soil sample and dispersion solution was then
shaken for 2 hours on a rotating table to properly disperse the sediment.
In recent years, laser diffractometry has generally replaced the traditional sieve-pipette
method for particle size analysis (Sperazza et al. 2004; Arriaga et al. 2006). Although the data
produced using these two methods are highly correlated (Arriaga et al. 2006), differences in the
estimation of the clay fraction do occur (Buurman et al. 1997). The laser method commonly
underestimates the amount of < 2 μm clay, when compared to the pipette method (Loizeau et al.
1994; Beuselinck et al. 1998). For this reason, Konert and Vandenberghe (1997) suggested that a
clay-silt break of 8 μm be utilized in laser diffractometry, in order to facilitate any comparisons
with traditional particle size analysis data. In-house data analyses produced the highest
correlations between clay contents determined by pipette vs. laser diffractometry when the claysilt break for the latter is set at 6 μm (Hobbs et al. 2011). This is the procedure followed in this
study.
In order to make the data set more robust and produce kriged surfaces of greater accuracy
with less error, an additional 59 samples collected by M. Bigsby and R. Schaetzl during previous
fieldwork campaigns were added to my data set. The Bigsby and Schaetzl samples were
collected using the same procedures as noted above, and incorporate many of the samples in the

54

southwestern portion of the broad study area. The aggregated data set included data from 267
sites (Figure 4.2).

Figure 4.2: The distribution of loess samples across the broad study
area, including a few points taken north and south of the study area.

55

4.3 Post Processing
After psa was performed on each sample, the particle size data were exported and
formatted for spreadsheet software. Excel (Microsoft Inc., Redmond, WA) was used to plot
continuous textural curves from 0 to 2000 micrometers. Using ArcMap 10 (ESRI 2011)
software, data for each sample (e.g., loess thickness, loess quality, and percent - sand, silt, clay,
very fine sand, medium silt, fine silt, etc.) were entered into a GIS attribute table, according to
site number. The ArcMap shapefile and field notes were re-examined to clarify proper
geographic location and to confirm that the site included the target location criteria listed above.
Using the geostatistical wizard module of ArcMap (ESRI 2011), data for the various
particle size data fractions were spatially interpolated using ordinary kriging with a smoothing
factor of 0.6 (Oliver and Webster 1990; Hobbs et al. 2011; Scull and Schaetzl 2011). Separate
surfaces were created for several of the major loess texture variables in order to examine the
spatial variations within the broad study area and to identify which of the variables were most
useful for loess classification/discrimination. Map products included multiple continuous surface
maps, using isolines to illustrate loess cap thickness, loess quality, mean and modal grain size,
ratios of raw textural data, and clay-free contents of numerous particle size fractions for the
Peshekee Loess.
5. Results and Discussion
In the past, soil maps have identified loess as a surficial deposit in Baraga, Iron, and
Marquette Counties, Michigan (Flint 1971; Berndt 1988; Linsemier 1989; Schwenner 2007;
Scull and Schaetzl 2011). However, until now, the loess mantle’s thickness, textural character,
and potential source area(s) have not been the focus of mapping or research efforts. This chapter,
therefore, focuses on characterizing the loess located in Baraga, Iron, and Marquette Counties –
the region of loess formally known as the Iron County Loess Sheet and the Peshekee Loess Sheet
56

(Scull and Schaetzl 2011). I will begin by presenting and discussing how the research data set
was “cleaned-up” and the logistics behind the clean-up process. The remaining sections of this
chapter center on the loess itself. First, I will introduce and define several “eolian textural curve”
types, for samples found within the broad study area. I will then individually illustrate the eolian
textural curve type categories and their distributions within the broad study area. Next, I will
present and discuss the spatial characteristics of the Peshekee Loess. In this section, trends in
loess thickness and distributions of various silt and fine sand contents will be shown using filled
contour, continuous surface maps, created by ordinary kriging in ArcMap10. I will then discuss
the distribution of modal particle size values throughout the study area. Next I will name and
characterize four smaller, loess “sections” within the study area. Finally, by combining the eolian
textural curve type information with the trends in loess thickness, modal particle size, and the
distribution of silt and sand contents, I will suggest presumed source area(s) for the four loess
sections within the study area, and defend these interpretations.
5.1 Data Set “Clean-up”
To my knowledge, this thesis is the first study to evaluate loess deposits situated at the
extreme margins of much larger and thicker - regional scale - loess region. Here, at the margins
of the broad study area, loess deposits are thin (between ~ 20 and 80 cm), highly dissected, and
vary considerably in texture. Many sites/soils completely lack eolian materials (more common in
the northern regions of the broad study area), while other sites/soils are questionable, as to
whether or not the surficial sediment was deposited by eolian processes - judging by hand-field
texturing. Thus, many sites were not sampled due to site criteria and/or the absence of loess, and
other sites were sampled even though they were, at the time, debatable. The sediments at some
sites were uncertain because it was difficult to field-determine the surface texture; I was unable
to estimate the ratio between the 25-75 µm fraction to 250-1000 µm fraction, simply by hand57

field texturing. As a result of these sampling strategy issues, the loess sample data set required a
second look, akin to a quality control exercise. Therefore, after each sample was processed for
particle size analysis (psa), it was evaluated before being incorporated into the final data set. This
evaluation process, from here on out, is referred to as the data set “clean-up” process.
The data set clean-up process began by analyzing each sample’s continuous textural
curve, in order to assure that the sample had an abundance peak within the 25-75 µm fraction.
This type of textural signature is typical of silty eolian materials (Smith 1942; Ruhe 1954; Frazee
et al. 1970; Olson and Ruhe 1979; Smalley and Smalley 1983; Schaetzl and Hook 2008; Schaetzl
2008; Schaetzl and Loope 2008; Scull and Schaetzl 2011). A sample that had a continuous
textural curve with a mode in the 250-1000 µm fraction would not be suggestive of eolian
sediment, but instead is more likely to have been glacial outwash and/or glacial till. Samples that
were believed to have eolian signatures were found to have either a single textural mode in the
25-75 µm fraction, or a bimodal textural curve - with a dominant mode in the 25-75 µm fraction
and a secondary mode, usually in the 250-1000 µm fraction. The 25-75 µm fraction likely
signifies the eolian sediment within each sample, whereas the peak in 250-1000 µm fraction
points to contributions from the underlying parent material. Thus, the next step in the clean-up
process was to categorize or assign each sample to an eolian textural curve (ETC) type. As a
result of this analysis procedure, five dominant ETC types emerged:
Type 1 - Unimodal silt curve: one mode in the 25-75 µm fraction.
Type 2 - Bimodal silt curve: two modes, with the dominant mode in the 25-75 µm
fraction and a second mode in the 250-1000 µm fraction.
Type 3 - Bimodal sand curve: two modes, with the dominant mode in the 250-1000 µm
fraction and a second mode in the 25-75 µm fraction.
Type 4 - Silt shoulder curve: one mode in the 25-75 µm fraction, and a change in the
slope of the textural curve near the 250 µm fraction. (For example, the curve has
58

a steep negative slope throughout the 125-250 µm fraction, but it transitions to a
gradual slope near the 250-500 µm fraction. The area of gradual change is
referred to as a “shoulder” in this thesis.
Type 5 - Sand shoulder curve: one mode in the 250-1000 µm fraction, and a change in
the slope of the textural curve near the 25-75 µm fraction. (For example, the
curve has a moderately steep positive slope throughout the 25-75 µm fraction,
but it transitions to a steep positive slope in the 250-1000 µm fraction).
Some samples did not have a single mode in the 25-75 µm fraction or the 250-1000 µm fraction,
but instead had a single mode between ~ 100 µm and 125 µm. Samples with a textural mode
within the 100-125 µm fraction do not match any of the five ETCs described above. This type of
sediment texture is usually not associated with eolian processes, because it is likely too coarse
for loess. Likewise, the sediment is too fine to be typical of the local glacial till and outwash.
Samples that had a single mode within the 100-125 µm fraction did not have a typical continuous
textural curve representative of the loess deposits within the broad study area (Figure 5.1). Thus,
these samples that did not exhibit textural curve types 1-5 and had a mode within the 100-125
µm fraction were considered to have a type 0 ETC.
Type 5 and 0 samples have very similar continuous textural curves, however, type 5
samples have a mean modal particle size of ~ 243 µm whereas the type 0 samples have a mean
modal particle size of ~ 102 µm (Figures 5.1 and 5.2). Eolian textural curve types 1-5 have a
mode or shoulder curve usually in one or two size fractions, i.e., in the 25-75µm fraction and/or
the 250-1000 µm fraction. However, type 0 curves do not have a modal size in either of these
two size fractions. Type 0 samples have a mode within the 100-125 µm fraction, suggesting that
they have a different sedimentologic origin (Figures 5.1 and 5.2). After analyzing each sample’s
continuous textural curve and classifying the loess samples to an eolian textural curve type,
seven samples had type 0 ETCs (Figure 5.1). Field notes revealed that these seven samples had
either a loess quality of 0, as often indicated by a hand-field texture that felt too coarse for loess

59

or a high pit/mound density; suggesting a poorly preserved or contaminated and mixed loess
deposit (Table 5.1). Floraturbation may have mixed the original loess deposits to a point where
the deposits no longer have a peak in the 25-75 µm fraction.

Figure 5.1: Samples curves that were eliminated from the kriging data set, as a
result of not having a type 1-5 ETC. See Table 5.1 for more details.

60

Figure 5.2: Differences between type 0 and type 5 eolian textural curves. Mean
values are shown.
Table 5.1: Characteristics of type 0 eolian textural curves.
Type 0
ETC
(n=7)

Modal
particle size

cm
Min.
Max.
Mean
Standard
Deviation

Coarse silt
(35-50 µm)

Fine sand
(125-250 µm)

Fine through
medium sand
(125-500 µm)

Loess
thickness

µm

20.0
60.0
39.3

78.9
163.0
101.9

5.9
10.4
8.9

18.1
32.2
21.9

29.6
49.1
35.9

11.5

26.5

1.5

4.6

6.2

Percent Clay-free

61

Six out of the seven type 0 samples were collected within the Peshekee Highlands physiographic
region, where textural curve types ranged from 1 to 5 and field notes revealed that loess deposits
were often absent (Figure 5.3). There was only one type 0 ETC near the southern margins of the
broad study area within the Michigamme Bedrock Terrain physiographic region. An examination
of the approximate site location in ArcMap suggested that this sample was likely taken from the
shoulder slope of a bedrock-upland, instead of the appropriate summit position. The sub-optimal
landscape position may have caused post-depositional processes to incorporate more sand into
any loess that may have initially existed there, or may have exacerbated the erosion of loess;
leaving behind very little loess and/or mainly the coarsest eolian material. In sum, seven type 0
ETC samples were eliminated from the Peshekee Loess final data set because these seven
samples were not representative of eolian materials within the broad study area.

Figure 5.3: The spatial distribution of ETC types 0-5 throughout the broad study
area.
62

This thesis utilized a unique eolian soil texture sample strategy, which was necessitated
because of the landscape’s high-relief, and thin and discontinuous eolian mantle. The sampling
technique involved subjective measures, e.g., loess thickness and the presence/absence of loess,
and therefore also required an innovative data set clean-up process. After running particle size
analysis on all samples, it was evident that not all samples have similar textural curves. Thus, I
grouped the loess samples into five dominant eolian textural curve types. The five continuous
textural curves are defined as ETC type categories, one of which has been assigned to each soil
sample taken within the broad study area.

63

5.2 Characteristics of Loess within the Broad Study Area
As explained above, the soil (eolian sediment) samples collected within the broad study
area, generally exhibit one of five eolian textural curve types. Each curve has either a unimodal,
bimodal, or shoulder shape. In this section, I will define, compare, illustrate and discuss the five
ETC types found within the broad study area. In addition, I will discuss the mean modal particle
size fractions and the percentage of various particle size fractions within each ETC category.
Lastly, I will illustrate the spatial distribution of eolian textural types throughout the broad study
area.
5.2.1 Eolian Textural Curve Type Categories
5.2.1.1 Type 1: Unimodal Silt Curves
Of the 267 loess samples from the broad study area, 30 were assigned a type 1 eolian
textural curve. These samples are the siltiest loess samples; they have a unimodal curve with the
modal particle size in the 25-75 µm fraction (Table 5.2). Type 1 samples have a mean modal
particle size in the very coarse silt fraction (48.7 µm) (Figure 5.4; Table 5.2). The loess
thicknesses for type 1 ETC sites ranged from 20 cm to 95 cm, respectively, and the mean loess
thickness for these 30 sites was 40 cm (Table 5.2).
In the past, loess has commonly been defined by its size (between ~ 20 and 75 µm),
transport mechanism, source, and preservation (Smith 1942; Ruhe 1954; Frazee et al. 1970;
Olson and Ruhe 1979; Smalley and Smalley 1983; Schaetzl and Hook 2008; Schaetzl 2008;
Schaetzl and Loope 2008; Stanley and Schaetzl 2011). Type 1 samples have a range of modal
particle size fractions, from medium silt (~ 28 µm) to coarse, very fine sand (~ 79 µm), and a
mean modal particle size of 48.7 µm (Table 5.2). Generally, the modal particle size of these
samples is within the traditional loess size fraction. In terms of a source, the broad study area

64

was repeatedly glaciated during the Pleistocene Epoch (Hughes and Merry 1978; Clayton 1984;
Attig et al. 1985; Peterson 1985; Lowell et al. 1999) supplying both silt and sand sized
sediments, along with strong winds for possible eolian transport. The uplands within the broad
study area were likely the first stable landscapes, and thus, would have provided a sink and/or a
surface that could have preserved eolian sediment. Thus, type 1 samples support data found in
NRCS county-level soil surveys, Flint (1971), Scull and Schaetzl (2011), and Bigsby (2010); siltrich deposits in this area are found overlying bedrock or drift, they are often found on uplands,
and likely originated as loess.
Figure 5.4 illustrates the textural curves of all 30 type 1 samples from the broad study
area. These samples have a mean total clay-free silt content (2-50 µm) of ~ 54 percent and have a
mean medium sand content (250-500 µm) of ~ 4 percent (Table 5.2). Type 1 samples have a very
high percentage of silt and a low percentage of the medium and coarser sand fractions, compared
to sample types 2-5, respectively (Table 5.2). Sample types 2-5 have mean total clay-free silt
contents of ~ 49, ~ 31, ~ 46, and ~ 26 percent, respectively, and have mean medium sand
contents that range between ~ 10 and 20 percent (Tables 5.2, 5.3, 5.4, 5.5 and 5.6). Thus,
samples assigned a type 1 ETC are different texturally and likely in other characteristics, e.g.,
thickness and preservation, than are those samples assigned types 2-5.

65

Figure 5.4: The continuous textural curves of the 30 samples
assigned a type 1 ETC, with the blue line representing the mean
textural curve. See Table 5.2 for more details.
Table 5.2: Characteristics of type 1 eolian textural curves.
Modal
Type 1
Loess
Fine silt
Medium silt
particle
ETC
thickness
(12-25 µm) (25-35 µm)
size
(n=30)
cm
µm
Min.
Max.
Mean
Standard
Deviation

Coarse silt
(35-50 µm)

Silt
(2-50 µm)

Fine very
fine sand
(50-75 µm)

Medium sand
(250-500 µm)

Percentage clay-free

20.0
95.0
40.0

27.8
78.5
48.7

10.0
34.7
18.1

8.3
19.9
13.0

11.3
18.2
15.6

34.5
84.7
53.8

11.5
22.0
16.7

0.0
8.3
4.1

15.2

12.1

5.7

3.0

1.6

11.4

2.3

1.8

66

5.2.1.2 Type 2: Bimodal Silt Curves
Type 2 eolian textural curves are the most common textural curve type within the broad
study area (n = 122) (Table 5.3). These samples have less total clay-free silt than type 1 samples.
Type 2 samples have a mean clay-free silt percentage of ~ 48 percent, whereas type 1 samples
have a mean of ~ 54 percent (Tables 5.2 and 5.3). Most importantly, type 2 ETC samples have a
bimodal continuous textural curve, whereas type 1 samples generally have a unimodal textural
curve (Figures 5.4 and 5.5). Type 2 samples have a dominant modal particle size in the very
coarse silt fraction (35-50 µm), similar to type 1 samples, but they also have a secondary peak,
usually in the medium sand fraction (250-500 µm) (Figure 5.5). Therefore, the main attribute that
separates type 2 from type 1 samples is the presence of a second mode. Because the second mode
is in a sand fraction, type 2 samples have an overall higher percentage of sand compared to type
1 samples. For example, the type 2 samples have a mean clay-free medium sand percentage of ~
10, whereas type 1 samples have a mean of ~ 4 percent clay-free medium sand (Table 5.3).
However, there is no notable difference in loess thickness between these two curve types; both
type 1 and 2 ETC sites have a mean loess thickness of ~ 40 cm (Tables 5.2; 5.3).
The distinct bimodality of particle size distributions of type 2 samples, as shown in
Figures 5.5 and 5.6, almost certainly resulted from the admixture of coarse sands from below the
loess deposit, into the loess proper. The coarser sediments from the underling glacial drift have
presumably been mixed with the loess through various forms of bioturbation and frost action
(Schaetzl and Hook 2008). For example, soil infauna such as ants, termites, earthworms,
badgers, gophers and moles burrow in the soil as a means of finding food and providing shelter
for hibernation or reproduction (Thorp 1949; Hole 1981; Carpenter 1953; Van Nest 2002). These
infauna species often move soil sediments within a soil profile, as a result of burrowing and
ingesting soil particles (Thorp 1949; Hole 1981; Tyler et al. 2001). Many soils that lack discrete
67

horizon boundaries often owe their morphologies to long-term mixing of their upper layers by
infauna, especially worms and ants (Langmaid 1964). Within the broad study area, where loess
deposits are thin and where the lithologic discontinuity to the underlying sediment is well within
the soil profile, infauna bioturbation is likely the main process that mixes the underlying
sediments into the loess deposits.
In the field, at a small number of sites, samples were recovered from both the loess and
the underlying sediment in order to provide more detail about the sediments below the lithologic
discontinuity. As an example, at sample site 870, one textural sample was taken from the
identified loess deposit and another sample from below the loess. Figure 5.5 compares the
continuous textural curves of the loess deposit and sediment below. The loess sample (which is a
composite sample of the entire loess column) exhibits a dominant mode within the 25-50 µm
fraction, and a second peak in the 250-500 µm fraction, i.e., a type 2 ETC. However, the sample
at depth, in the drift, is unimodal, with a very low abundance of silt, and a mode within the 250500 µm fraction (Figure 5.5). The mode of the deep sample is roughly at the same location along
the size fraction as the secondary mode of the loess sample, ranging from 250 to 500 µm. Thus,
the secondary mode in the loess sample is likely attributed to in-mixing of underlying sands into
the loess, either during loess deposition or by pedoturbation processes, e.g., bioturbation,
cryoturbation, that occurred later or during soil development.
Data in Figures 5.5 and 5.6 suggest that the original loess was much siltier and had less
sand than at present, perhaps like the type 1 ETCs shown in Figure 5.4. The relatively uniform
particle-size distributions within the silt fractions also supports this conclusion, i.e., the mean
contents of the 12-25 µm fraction, 25-35 µm fraction, and 35-50 µm fraction (17.3, 11.6, and
13.1 percent, respectively) are relatively similar for type 2 samples. Likewise, type 1 samples

68

have relatively the same uniform particle-size distributions within the same silt-size fractions,
with the mean contents being 19.0, 13.0, and 15.6 percent, respectively. In addition the mode of
the sample at depth is within the 250-500 µm fraction, which is the same size fraction as the
secondary mode of the type 2 loess deposits (Figure 5.6). Therefore, because type 2 samples
have a mode within the 25-75 µm fraction, similar to type 1 samples, this research suggest this
25-75 µm fraction likely originated as loess. Conversely, particles within the 250-500 µm
fraction likely did not originate as loess, but instead originated as glacial till or outwash, due to
their sediment size.

Figure 5.5: The continuous textural curves of two samples taken from site 870; the
yellow line showing the continuous textural curve of the loess and blue showing the
sample’s continuous texture at depth.

69

Figure 5.6: The continuous textural curves of the samples assigned a
type 2 ETC, with the blue line representing the mean textural curve.
See Table 5.3 for more details.
Table 5.3: Characteristics of type 2 samples.
Modal
Type 2
Loess
Fine silt
particle
ETC
thickness
(12-25 µm)
size
(n=122)
cm
µm
20.0
25.5
7.2
Min.
85.0
74.6
34.3
Max.
40.6
41.5
17.3
Mean
Standard
13.4
7.6
4.3
Deviation

Medium silt
(25-35 µm)

Coarse silt
(35-50 µm)

Silt
(2-50 µm)

Percentage clay-free
8.9
32.0
19.7
79.3
13.1
48.8

7.4
17.4
11.6
2.1

2.5
70

8.8

Fine very
fine sand
(50-75 µm)

Medium sand
(250-500 µm)

9.0
22.1
13.2

1.9
18.0
10.3

2.6

4.0

5.2.1.3 Type 3: Bimodal Sand Curves
A total of 28 samples were assigned a type 3 eolian textural curve; these samples, like
type 2 ETCs, have a bimodal continuous textural curve. Type 3 samples have a mean clay-free
silt percentage of ~ 31 percent, and have a mode in the very coarse silt fraction (~ 54 µm).
However, type 3 samples are different from type 2 samples, because their dominant mode, or
larger peak is in the sand fraction (mean of ~ 280 µm), as opposed to the silt fraction, like type 1
and 2 textural curves. Thus, type 3 samples appear to represent mainly sandy sediment with
varying degrees of silt admixture, as opposed to type 2 samples, which appear to be silty
sediment with sandy admixtures (Figures 5.6 and 5.7). Type 3 samples have a mean clay-free
medium sand content of ~ 20 percent, whereas types 1 and 2 samples have a mean of ~ 10
percent and ~ 4 percent (Tables 5.2, 5.3 and 5.4). Type 3 ETC sites also had a thinner mean loess
thickness (~ 33 cm) compared to ETC type sites 1 and 2 (~ 40 cm), which could help explain
their higher sand content percentages (Tables 5.2, 5.3 and 5.4). The thinner loess deposits
generally have a smaller amount of silty sediment available for mixing into the initial sandy
sediment, than do the thicker loess deposits. The large sand mode in type 3 samples is typically
between 250 and 500µm and is larger than their silt mode. At these sites, the loess deposits are
likely thin enough that they have been thoroughly mixed into the sandy outwash below, as is
shown on the particle size curves by a small, secondary mode within the silt fraction (Figure
5.7).

71

Figure 5.7: The continuous textural curves of the samples assigned a
type 3 ETC, with the blue line representing the mean textural curve.
See Table 5.4 for more details.
Table 5.4: Characteristics of type 3 samples.
Modal
Type 3
Loess
Fine silt
particle
ETC
thickness
(12-25 µm)
size
(n=28)
cm
µm
0.0
27.7
5.6
Min.
80.0
415.3
23.0
Max.
32.5
280.5
14.2
Mean
Standard
14.9
80.3
4.5
Deviation

Medium silt
(25-35 µm)

Coarse silt
(35-50 µm)

Silt
(2-50 µm)

Percentage clay-free
6.6
22.0
10.6
43.8
8.5
30.6

5.4
9.2
7.1
0.9

0.8
72

5.1

Fine very
fine sand
(50-75 µm)

Medium sand
(250-500 µm)

6.4
11.8
9.8

14.7
30.0
20.1

1.4

3.2

5.2.1.4 Type 4: Silt Shoulder Curves
The type 4 eolian textural curve is the second most common continuous textural curve
type within the broad study area (n=68). Type 4 samples are a hybrid between a type 1 (unimodal
silt curve) and a type 2 (bimodal silt curve). Type 4 samples have a unimodal continuous textural
curve with the modal particle size in the 25-75 µm fraction, thus, resembling a type 1 sample
(Figures 5.4 and 5.8). However, similar to type 2 and 3 samples, type 4 samples have more sand
than do type 1 samples. Type 4 samples are comparable to type 2 samples with both having a
mode within the 25-75 µm fraction and an increase in abundance within the 250-500 µm
fraction. The difference between the type 4 and 2 samples is the amount of sand; type 4 samples
have a shoulder curve within the 250-500 µm fraction, and thus usually have less sand contents
compared to the type 2 samples, which have a “true” secondary mode within the 250-500 µm
fraction (Figures 5.6 and 5.8). For this reason, type 4 ETCs are described as having a silt
shoulder curve – a silt mode but a shoulder in the sand fraction.
Type 4 samples are unimodal in the 25-75 µm fraction with a change in slope usually
near the 125 µm fraction where the gradient of the line changes from a steep negative slope, to a
more gradual/gentle slope throughout the 250-500 µm fraction (Figure 5.8). Sun et al. (2004)
described a shoulder curve as a “S curve” when analyzing continuous textural curves. Type 4
samples have a mean total clay-free silt content of ~ 46 percent, but as expressed in their
shoulder curve, they also have a mean clay-free medium sand content of ~ 9 percent (Table 5.4).
Type 4 sample sites have a mean loess thickness of ~ 34 cm, which is similar to type 3 sample
sites (mean of ~ 33 cm), but type 4 sites exhibit thinner loess than both type 1 and 2 sites (mean
of ~ 40 cm) (Tables 5.2, 5.3, 5.4 and 5.5).

73

Figure 5.8: The continuous textural curves of the samples assigned a
type 4 ETC, with the blue line representing the mean textural curve.
See Table 5.5 for more details.
Table 5.5: Characteristics of type 4 samples.
Type 4
Loess
Modal
Fine silt
ETC
thickness particle size (12-25 µm)
(n=68)
cm
µm
20
27.4
7.5
Min.
63
79.1
27.3
Max.
34.4
51.0
16.1
Mean
Standard
11.2
12.1
5.3
Deviation

Medium silt
(25-35 µm)

Coarse silt
(35-50 µm)

Silt
(2-50 µm)

Percentage clay-free
9.2
29.6
16.5
65.3
12.9
45.7

7.1
14.9
10.8
1.9

1.6
74

8.5

Fine very
fine sand
(50-75 µm)

Medium sand
(250-500 µm)

10.6
18.6
14.4

4.6
16.1
9.0

1.7

2.6

Sample site 1086 is another location within the broad study area where samples were
collected from both the identified loess deposit and the subjacent sediment below the loess.
Figure 5.9 compares the continuous textural curves of the loess deposit with the sediment below.
The loess sample has a mode within the 25-75 µm fraction and a shoulder near the 250-500 µm
fraction (thus, it has a type 4 ETC), whereas the sample at depth is unimodal, with a low
abundance of silt and a mode within the 250-500 µm fraction (Figure 5.9). These curves suggest
that type 4 samples may have originally been much siltier, but have been mixed with the coarser
underlying parent material, similar to type 2 samples, but the process has not occurred to as great
an extent.

Figure 5.9: The continuous textural curves of two samples taken from site 1086; the
yellow line is the continuous textural curve of the loess and blue is the continuous textural
curve for the sediment at depth.

75

5.2.1.5 Type 5: Sand Shoulder Curves
The type 5 eolian textural curve category is the least common textural curve within the
broad study area (n=12). Similar to type 4 samples, type 5 sample have a unimodal shoulder
curve, but are categorized differently because the sample’s modal particle size is within the 250500 µm fraction (medium sand). Type 5 samples do not have a true second mode the slope of the
textural curve usually changes within the 25-75 µm fraction. Hence, these samples are described
as having a sand shoulder curve (Figure 5.10). Type 5 textural curves exhibit a gradual and
positive slope throughout the 25-75 µm fraction, documenting that these samples have an
abundance of particles within the 25-75 µm fraction (Figure 5.10). Type 5 samples have a mean
total clay-free silt content of ~ 26 percent, but they also have the highest percentage of total clayfree sand (mean of ~ 74 percent) (Table 5.6). The type 5 ETC sample sites have the smallest
mean loess thickness (~ 28 cm). These thin loess deposits have likely experienced more mixing
during and after loess deposition, than did the thicker loess deposits. Moreover, vectors for
mixing the loess with the underlying parent material are likely more efficient in thinner loess
deposits than in thicker loess deposits. Therefore, the continuous textural curves of type 5
samples show that these samples are dominantly sandy sediments with a low abundance of
sediment from the 25-75 µm fraction.

76

Figure 5.10: The continuous textural curves of the samples assigned
a type 5 ETC, with the blue line representing the mean textural curve.
See Table 5.6 for more details.
Table 5.6: Characteristics of type 5 samples.
Modal
Type 5
Loess
Fine silt
particle
ETC
thickness
(12-25 µm)
size
(n=12)
cm
µm
0.0
151.3
9.0
Min.
40.0
324.8
24.4
Max.
27.7
242.6
14.2
Mean
Standard
10.0
60.9
3.7
Deviation

Medium and
Coarse silt
(25-50 µm)

Medium sand
(250-500 µm)

Sand
(50-2000
µm)

8.7
16.9
13.1

Percent Clay-free
17.6
16.5
25.0
35.0
22.0
26.0

13.8
30.6
20.0

65.0
83.5
74.0

2.1

5.2

2.3

4.3

5.2

77

Silt
(2-50 µm)

Fine sand
(125-250 µm)

In summary, loess has various textural signatures throughout the broad study area. In this
thesis, I have divided loess samples from the braod study area into five ETC categories.
Nonetheless, there are two dominant eolian textural curve types:
(1) Those with high contents within the 25-75 µm fraction and a secondary peak (or
shoulder) within the 250-500 µm fraction (types 1, 2 and 4)
(2) Those with high contents within the 250-500 µm fraction and a secondary peak (or
shoulder) within the 25-75 µm faction (types 3 and 5).
Eolian textural curve types 1, 2, and 4 have a mean modal particle size within the 42-51 µm
fraction (Figure 5.11). Conversely, eolian textural curve types 3 and 5 have a mean modal
particle size within the 243-281 µm fraction (Figure 5.11).

Figure 5.11: Summary diagram of the five ETC types and their
mean continuous textual curves.
78

5.2.1.6 Distribution of Sites with Different Textural Curves
Maps of the spatial distribution of ETC types within the broad study area could
potentially suggest a relationship between eolian textural type, loess thickness, and distance from
presumed source area(s). Figure 5.12 shows the sample locations within the broad study area,
with each sample labled by their assigned ETC type. It is immediately evident that the map lacks
a promiant spatial trend with respect to ETC type, across the broad study area. However, there
are a few noteworthy spatial trends of ETC type curves within smaller parts of the broad study
area, e.g., there are small concentrations of type 1 samples in the center of the broad study area,
particularly east of the Baraga Plains (Figures 5.12 and 5.13). Both type 2 and 4 ETCs (which
have the second highest total silt contents) are mapped throughout the entire broad study area
(Tables 5.3 and 5.5; Figure 5.12). Type 2 ETCs dominate the southern portion of the broad study
area and comprise ~ 82 percent of the samples within Iron County (Figures 5.12 and 5.13),
whereas type 4 samples are mostly located in Baraga and Marquette Counties. Lastly, both type
5 and type 3 samples, which have high sand contents and low silt contents, are mainly mapped in
the high-relief and northern regions of the broad study area; all type 5 samples (unimodal sand
shoulder curve) are mapped within the high-relief Peshekee Highlands (Figures 5.12 and 5.13).
Figure 5.14 shows the distribution of samples with high silt contents (roughly the 25-75 µm
fraction) vs. those with high sand contents (roughly the 150-500 µm fraction).

79

Figure 5.12: The distribution of soil (loess) samples taken within the broad study area, symbolized by their eolian
textural curve type.
80

Figure 5.13: The distribution of ETC types plotted against a background of local relief, within the broad study area.
81

Figure 5.14: The distribution of eolian textural type curves with generally high silt contents and low sand contents (labeled in
red), and high sand contents and low silt contents (labeled in green).
82

Eolian textural curve types with high silt contents (types 1, 2, and 4) are representative of
~ 85 percent of the sites within the broad study area (Figure 5.14). This finding confirms that
loess is widespread across the broad study area, and validates the NRCS soil survey data and the
conclusions of Scull and Schaetzl (2011). Although samples with high silt contents are located
across the entire, broad study area, they particularly dominate the southern and western regions
(Figure 5.14). Conversely, samples with high sand contents (types 3 and 5) are uncommon
(comprising ~ 15 percent of the sites) and are mainly mapped in the northeastern regions of the
broad study area (Figure 5.14). Figure 5.15 shows that ~ 88 percent of the “high sand” samples
(types 3 and 5) are located within the high-relief Peshekee Highlands. The high sand content
samples have a mean loess thickness of ~ 30 ± 13.8 cm, whereas the high silt content sample
sites have a mean loess thickness of ~ 38 cm (Tables 5.2, 5.3, 5.4, 5.5 and 5.6). Using a t-test
(assuming unequal variances), the loess thicknesses associated with high silt and high sand
content groups were significantly different at P < 0.001.

83

Figure 5.15: The distribution of eolian textural type curves with high silt and high sand contents, plotted against relief
within the broad study area.
84

In summary, type 1 samples are the siltiest (with a mean of ~ 54 percent within the 25-75
µm fraction) and are located on sites with the thickest loess (a mean loess thickness of ~ 40 cm).
The most prevalent textural curve type sampled within the broad study area is type 2 (bimodal,
with a dominant mode within the 25-75 µm fraction). Second most common is the type 4 ETC.
Like the type 2 samples, the modal particle size of the type 4 samples is within the 25-75 fraction
with a second peak or shoulder within the 250-500 µm fraction. Eolian textural curve types 3 and
5 are not as common, which could be due to the field sampling procedure or a bias in field handtexturing. For example, I generally avoided sites that had a surface texture similar to ETC types 3
and 5 because it was difficult to determine whether or not a site/sample had a bimodal texture in
the field. Figure 5.15 shows that ETC types 3 and 5 are dominantly mapped within the highrelief Peshekee Highlands and Tables 5.3, 5.4, 5.5, and 5.6 show that these loess deposits are
generally thinner (mean of ~ 30 cm). Where loess depositional events were short-lived, original
surface sediments were likely being mixed during loess deposition. Moreover, after deposition,
where loess deposits are thinner, vectors for mixing underlying sediments with the loess, such as
bioturbation and cryoturbation, are likely more efficient than in thicker loess deposits. Thus, in
the northern regions of the broad study area where loess deposits are typically thinner, it is more
common for loess samples to be dominantly sandy (Figure 5.15).

85

5.3 Spatial Characteristics of Loess within the Study Area
In the past, loess deposits have been associated with a wedge-shape, with both loess
thickness and particle size decreasing from the loess source (Smith 1942; Ruhe 1954; Frazee et
al. 1970; Olson and Ruhe 1979; Muhs and Bettis 2000). This section will examine the spatial
characteristics of loess within the study area - defined by the extent of the loess sample points and determine whether the loess characteristics are similar to the spatial trends of thicker loess
deposits in the Great Lakes Region. I will first focus on three traditional characteristics of loess
by showing and discussing interpolated maps of (1) loess thickness, (2) silt distribution, and (3)
sand distribution. Then, because many of the loess deposits within the study area are comprised
of two particle size fractions; 25-75 µm fraction and the 250-500 µm fraction, maps of these
fractions will be discussed.
5.3.1 Loess Thickness Data
Within the study area sampled sites exhibited loess thicknesses that ranged between
approximately 80 and 20 cm (Figure 5.16). Loess is generally thickest (> 80 cm thick) in
southern and southwestern Iron County. It progressively thins towards northeastern Marquette
County, where loess deposits are commonly < 20 cm thick (Figure 5.16). This thickness range is
similar to reports of loess thicknesses in the western Upper Peninsula by Scull and Schaetzl
(2011) and Bigsby (2010). The interpolated loess thickness trend (Figure 5.16) also matches well
with the loess thicknesses produced using NRCS soils data (Figure 3.12). However, anomalously
thick areas of loess, not before documented, do occur in the western and north-central regions of
the study area. For example, east of the Baraga Plains, loess is estimated to be between 48 and 58
cm thick, compared to the surrounding area where loess thicknesses are estimated to be between
33 and 36 cm (Figure 5.16). Similarly, in the north-central region of the study area, near the
Peshekee River, there is a small area of loess estimated to range between 48 and 58 cm in
86

thickness, whereas loess deposits in the surrounding areas are generally between 12 and 21 cm
(Figure 5.16). Schaetzl and Liebens (1992, 1993) reported on the thick loess near the Peshekee
River, although this work was never published. With the exception of a few anomalously thick
areas, the loess deposits within the study area generally exhibits a wedge-shaped thickness trend
(Figure 6.16), with the thickest loess in the south, and thinning towards the north. This trend is
consistent with Midwestern loess literature (Smith 1942; Olson and Ruhe 1979; Muhs and Bettis
2000; Schaetzl and Hook 2008).

Figure 5.16: Interpolated loess thickness map for the study area.
87

5.3.2 Particle Size Analysis of the Peshekee Loess
5.3.2.1 Spatial Trend in Silt Contents
Samples with the highest total clay-free silt contents (2-50 µm fraction) generally occur
in two regions of the study area: (1) in central Iron County, near the southern limits of the study
area, and (2) in southern Baraga County, in the central and western regions of the study area
(Figure 5.17). The distribution of clay-free fine silt (2-25 µm) has a similar spatial trend as the
distribution of total silt contents, with the highest percentages occurring in the southern, central
and western regions of the study area (Figure 5.18). The spatial distribution of silt contents
changes, however, when mapping the distribution of clay-free medium silt (25-35 µm fraction)
and coarse silt (35-50 µm fraction) (Figures 5.19 and 5.20). Figure 5.19 shows that the southern,
central, and western regions of the study area have the highest percentages (~ 20) of medium silt,
whereas Figure 5.20 illustrates that the southeastern regions of the study area (in western
Marquette County) has the highest percentages (~20) of coarse silt. The spatial distribution of
medium silt is a hybrid between the distribution of fine silt and coarse silt.
The loess surface texture map (Figure 3.11), created using NRCS data, matches well with
the distribution of silt contents in the southern regions of the study area – that is, loessal soils
here are dominantly being mapped with soils that have a silt loam surface texture, e.g., the
Goodman and Wabeno soil series. The NRCS loess surface texture map also shows that sandier
loessal soils (e.g. Keewaydin series) are usually mapped in the eastern regions of the study area;
this parallels the higher coarse silt percentages that occur on the eastern regions of the study area
(Figure 5.20). Conversely, the increased siltiness of the loessal soils in southern Baraga County
and directly east of the Baraga Plains is not as evident from the NRCS surface texture map, as
these loessal soils are mainly mapped as Champion which has a cobbly silt loam surface texture

88

(Figures 3.11 and 5.17). Champion soils are widely mapped throughout the region, even in areas
that do not have a cobbly silt loam surface texture.

Figure 5.17: Interpolated map of clay-free silt contents in the study area.

89

Figure 5.18: Interpolated map of clay-free fine silt contents in the study area.

90

Figure 5.19: Interpolated map of clay-free medium silt contents in the study area.

91

Figure 5.20: Interpolated map showing the distribution of clay-free coarse silt contents.

92

The loessal soils in eastern Baraga County are mainly mapped within the Champion,
Michigamme, Petticoat, and Dishno series, all of which have cobbly silt loam surface textures.
The interpolated loess thickness map suggests that loess deposits in eastern Baraga County tend
to be thinner compared to northern Iron County or the southern regions of the study area (Figure
5.16). In periglacial environments, where soils are exposed to intense frost action, coarse
fragments are often pushed or pulled to the surface (Corte 1963; Inglis 1965). The term
upfreezing is used when water in soils expands as it freezes, and ice lenses develop under clasts
within the soils. Thus, the soils expand when they freeze; as a result, cavities develop under
individual clasts, and when the soil thaws, the cavities can become partially infilled with
sediment. The infilled cavities then prevent these clasts from returning to their original location,
resulting in a net upward movement of large clasts. High moisture contents and large clasts favor
high rates of upfreezing (Anderson 1988). Thus, upfreezing is optimal in regions with a cool
climate and where soils are forming in dominantly silty sediment, with few, but large, clasts.
Within the study area, where thin loess deposits are prevalent and frost can penetrate > 50
cm (Schaetzl and Isard 1996); upfreezing has likely mixed cobbles from the underlying parent
material with the overlying silt-rich loess, forming soils that have cobbly silt loam surface
textures. Thus, there is an excellent correlation between the NRCS loessal soil surface texture
map and the interpolated loess thickness map with the cobbly surface texture modifier, i.e.
usually where loess is estimated to be thin and silty (Figures 3.11 and 3.12) the NRCS had
mapped a loessal soils with a “cobbly” surface texture modifier (e.g., Champion, Michigamme,
Petticoat, and Dishno) (Figure 3.11). In the southern regions of the study area, where loess
deposits are generally thicker, loessal soils are dominantly mapped without a “cobbly” surface

93

texture modifier (e.g., Goodman and Wabeno) (Figures 3.11 and 5.16). Here, cryoturbation
likely has been less able to mix cobbles from the underlying drift into the loess deposits.
5.3.2.2 Spatial Trends in Sand Contents
For the study area, interpolated maps showing the distribution of sand contents of the
loess deposits suggests that in western and northern Marquette County, they have the highest
sand contents (Figures 5.21, 5.22, 5.23 and 5.24). For example, Figures 5.21 shows that loess
deposits in western Marquette County generally have the highest percentages (~ 22) of sand
contents within the 50-75 µm fraction. Moreover, Figure 5.22 shows that these deposits in
western Marquette County also have the highest contents within the 75-125 µm fraction. This
research indicates that northern and western Marquette County generally has coarser loess than
does Iron and Baraga Counties. These data support the decision by NRCS to develop a new soil
series for the Marquette County soil survey. In Marquette County, loessal soils dominantly have
a cobbly fine sandy loam surface texture and are mapped in the Keewaydin series, whereas in
Baraga and Iron counties loessal soils dominantly have a silt loam or cobbly silt loam surface
texture and are mapped in the Goodman, Wabeno, Champion, Michigamme, Petticoat, and
Dishno series.
A comparison of the NRCS loessal soils surface texture map (Figure 3.11) to the
interpolated loess thickness map (Figure 5.16) shows that the Keewaydin series is usually
mapped where loess deposits are thin (between ~ 21 and 28 cm). Loess mixing vectors, such as
bioturbation and cryoturbation, likely are more efficient in incorporating the coarser underlying
sediments into these thin loess deposits, resulting in a coarser loess deposit. However, generally
the finer (eolian) fractions of the loess deposits in Marquette County are larger in particle size,
i.e., have a larger modal value, than the loess deposits in Baraga and Iron Counties. The
relatively coarser loess deposits could have been caused by different source areas, such as more
94

local sources, compared to the loess in Baraga and Iron Counties, which could have had more
distant source areas.

Figure 5.21: Interpolated map showing the distribution of clay-free fine, very fine
sand contents.

95

Figure 5.22: Interpolated map showing the distribution of clay-free coarse, very fine
sand contents.

96

Marquette County not only has some of the coarsest loess within the study area (when
examining the 50-75 µm fraction only), but loess samples here generally also have the highest
percentages of fine, medium, and coarse sand contents. Particles within the 125-1000 µm
fraction have probably been mixed via bioturbation and/or cryoturbation into the loess column
from the underlying sandy drift material. Interpolated maps show that northwestern Marquette
County loess deposits generally have the highest percentages of contents within the 125-175 µm,
175-250 µm, and 250-500 µm fractions (Figures 5.23, 5.24 and 5.25). Loess deposits in
northwestern Marquette County are estimated to have ~ 12 percent fine, fine sand (125-175 µm
fraction), ~ 13 percent coarse, fine sand (175-250 µm fraction), and ~ 31 percent medium sand
(250-500 µm fraction) (Figures 5.23, 5.24 and 5.25). Conversely, the loess samples in Baraga
and Iron Counties generally have less than 5 percent of their loess contents within these size
fractions. Loess deposits in the northern regions of the study area are generally thinner (Figure
5.16), facilitating the likelihood that the loess has been mixed with the underlying parent
material. Moreover, the spatial pattern of samples with high sand contents matches well with the
distribution of ETC types. Loess deposits in northwestern Marquette County usually have a
textural curve that is bimodal with one mode in the 250-1000 µm fraction or they are unimodal
with a mode in the 250-1000 µm fraction (i.e., ETC types 3 and 5) (Figures 5.14 and 5.15). The
interpolated maps (Figures 5.21, 5.22, 5.24 and 5.25) show a distribution of sand contents within
the loess area that agrees with the NRCS loessal soil series maps. Loessal soils in northwestern
Marquette County are generally mapped as Keewaydin, which is described as having a cobbly,
fine sandy loam surface texture. Goodman, Wabeno, Champion, Michigamme, Petticoat, and
Dishno soils are mainly mapped in Baraga and Iron Counties; they are loessal soils with silt loam
or cobbly silt loam surface textures (Figures 3.11, 5.16, 5.23, 5.24 and 5.25). In sum, both the

97

interpolated sand contents maps and the NRCS loessal soil series maps show that loess deposits
in the northern regions of the study area are generally thinner and texturally coarser than the
deposits in the western and southern regions of the study area.

Figure 5.23: Interpolated map showing the distribution of clay-free fine, fine sand
contents.

98

Figure 5.24: Interpolated map showing the distribution of clay-free coarse, fine sand
contents.

99

Figure 5.25: Interpolated map showing the distribution of clay-free medium sand
contents.

100

5.3.2.3 Spatial Trends in Modal Particle Size
As a whole, the study area has thin (mean thickness of ~ 37 cm) loess deposits (Figure
5.16). Generally, the loess deposits have bimodal continuous textural curves, with particle size
peaks in both within the 25-75 µm fraction and in the 250-500 µm fractions (Figures 5.4, 5.6,
5.7, 5.8 and 5.10). The textural bimodality of these loess deposits, particularly the wide disparity
in sizes between the two modes, suggests that the sediments making up the loess deposits are
polygenetic. I am suggesting that the silt size particles likely originated as loess, while the
coarser sandy particles originated as glacial outwash or till; after loess deposition, the coarser
underlying sediments have been mixed with and into the overlying loess. The modal particle size
within each particle size fraction (e.g., silt and sand) is one of the most important characteristics
of these loess deposits, and thus, should be analyzed/mapped separately in order to identify
potential source areas. Therefore, two maps were compiled in order to further support the
hypothesis that the silt or fine, very fine sand contents of the soil samples originated as loess, and
the coarser contents are of a different origin.
Within the study area, ~ 95 percent of the samples have a modal particle size within the
26-99 µm fraction. Additionally, ~ 60 percent of the samples have a bimodal continuous texture
curve, with one peak in the 26-99 µm fraction and another peak in the 200-780 µm fraction. If
the contents within 26-99 µm fraction originated via eolian processes, as would seem logical,
then this size fraction should display a clear and explainable spatial trend in particle size, from a
potential source area. Conversely, lack of a spatial trend should be evident when interpolating
particle size distributions within the 200-780 µm fraction, because these particles most likely did
not originate via eolian processes, but instead by direct glacial processes.
The map of “silt modes” was compiled by assigning each sample its modal value within
the 26-99 µm fraction, as reported by the Malvern Mastersizer 2000. This mode was then added
101

to the GIS shapefile (e.g., Figure 5.26). The “sand mode” map was produced in similar fashion,
using the data from the 200-780 µm fraction (e.g., Figure 5.27). However, not all samples were
used for each modal map, because some samples have a shoulder curve, and thus, only one
“true” mode. Nonetheless, each sample has at least one mode, and therefore each sample point
was assigned to at least one mode in the data set. Sample types 1-4 were used to generate the silt
mode map (Figure 5.28), and sample types 2, 3 and 5 were used to generate the sand mode map
(Figure 5.29).

Figure 5.26: As an example, this type 3 sample taken from the study area was assigned a
48 µm silt mode and a 415 µm sand mode, for the two modal maps.

102

Figure 5.27: As an example, this type 2 sample taken from the study area was assigned a
48 µm silt mode and a 300 µm sand mode, for the two modal maps.

103

Figure 5.28: A map illustrating modal particle size of eolian sediment in the study area,
using sample modes within the 26-99 µm fraction. Samples without a “true” mode within
the 26-99 µm fraction, such as type 5 ETCs, were not used in this kriging exercise.
.

104

Figure 5.29: A map illustrating modal particle size of coarser sediment in loess samples,
within the study area. Samples without a “true” mode in the 200-780 fraction, such as
types 1 and 4 ETCs, were not used in this kriging exercise.
.

105

The silt mode map (Figure 5.28) illustrates a prominent southwest to northeast spatial
trend, with the southern regions of the study area having samples with the finest modal particle
size (~ 26 µm), and the northeastern region having samples with a coarser mode (~ 99 µm).
Loess deposits generally become progressively finer in texture away from the source area(s),
thus, the silt mode map points to a predominantly northeastern source area for the Peshekee
Loess.
The sand mode map (Figure 5.29) depicts no clear spatial trend, and instead shows
several regions with coarser sand modes (~ 780 µm). The sand mode map provides evidence that
the coarser contents within the loess samples were not transported via eolian processes. Many of
the regions with large sand modes are located in the southern portion of the study area, where
loess samples are also known to have the finest silt modes. Within the study area, it is possible to
have areas where there is a high percentage of silt or fine, very fine sand contents in addition to
medium and coarse sand contents because loess deposits in this area often have a bimodal
textural curve (e.g., type 2 ETCs). The modal maps provide additional evidence that the
underlying, coarser sediments, likely originated before the loess, and have since been mixed into
the loess.
Loess deposits are generally thicker and coarser near their source and become thinner and
finer-textured downwind (Smith 1942; Frazee et al. 1970; Olson and Ruhe 1979; Muhs and
Bettis 2000). The silt modal particle size map suggests a dominantly northern source area, which
stands in opposition to the interpolated loess thickness map, that suggests a southern source
area(s). In this section, I will attempt to explain this apparent disparity.
Loess deposits are usually thickest in the southern regions of the study area and
progressively thin towards the north, whereas modal particle size is coarsest in the north and

106

becomes progressively finer towards the south (Figures 5.16 and 5.28). If the dominant source
area(s) is (are) in the southern parts of the study area, or even areas beyond it to the south, as the
loess thickness map suggests, then silt modal particle size should progressively become finer
towards the north. However, the silt mode map shows that the particle size mode of loess
samples within the study area gradually gets coarser towards the north. These seemingly
contradictory spatial trends could be a result of multiple and heterogeneous source areas within
the study area, i.e., there are actually multiple loess sections within the study area. I believe that,
despite the apparent conflict between the loess thickness and modal particle size trend maps, the
loess deposits in the southern regions of the study area were likely sourced from landscapes
outside (and south) of the study area. The loess deposits here, particularly those in Iron County,
probably originated from more distant sources. Evidence for this is the dominantly smaller
particle size fraction (~ 26 µm) (Bigsby 2010). I also suggest that sandy, but thin, loess deposits
in the northern regions were sourced from more localized landscapes, which would explain the
dominantly coarser particles (~99 µm). This topic is discussed in more detail below.
5.4 Peshekee Loess Individual Sections and Potential Source Areas
Spatial characteristics of loess deposits not only aid in the identification of a sediment as
eolian, but can also point to provenance (Schaetzl and Hook 2008). At a large scale, the study
area shows several spatial trends in loess thickness, silt and sand contents, and modal particle
size, even though some of these trends appear to be contradictory across the larger region
(Figures 5.16, 5.17, 5.21 and 5.28). One way to alleviate this potential problem is to examine the
loess characteristics at a larger scale and determine whether or not the textural character and
thickness of individual loess deposits varies between these smaller regions within the larger loess
region. Here, within the study area, this approach may be advantageous, because the loess is
highly variable spatially, and loess is absent at many sites. Analyzing the Peshekee Loess at a
107

larger scale could potentially tease out the conflict between modal particle-size and loess
thickness, and point to more than one source area for the loess.
In this thesis, each smaller loess region, within the larger study area, is referred to as a
“section.” The study area was divided into four sections, in order to better differentiate the types
of loess deposits within it, and to suggest possible, unique source areas for each section. The four
sections were named for their unique loess character and for a physical or cultural feature of
prominence that is within that section. From south to north, the four sections are Amasa,
Covington, Republic, and Champion (Figures 5.30 and 5.31). The delineation of these sections
was largely driven by spatial patterns of loess thickness, modal particle-size, percentage of
contents within the 25-75 µm fraction (i.e., the loess), and the percentage of contents within the
125-500 µm fraction (i.e., the in-mixed sand) (Table 5.7; Figures 5.16, 5.17, 5.21 and 5.28).
Below is a discussion of the specific criteria and logic that were used to define the section
boundaries.
The boundary between the Amasa and Covington sections follows the isoline that
represents 63 percent clay-free silt content (2-50 µm fraction). This boundary was used because
the loess samples within these two sections generally have a high percentage of clay-free silt, in
comparison to the Champion and Republic sections, which have less. The Champion section
boundary, where it abuts the Covington and Republic sections, follows the isoline that represents
39 percent clay-frees silt content (Figure 5.30). The Republic section boundary, where it is
adjacent to the Amasa and Champion sections, is marked by the isoline that represents 16 percent
clay-free fine, very fine sand content (Figure 5.31).
Not all areas within the study area fall within a section boundary, proper, using the above
boundary definition criteria. Because it was deemed necessary for all areas within the study area

108

to be classified within a section, some of the section boundaries were therefore slightly adjusted.
For areas where there exists a gap between two section boundaries, the gap was divided into two
roughly equal parts; half of that area was given over to each of the two bounding sections. This
“cut” operation was done in a GIS, by visual inspection. Moreover, for discussion purposes, only
the samples that are representative of the loess in that section (i.e., those labeled with a white
point) are illustrated in Figures 5.30, 5.31, 5.32, 5.36, 5.40, and 5.41. Most of these samples
occur within the “original” section boundaries. The remaining segments of this chapter will focus
on these four sections. I will illustrate and discuss the characteristics of the loess deposits within
the four sections, and suggest possible source area(s) for them.
Table 5.7: Characteristics of the four major loess sections within the study area.
Amasa
Section (n=37)

Covington
Section (n=77)

Republic
Section (n=19)

Champion
Section (n=126)

Most common
ETCs

Type 2
(89.2%) and
type 1 (5.4%)

Type 2 (42.9%)
and type 4
(35.1%)

Type 2
(52.6%) and
type 1 (36.8%)

Type 2 (36.5%)
and type 4
(29.4%)

Mean Loess
thickness

52.4 ± 13.1 cm

36.9 ± 13.9 cm

42.5 ± 11.8 cm

32.9 ± 10.3 cm

Mean modal
particle-size
within the 26-99
µm fraction

38.6 µm

40.2 µm

49.8 µm

52.9 µm

Mean %
contents within
the 25-75 µm
fraction

38.5%

39.6%

47.2%

33.0%

Mean %
contents within
the 250-500 µm
fraction

9.9%

8.2%

5.5%

13.4%

109

Figure 5.30: Map of interpolated clay-free silt contents and a comparison of loess samples’
textural curves within four loess sections, in the study area.
110

Figure 5.31: Map of interpolated clay-free fine, very fine sand (50-75 um) contents and a
comparison of loess samples’ textural curves within four loess sections, in the study area.
111

5.4.1 Amasa Section
The loess deposits near the southern margins of the study area are some of the thickest
and siltiest deposits within the study area (Tables 5.7 and 5.8). I have named these loess deposits,
in Iron County, the Amasa section. The most common eolian textural curve in this section is the
type 2; ~ 89 percent (n= 37) of the samples recovered within this section exhibited a type 2 ETC
(Table 5.7). The most common eolian textural curve for the remaining sections was also a type 2,
however their percentage of samples with this type curve are all lower and range between ~ 53
and 37 percent. Moreover, the Amasa section has the greatest loess thicknesses, with a mean
thickness of 52.4 ± 13.1 cm, whereas to the remaining three sections have mean loess thicknesses
that range between ~ 33 and 43 cm (Tables 5.7 and 5.8; Figure 5.16). Several continuous textural
curves selected from the Amasa section show that loess deposits here usually have a modal
particle-size within the 25-50 µm fraction, and small peak in the 250-500 µm fraction (Figure
5.32). The mean modal particle-size (within the 26-99 µm fraction) in the Amasa section is 38.6
µm, which is a much finer mean modal particle-size, as compared to the Covington, Republic
and Champion sections, with a mean ranging between 40.2 and 52.9 µm (Tables 5.7 and 5.8).

112

Table 5.8: A table ranking the four loess sections, within the study area, by calculating
each of the sections mean loess thickness and the mean of several particle-size fractions.
Loess
characteristic

Amasa
Section
(n=37)

Covington
Section
(n=77)

Republic
Section
(n=19)

Champion
Section
(n=126)

Rank
(Red being the highest and pink being the lowest)

Mean:
Loess
thickness

52.4%

36.2%

42.5%

32.9%

Medium silt
(25-35 µm)

12.3%

12.4%

12.6%

9.1%

Coarse silt
(35-50 µm)

13.4%

13.7%

16.3%

11.1%

Total silt
(2-50 µm)

52.2%

52.6%

49.6%

38.6%

Fine, very
fine sand
(50-75 µm)

12.8%

13.5%

18.2%

12.8%

Medium sand
(250-500 µm)

9.9%

8.2%

5.5%

13.4%

Coarse sand
(500-1000
µm)

5.8%

4.1%

3.1%

6.2%

113

Figure 5.32: An example of four continuous textural curves from the Amasa section. Note
that the textural curves are very similar and all have a mode within the coarse silt fraction
(35-50 um).

114

Greater loess thicknesses in the Amasa section, coupled with finer textures, suggest that
the loess deposits here are a distal part of the thicker, Iron County Loess. The Iron County Loess
is located primarily in Iron County, Michigan, and extends a few kilometers into Baraga,
Gogebic, and Marquette Counties, and into bordering counties in Wisconsin. Bigsby (2010) and
Scull and Schaetzl (2011) showed that the loess deposits in Iron County are thickest in southern
and central Iron County (between ~ 50 and 85 cm thick). Early evidence from Bigsby (2010)
suggested that the Iron County Loess was receiving eolian sediment from multiple source areas;
each source area was likely contributing different sizes of eolian sediment. The loess deposits are
presumably thickest in southern and central Iron County because the landscape was a glacial reentrant (Bigsby 2010). Here, the landscape is bordered by several ice-marginal positions formed
by four separate glacial lobes (Attig et al. 1985; Peterson 1985, 1986).
The uplands in southern and central Iron County would have likely been ice-free and
stable earlier than the surrounding areas, in addition to being at higher elevations, allowing more
time for loess to accumulate there. Although Bigsby’s (2010) work is on-going, the Vilas County
Outwash Plains, and the morainic features bordering Iron County, are thought to have been
sources for the ICLS. The Vilas County Outwash Plains are the southern-most (and largest)
source area identified by Bigsby (2010) for the Iron County Loess. Strong, chaotic winds likely
deflated the fine-grained sediments from these outwash plains and deposited them on stable
uplands, e.g., drumlins in Iron County. Once the Green Bay, Michigamme, Keweenaw, and
Ontonagon lobes retreated north from the Sagola, Republic, and Watersmeet ice-marginal
positions, these morainic landscapes could have contributed additional fine-grained sediments to
the Iron County Loess (Figures 5.33 and 5.34). Because southern and central Iron County likely
became subaerial first, loess deposits are generally thicker here. Figures 5.34 and 5.35 illustrate

115

the loess thickness trend and spatial distribution of silt contents within the Amasa section, in
relation to the entire Peshekee Loess, the Vilas County Outwash Plains, and the ice-marginal
positions; these maps are overlain onto the NRCS soil surface texture map.
My research provides additional data and supports the prior work of Bigsby (2010) on the
Iron County Loess. My work has joined the Peshekee Loess and the Iron County Loess, in the
Amasa section, which is likely the distal end of the thicker, and more extensive, Iron County
Loess. This research suggests that the loess in the Amasa section came from a distant source
because of its fine modal particle-size. The Vilas Outwash Plains were likely a major source area
for the Amasa section because (1) the outwash sediments were exposed earlier, as compared to
the surrounding areas, (2) the finer modal particle-size of the Amasa loess suggests a distant
source, and (3) the interpolated loess thickness. Bisgby (2010) as well as and Scull and Schaetzl
(2011) showed that loess progressively thickens towards southern Iron County (Figures 5.34 and
5.35).

116

Figure 5.33: Ice-margin positions in the western Upper Peninsula of Michigan and loess
sections within the study area. The ice-margins were determined using the work of Attig et
al. (1985), Peterson (1985, 1986) and various NRCS county-level soil surveys.

117

Figure 5.34: Map of interpolated loess thickness, overlain on the NRCS soil surface
texture map. The four smaller loess sections are shown in relation to the potential source
areas.
118

Figure 5.35: Map of interpolated distribution of clay-free silt contents, overlain onto
the NRCS soil surface texture map. The four smaller loess sections are shown in
relation to the potential source areas.
119

5.4.2 Covington Section
North of the Amasa section is the Covington section, located mainly in southern and
eastern Baraga County. Loess deposits in the Covington section generally are silty, similar to
those in the Amasa section (Tables 5.7 and 5.8). However, loess in the Covington section is
generally thinner than in the Amasa section (Table 5.7). The mean loess thickness for the
Covington section is 36.9 ± 13.9 cm and the mean total content within the 25-75 µm fraction is
39.6 percent; in contrast, the Amasa section has a mean loess thickness of 52.4 ± 13.1 cm and a
mean total content within the 25-75 µm fraction of 38.5 percent (Table 5.7). Texturally, the
Covington loess deposits usually have a bimodal textural curve or a shoulder curve, with a modal
particle-size within the 35-50 µm fraction and a small, secondary peak in the 250-500 µm
fraction (Figure 5.36). The first and second most common ETCs within the Covington section
are type 2 and type 4; ~ 43 percent of the samples within this section have a type 2 ETC and ~ 35
percent of the samples exhibit a type 1 ETC (Table 5.7). The mean modal particle-size within the
26-99 µm fraction for the Covington section is 40.2 µm, which is similar to the Amasa section
where the mean modal particle-size within the 26-99 µm fraction is 38.6 µm (Table 5.7).
Conversely, the Republic and Champion sections have a mean modal particle-size within the 2699 µm fraction of ~ 50 µm and 53 µm, respectively, which is much coarser than loess in the
Covington and Amasa sections (Table 5.7).

120

Figure 5.36: An example of four continuous textural curves from the Covington section.
Note that the textural curves are very similar and all have a mode within the coarse silt
fraction (35-50 µm).

121

Figures 5.17, 5.18, 5.19, 5.20, and 5.21 show that the Covington section may have a
western source area because this region generally has high percentages of contents within the 2575 µm fraction, compared to the surrounding areas. The Ontonagon Clay Plains is an extensive
physiographic region (~200,000 hectares) ~ 12 km west of the Covington section (Figures 5.34
and 5.35; Peterson 1985; Schaetzl et al. (in prep.)) that is largely comprised of fine-textured tills
and lacustrine sediments (Figure 5.37). Soils there have silt loam, silty clay loam, and clay loam
surface textures, e.g., the Negwegon, Amnicon, and Cuttre series (Eversoll and Carey 2010).
Generally the Ontonagon Clay Plains are a low to moderate relief landscape, but relief is much
higher where the Ontonagon River and its tributaries have incised the landscape and created
deep, V-shaped valleys (Figure 5.37). The dense network of steep and unstable slopes on the
Ontonagon Clay Plains has exposed a plethora of fine-grained sediments. The Covington
section’s fine modal particle-size and the high percentage of contents within the 25-75 µm
fraction, directly east of the Ontonagon Clay Plains, suggest that ~ 11,500 cal. yr BP, after the
Marquette Phase (Hughes and Merry 1978; Lowell et al. 1999; Pregitzer et al. 2000), when
vegetation was scarce, westerly winds likely deflated the fine-grained sediments from the
Ontonagon Clay Plains and later deposited them on distant, stable-uplands, within the Covington
section (Figure 5.35).
Additionally, a smaller and more local source area was likely the Baraga Plains, located
east of the Ontonagon Clay Plains and adjacent to the Covington section (Figures 5.38 and 5.39).
The Baraga Plains are a low relief, sandy, dry, outwash or glaciolacustrine plain. Similar to the
Ontonagon Clay Plains, the Baraga Plains are likely associated with the Marquette Phase (Barrett
et al. 1995; Arbogast and Packman 2004; Figures 5.16, 5.34 and 5.39;).The anomalously thick
loess deposits of the Covington section, located directly east of the Baraga Plains, suggest that

122

this small outwash plain (~ 7,000 hectares) could have also contributed fine-grained sediments to
the loess deposits of the Covington section (Figures 5.38 and 5.39). Moreover, Arbogast and
Packman (2004) documented active sand dunes ~ 7,800 cal. yr BP on the Baraga Plains. During
this dune activation period, saltating eolian sand may have facilitated down-wind transportation
of the finer-grained sediments, similar to the model Mason et al. (1999) described.

Figure 5.37: The unstable, deeply incised Ontonagon Clay Plains, showing V-shaped
valleys and silt loam and clay loam upper soil textures.

123

This research suggests that the loess in the Covington section came from both distant and
local sources. The Ontonagon Clay Plains were likely the dominant distant source area for the
Covington section because of (1) the east-west, tongue-shaped distribution of silt contents within
this area, and (2) the overall fine modal particle-size of this loess (Figure 5.35). Although it
likely contributed less, the Baraga Plains could have also contributed fine-grained sediments to
the Covington section due to the anomalously thicker loess deposits located directly southeast of
the Baraga Plains (Figure 5.38).

Figure 5.38: Map showing loess thicknesses and the anomalously thick loess deposits that
are located within the Covington section, just southeast of the Baraga Plains.

124

Figure 5.39: The Baraga Plains, a low-relief outwash plain with scattered sand dunes.
Photograph by R. Schaetzl.

125

5.4.3 Republic Section
The Republic section is located near the eastern and southeastern margins of the study
area, in western Marquette County. Of the four loess sections in the study area, loess in the
Republic section has the second highest mean thickness (42.5 ± 11.8 cm) and 25-75 µm fraction
contents - 47.2 percent (Tables 5.7 and 5.8). The first and second most common ETC types
within the Republic section are the type 2 and type 1 curves; ~ 53 percent of the samples
collected in this section have a type 2 ETC and ~ 37 percent have a type 1 ETC (Table 5.7;
Figure 5.40). Typical continuous textural curves from the Republic section show that loess
deposits here usually have a high abundance of coarse silt (50 µm) and fine, very fine sand (5075 µm), often with a small, secondary peak in medium sand (250-500 µm) fraction (Figure 5.40).
The main attribute that distinguishes the Republic section from the Amasa section is the modal
particle-size of the loess samples. The Republic section has a mean modal particle-size of 49.8
µm, whereas the loess in the Amasa section is much finer with a mean modal particle-size of
38.6 µm (Table 5.7). I suggest that the coarser-textured loess in the Republic section is due to
more localized source area(s) composed of coarser sediments.

126

Figure 5.40: Examples of continuous textural curves from the Republic section. Note that
the textural curves are very similar and all have a mode within the fine, very fine sand
fraction (50-75 um).

127

The Michigamme River, Republic Moraine, Sagola Moraine, and the Gwinn Sandy
Terrain are all near the central and eastern margins of the Republic section, and were likely local
source areas for the loess in the Republic section (Figures 5.33, 5.34 and 5.40). The Michigamme
River was likely carrying silt-rich, glacial meltwater while the Michigamme and Green Bay
lobes stalled at the Republic and Sagola ice-margin positions. During low-flow events when
discharge was minimal, the river’s floodplain and terrace sediments would have been exposed.
Strong, chaotic winds presumably could have deflated the fine-grained floodplain and terrace
sediments and deposited them on adjacent stable-uplands within the Republic section.
Peterson (1986) describes the Republic Moraine as being composed of gravelly and
sandy, brown till. He also suggested that the Sagola Moraine was built by the Green Bay Lobe
and consists of calcareous red drift composed of ice-contact stratified sand and gravel (Figure
5.33). The red color of the till suggests that the ultimate source of the drift was the red
Precambrian sandstones and shale from the Lake Superior basin (Attig et al. 1985; Peterson
1986). During the retreat of the Michigamme and Green Bay lobes from the Republic and Sagola
ice-margin positions, strong, katabatic winds may have winnowed-out the fine-grained sediments
from the coarse-textured glacial till (Figure 5.31; Bigsby 2010). These fine-grained sediments
were then likely deposited within the Republic section, on near-by, stable-uplands, south of the
Republic and Sagola moraines.
The Gwinn Sandy Terrain is located in central Marquette County, east of the Republic
section, and may also have been a source area for the loess in the Republic section (Figures 5.34
and 5.35). The Gwinn Sandy Terrain is dominated by a sandy head-of-outwash, distal to the
Watton ice-margin position (Figures 3.9 and 5.33). The topography of this region includes
hummocky areas, incised stream valleys, small bedrock outcrops and bedrock-influenced terrain,

128

and low, rolling terrain on sandy thick drift (Table 3.3; Schaetzl et al. (in prep.)). Again, strong,
katabatic winds may have winnowed-out the fine-grained sediments from the coarser outwash
and deposited them on adjacent, stable-uplands, as they did from the Vilas County Outwash
Plains.
In sum, loess in the Republic section was probably generated from local source areas.
Evidence for this conclusion is the coarser nature of the loess deposits (Figure 5.31).
Nonetheless, this section likely had multiple source areas due to the thick loess deposits that exist
there (Table 5.7; Figure 5.16).
5.4.4 Champion Section
The Champion section is the northern-most loess section within the study area. The
Champion section is within the Peshekee Highlands physiographic region, where the relief is
generally very high, in some places more than ~ 205 m per 250 m radius area, with an average of
~ 30 m. Conversely, the Michigamme Bedrock Terrain has an average local relief of ~ 13 m, and
is where the Amasa, Covington, and Republic sections are located (Figures 3.8 and 5.28). Loess
deposits in the Champion section are generally thinner and have a coarser modal particle-size
than the Amasa, Covington, and Republic sections (Tables 5.7 and 5.8). The mean loess
thickness for the Champion section is 32.9 ± 10.3 cm, whereas the Amasa, Covington, and
Republic sections have mean thicknesses of ~ 52, 37, and 43 cm, respectively (Table 5.7). The
Champion section’s mean modal particle-size within the 26-99 µm fraction is 52.9 µm,
compared to the Amasa, Covington, and Republic sections that have mean modal particle-size in
the 26-99 µm fraction of 38.6, 40.2, and 49.8 µm, respectively (Table 5.7). Moreover, the loess
deposits within the Champion section are generally much sandier than are the loess deposits of
the Amasa, Covington, and Republic sections (Tables 5.7 and 5.8). The Champion section’s
mean content within the 250-500 µm fraction is 13.4 percent, whereas the mean 250-500 µm
129

fraction content within the Amasa, Covington, and Republic sections are 9.9, 8.2, and 5.5
percent, respectively (Table 5.7). Type 2 and 4 ETCs are the most common ETCs for the
Champion section, which has the lowest percentage of both type 1 and 2 curves, and the highest
percentage of the type 5 and 3 ETCs (Table 5.7). Figure 5.41 shows that the continuous textural
curves of loess deposits within this section generally vary considerably across even short
distances, and that the modal particle-size of these loess deposits fluctuates around 50 µm. This
is likely a product of these Champion loess deposits having several, small, local source areas.
Thus, the modal particle-size will vary, in addition to the thicknesses, depending on distance and
direction from the source areas.

130

Figure 5.41: Examples of continuous textural curves from the Champion section. Note that
the textural curves of loess deposits within this section generally vary considerably across
short distances, and that the modal particle-size of these loess deposits is generally within
the fine, very fine sand fraction (50-75 µm).
.

131

Loess within the 50-75 µm fraction, likely did not travel far, and thus, like the Republic
section, the Champion section’s loess deposits were presumably from local source areas. Small,
interspersed glacial meltwater streams and outwash plains were likely the dominant sources of
loess for the Champion loess deposits. The Peshekee River flows through the western part of the
Champion section, and Figure 42 shows that there are several anomalously thick loess deposits
located near the Peshekee River, which has been associated with the Marquette Advance
(Hughes and Merry 1978; Farrand and Drexler 1985; Barrett et al. 1995; Lowell et al. 1999).
When the Michigamme Lobe retreated north from the Marquette Phase, ice-marginal position,
the stream presumably carried glacial meltwater, and was choked with both coarse- and finegrained sediments. During low-flow events, the finer sediments would have been exposed on the
river’s floodplain and terraces (Figures 5.43 and 5.44). Eventually strong winds could have
deflated the fine-grained sediments and deposited them on adjacent uplands within the Champion
section (Figures 5.42 and 5.44; Schaetzl and Liebens 1992, 1993).

Figure 5.42: Map showing the anomalously thick loess deposits that are
located near the Peshekee River, within the Champion section.
132

Figure 5.43: Peshekee River terrace and terrace sediments. Photograph by R.
Schaetzl.

Figure 5.44: The Peshekee River - an underfit stream. Photograph by R. Schaetzl.

133

Moreover, small interspersed outwash plains were also likely source areas for the Champion
section. Although the kriged maps of the distributions of loess thicknesses and silt contents do
not provide evidence of this assertion, field observation and NRCS parent material maps show
that there are numerous low-relief, sandy, outwash plains interspersed throughout the Peshekee
Highlands and Champion section - generally in the areas between bedrock knobs (Figure 3.9).
The coarse nature of the loess deposits within Champion section suggests that the loess did not
travel far and was likely in suspension for a short period of time. The coarser and sandier
outwash plains would have presumably been the first to dry-out and thus allow for strong winds
to deflate the fine-grained sediments and later deposit them on near-by stable uplands. This
region’s high-relief and rugged topography did not allow for saltation to occur very far and
hence, the very fine sands are likely from local source areas. Additionally, there is strong
evidence of established vegetation prior to the Marquette advance - the Lake Gribben forest bed,
near palmer in Marquette County (Lowell et al. 1999). Thus, the uplands within the Peshekee
Highlands could have had small amounts of vegetation which may have acted as a trap/sink for
fine-grained sediments that were being transported by wind.
5.5 Summary
The study area is an assemblage of several smaller, overlapping loess regions that likely
had several heterogeneous source areas. Thus, loess textures and thicknesses usually vary
between sections. The sections with thicker loess deposits are generally within the Michigamme
Bedrock Terrain, and usually have a low percentage of sand. Conversely, the Champion section,
which is within the Peshekee Highlands, generally has thinner loess deposits and these deposits
are sandier.

134

6. Conclusions
Wind-blown silt has been recognized over a great deal of the central United States. In the
core of the region, ranging from Illinois to Nebraska, loess deposits may exceed several meters
in thickness and usually have one dominant source area, e.g., the Mississippi or Missouri River
floodplain. At the distal edges of these large loess regions, the deposits are, generally, much
thinner and are very discontinuous. Many argue that, in such areas, the loess may have been
sourced from different areas and landforms. These thin loess deposits are also patchy and vary in
texture from site to site.
Recent research demonstrates that loess deposits also occur in the upper Midwest. Prior
work by Schaetzl and Hook (2008), Schaetzl and Loope (2008), Schaetzl (2008), Hobbs et al.
(2011), Scull and Schaetzl (2011), and Stanley and Schaetzl (2011) provides evidence that loess
extends into both Michigan and Wisconsin. Within Michigan, the Peshekee Loess is at the
northeastern margin of the North American, midcontinent loess region. This research is the first
to document both the extent and textural characteristics of loess situated at the extreme margins
of a much larger and thicker, region of loess. Peshekee Loess is patchy, with marked variations
in both landscape topography and loess characteristics. Because the Peshekee Loess is
intermittent and highly variable in texture, a data set clean-up process was initiated for this study.
Five eolian textural curve types were found to dominant the study area, and each sample was
assigned to one of these specific textural categories. This categorization of eolian textural curve
types is unique and has never before been done by other loess studies.
The Peshekee Loess deposits often have a bimodal continuous textural curve, usually
with a primary modal particle-size within the 25-75 µm fraction and a secondary mode within
the 250-500 µm fraction. “Shoulder” textural curves are also common with Peshekee Loess, and
they are considered to be a hybrid between a true unimodal continuous textural curve and a
135

bimodal textural curve. Generally, within the study area, where the loess is thick and the relief is
moderate, e.g., the Michigamme Bedrock Terrain, or near the southern margins of the study area,
loess deposits contain more silt, e.g., ETC types 1, 2, and 4. Conversely, loess deposits usually
have a higher percentage of sand, e.g., ETC types 3 and 5, where they are thin and where relief is
high, such as in the Peshekee Highlands. Findings from this research suggest that thinner loess
deposits often have had more of the coarser, underlying sediment mixed into the loess column
than have the thicker loess deposits. Further research is needed to confirm this hypothesis and
determine the main vectors for this mixing.
Overall, within the study area, loess deposits are the thickest near the southern margins
and progressively become thinner towards the northeast. Moreover, loess deposits with the finest
modal particle-sizes are usually near the southern margins of the study area; coarser textured
loess deposits are towards the northeast. Field observations and maps showing the distributions
of eolian particle size fractions suggest that the study area is actually an amalgamation of several,
smaller loess regions.
The Peshekee Loess likely had both distant and local source areas, and several different
kinds of landforms and landscapes contributed eolian sediment. Because of the heterogeneous
nature of these source areas and the various amounts of in-mixed sand, the study area was
divided into four sections. From south to north they are the Amasa, Covington, Republic, and
Champion sections. The loess deposits within the Amasa section are thick and silty, and are
likely the distal part of the Iron County Loess, where loess deposits are generally thick and silty,
and extend a larger area. The dominant source area for the Amasa section was likely from the
south, e.g., the Vilas County Outwash Plains of Wisconsin. The loess deposits within the
Covington section are generally thin, with high percentages of silt and low sand content. The

136

dominant source area for these loess deposits was likely from the west, e.g., the Baraga Plains
and the Ontonagon Clay Plains. The loess deposits within the Republic section are generally
thick, with low percentages of sand. They have abundant coarse silt and fine, very fine sand. The
Republic loess deposits generally have a coarser modal particle-size, compared to the Amasa and
Covington sections. The dominant source areas for the Republic section were likely from both
the north and east, e.g., the Republic and Sagola moraines, the Michigamme River floodplain,
and the Gwinn Sandy Terrain. The Champion loess deposits, located within the Peshekee
Highlands, are generally thin, highly discontinuous, and sandy. Local source areas were likely
the dominant sources for this loess, e.g., small, interspersed glacial meltwater streams and
outwash plains. The Peshekee River floodplain was also likely a main source area for the loess of
the Champion section. This river would have carried glacial meltwater, choked with silt, and
thus, could have contributed fine-grained sediments ~ 11,500 cal. yr BP, while the ice margin sat
at the Marquette Moraine.
This research has two main conclusions: (1) it recognizes that thin loess deposits, on a
recently glaciated landscape, may have several source areas, and (2) it documents the effects of
mixing within thin loess deposits. This research also provides valuable insight into how to map
and categorize thin loess deposits, near ice-margin positions. These thin loess deposits may
provide important information on paleoenvironmental conditions during both glacial retreat and
loess deposition.

137

REFERENCES

138

REFERENCES

Anderson, S. P. 1988. The upfreezing process: experiments with a single clast. Geological
Society of America Bulletin 100 (610): 609-621.
Arriaga, F. J., and B. Lowery. 2006. A fast method for determining soil particle size distribution
using a laser instrument. Soil Science 171 (9): 663-74.
Arbogast, A. F., and S. C. Packman. 2004. Middle-Holocene mobilization of aeolian sand in
western upper Michigan and the potential relationship with climate and fire. The
Holocene 14 (3): 464-471.
Attig, J. W., Jr. 1984. The Pleistocene geology of Vilas County, Wisconsin (Ph.D. thesis):
Madison, Wisconsin, University of Wisconsin.
Attig, J. W., L. Clayton, and D. M. Mickelson. 1985. Correlation of late Wisconsin glacial
phases on the western Great Lakes. Geological Society of America Bulletin 96 (12):
1585-1593.
Barrett, L. R., J. Liebens, D. G. Brown, R. J. Schaetzl, P. Zuwerink, T. W. Cate, and D. S. Nolan.
1995. Relationships between soils and presettlement vegetation in Baraga County,
Michigan. American Midland Naturalist. 134: 264-285.
Bettis, E. A., D. R. Muhs, H. M. Roberts, and A. G. Wintle. 2003. Last Glacial loess in the
conterminous USA. Quaternary Science Reviews 22 (18-19): 1907-1946.
Beuselinck, L. G. Govers, J. Poesen, G. Degraer, and L. Froyen. 1998. Grain-size analysis by
laser diffractometry: comparison with the sieve-pipette method. Catena 32 (3-4): 193208.
Berndt, L. R. 1988. Soil survey of Baraga County Area, Michigan. USDA, Soil Conservation
Service. Washington, D.C.: U.S. Government Printing Office.
Bigsby, M. E. 2010. The characterization and possible origins of two loess sheets in the upper
great lakes region, USA, MS Thesis. Michigan State University, East Lansing.
Bornhorst, T. J., and D. Brandt. 2009. In: Schaetzl, R. J., J. T. Darden, and D. Brandt. (eds.).
Michigan Geography and Geology. Boston, MA: Pearson Custom Publishing.
Buurman, P. T., and C. C. Muggler. 1997. Laser grain-size determination in soil genetic studies.
1. Practical problems. Soil Science 162 (1): 211-218.
Card, K. D. 1990. A review of the Superior Province of the Canadian Shield, a product of
Archean accretion. Precambrian Research 48: 99-156.

139

Clayton, L. 1983. Chronology of Lake Agassiz drainage to Lake Superior, in: Teller, J.T.,
Clayton, L. (Eds.). Glacial Lake Agassiz. Geological Association of Canada Special
Paper 26: 309-332.
Clayton, L. 1984. Pleistocene geology of the Superior Region, Wisconsin: Wisconsin Geological
and Natural History Survey Information Circular.
Comer, P. J., D. A. Albert, H. A. Wells, B. L. Hart, J. B. Raab, D. L. Price, D. M. Kashian, R. A.
Corner, and D. W. Schuen. 1995. Michigan’s native landscape, as interpreted from the
General Land Office surveys 1816-1856. Rept. to the US EPA Water Division and the
Wildlife Division, MI Dept. of Natural Resources, Michigan Natural Features Inventory,
Lansing. 78 pp. and digital map. http://www.mcgi.state.mi.us/mgdl.
Corte, A. E. 1963. Particle sorting by repeated freezing and thawing. Science 142: 499-501.
Davis, D. W., and J. C. Green. 1997. Geochronology of the North American Midcontinent rift in
western Lake Superior and implications for its geodynamic evolution. Canadian Journal
of Earth Science 34: 476-488.
Derouin, S. A., T. V. Lowell, and I. Hajdas. 2007. Landscape evolution and deglaciation of the
Upper Peninsula, Michigan: An examination of chronology and stratigraphy in kettle lake
cores. J. Great Lakes Research. 33: 875-886.
Doonan, C. J., and J. R. Byerlay. 1973. Water Investigation 11, Ground water and geology of
Baraga County, Michigan. A report prepared in cooperation with the Water Resources
Division of the Geological Survey, United States Department of the Interior. Lansing,
Michigan.
Eichenlaub, V. L., J. R. Harman, F. V. Nurnberger, and H. J. Stolle. 1990. The Climatic Atlas of
Michigan. Univeristy Notre Dame Press. Notre Dame, IN.
Environmental Systems Research Institute (ESRI). 2011. ArcGIS Desktop: Release 10.
Redlands, CA.
Eversoll, S. J., L. M. Carey. 2010. Soil survey of Ontonagon County area, Michigan. USDA,
Natural Resources Conservation Service. Washington, D.C.: Government Printing Office.
Farrand, W. R. and C. W. Drexler. 1985. Late Wisconsinan and Holocene history of the Lake
Superior basin. In: Quaternary Evolution of the Great Lakes. Karrow, P. F., and P. E.
Calkin (eds.). Geological Association of Canada Special Paper 30: 17-32.
Fenneman, N. M. 1938. Physiography of Eastern United States. New York: McGraw-Hill.
Follmer, L. R. 1996. Loess studies in central United States: Evolution of concepts. Engineering
Geology 45 (1-4):287-304.

140

Forman, S. L., E. A. Bettis, T. J. Kemmis, and B. B. Miller. 1992. Chronologic evidence for
multiple periods of loess deposition during the late Pleistocene in the Missouri and
Mississippi River Valleys, United States: Implications for the activity of the Laurentide
Ice sheet. Palaeogeography, Palaeoclimatology, Palaeoecology 93 (1): 71-83.
Forman, S. L., and J. Pierson. 2002. Late Pleistocene luminescence chronology of loess
deposition in the Missouri and Mississippi river valleys, United States. Palaeogeography,
Palaeoclimatology, Palaeoecology 186 (1): 25-46.
Frazee, C. J., J. B. Fehrenbacher, and W. C. Krumbein. 1970. Loess Distribution from a Source.
Soil Science Society of America Journal 34 (1): 296-301.
Grimley, D. A. 2000. Glacial and nonglacial sediment contributions to Wisconsin Episodes loess
in the central United States. Geological Society of America Bulletin 112 (1-5): 14751495.
Hobbs, T., R. J. Schaetzl, and M. D. Luehmann. 2011. Evidence for periodic, Holocene loess
deposition in kettles in a sandy, interlobate landscape, Michigan, USA. Aeolian Research
3: in press.
Hughes, J. D., and W. J. Merry. 1978. Marquette buried forest 9,850 years old: American
Association for the Advancement of Science, National Meeting.
Inglis, D. R. 1965. Particle sorting and stone migration by freezing and thawing. Science 148:
1616-1617.
Johnson, W. H., and L. R. Follmer. 1989. Source and origin of Roxana Silt and Middle
Wisconsinan Midcontinent glacial activity. Quaternary Research 31 (1): 319-331.
Kilmer, V. J., and L.T. Alexander. 1949. Methods of making mechanical analysis of soils. Soil
Science 68:15-24.
Konert, M., and J. E. F. Vandenberghe. 1997. Comparison of laser grain size analysis with
pipette and sieve analysis: a solution for the underestimation of the clay fraction.
Sedimentology 44 (3): 523-535.
Larson, G. J., and R. J. Schaetzl. 2001. Origin and evolution of the Great Lakes. Journal of Great
Lakes Research 27:518-546.
Leigh, D. S., and J. C. Knox. 1993. AMS Radiocarbon Age of the Upper Mississippi Valley
Roxana Silt. Quaternary Research 39 (1): 282-289.
Leigh, D. L. 1994. Roxana silt of the Upper Mississippi Valley: Lithology, source, and
paleoenvironment. Geological Society of America Bulletin 106 (1): 430-442.

141

Leigh, D. S., and J. C. Knox. 1994. Loess of the Upper Mississippi Valley Driftless Area.
Quaternary Research 42 (1): 30-40.
Leverett, F. 1911. Surface geology of the northern peninsula of Michigan with notes on
agricultural conditions and water power. Michigan Geological and Biological Survey
Publication 7. Geological Series 5.
Linsemier, L. H. 1989. Soil survey of Dickinson County Area, Michigan. USDA, Soil
Conservation Service. Washington, D.C.: U.S. Government Printing Office.
Linsemier, L. H. 1997. Soil survey of Iron County Area, Michigan. USDA, Natural Resources
Conservation Service. Washington, D.C.: U.S. Government Printing Office.
Loizeau, J. L., D. Arbouille, S. Santiago, and J. P. Vernet. 1994. Evaluation of a wide range laser
diffraction grain size analyzer for use with sediments. Sedimentology 41 (1): 353-361.
Lowell, T. V., G. J. Larson, J. D. Hughes, and G. H. Denton. 1999. Age verification of the Lake
Gribben forest bed and the Younger Dryas Advance of the Laurentide Ice Sheet.
Canadian Journal of Earth Science 36:383-393.
Maat, P. B., and W. C. Johnson. 1996. Thermoluminescence and new 14C age estimates for late
quaternary loess in southwestern Nebraska. Geomorphology 17 (1): 115-128.
McKay, E. D. 1979. Wisconsinan loess stratigraphy of Illinois. In: Wisconsinan, Sangamonian,
and Illinoian stratigraphy in central Illinois. Midwest Friends of the Pleistocene Field
Conference Guidebook. Illinois State Geological Survey Guidebook 13. pp. 95-108.
Markewich, H. W., D. A. Wysocki, M. J. Pavich, E. M. Rutledge, H. T. Millard, F. J. Rich, P. B.
Maat, M. Rubin, and J. P. McGeehin. 1998. Paleopedology plus TL, Be-10, and C-14
dating as tools in stratigraphic and paleoclimatic investigations, Mississippi River Valley,
USA. Quaternary International 51-2 (1): 143-167.
Mason, J. A., E. A. Nater, C. W. Zanner, and J. C. Bell. 1999. A new model of topographic
effects on the distribution of loess. Geomorphology 28 (1): 223-236.
Mason, J. A. 2001. Transport Direction of Peoria Loess in Nebraska and Implications for Loess
Sources on the Central Great Plains. Quaternary Research 56 (1):79-86.
Mason, J. A., P. M. Jacobs, P. R. Hanson, X. Miao, and R. J. Goble. 2003. Sources and
paleoclimatic significance of Holocene Bignell Loess, central Great Plains, USA.
Quaternary Research 60 (1):330-339.
Mason, J. A., R. M. Joeckel, and E. A. Bettis. 2007. Middle to Late Pleistocene loess record in
eastern Nebraska, USA, and implications for the unique nature of Oxygen Isotope Stage
2. Quaternary Science Reviews 26 (1-6): 773-792.

142

Muhs, D. R., and E. A. Bettis. 2000. Geochemical variations in Peoria Loess of western Iowa
indicate paleowinds of midcontinental North America during last glaciations. Quaternary
Research 53 (1):49-61.
NOAA National Climatic Data Center. 2002. Michigan Monthly Normals of Temperature,
Precipitation, and Heating and Cooling Degree Days, 1971-2000.
Peterson, W. L. 1982. Preliminary surficial geologic map of the Iron River 1o x 2o quadrangle,
Michigan and Wisconsin. U.S. Geological Survey Open File Report pp. 82-301.
Peterson, W. L. 1985. Surficial geologic map of the Iron River 1o x 2o quadrangle, Michigan and
Wisconsin: U.S. Geological Survey Miscellaneous Investigations Series Map I-1360-C,
Scale 1:250,000.
Peterson, W. L. 1986. Late Wisconsinan glacial history of northeastern Wisconsin and western
upper Michigan. U.S. Geological Survey Bulletin 1652. 14 pp.
Pye, K. 1995. The nature, origin and accumulation of loess. Quaternary Science Reviews 14 (25): 653-667.
Pregitzer, K. S., D. D. Reed, T. J. Bornhorst, D. R. Foster, G. D. Mroz, J. S. Mclachlan, P. E.
Laks, D. D. Stokke, P. E. Martin, and S. E. Brown. 2000. A buried spruce forest provides
evidence at the stand and landscape scale for the effects of environment on vegetation at
the Pleistocene/Holocene boundary. Journal of Ecology 88: 45-53.
Reed, R. C., and J. Daniels. 1987. Bedrock geology of northern Michigan. State of Michigan
Department of Natural Resources. Map: 1: 500,000.
Roberts, H. M., D. R. Muhs, A. G. Wintle, G. A. T. Duller, and E. A. Bettis. 2003.
Unprecedented last-glacial mass accumulation rates determined by luminescence dating
of loess from western Nebraska. Quaternary Research 59 (4): 411-419.
Rodbell, D. T., S. L. Forman, J. Pierson, and W. C. Lynn. 1997. Loess and paleosol stratigraphy,
magnetic susceptibility and chronology of Mississippi Valley loess in western Tennessee.
Geological Society of America Bulletin 109 (9): 1134-1148.
Ruhe, R. V. 1954. Relations of the Properties of Wisconsin Loess to Topography in Western
Iowa. American Journal of Science 252 (1): 663-672.
Ruhe, R. V. 1956. Geomorphic surfaces and the nature of soils. Soil Science 82 (1): 441-455.
Ruhe, R. V. 1969. Quaternary Landscapes in Iowa. Ames, IA: Iowa State University Press,
255pp.
Ruhe, R. V., and C. G. Olson. 1980. Clay-Mineral Indicators of Glacial and Non-Glacial Sources
of Wisconsinan Loesses in Southern Indiana, USA. Geoderma 24 (1): 283-297.
143

Ruhe, R. V. 1983. Depositional environment of late Wisconsin loess in the midcontinental
United States. Late-Quaternary Environments of the United States 1: 130-137.
Rutledge, E. M., L. T. West, and M. Omakupt. 1985. Loess deposits on a Pleistocene age terrace
eastern Arkansas. Soil Science Society of America Journal 49 (5): 1231-1238.
Rutledge, E. M., M. J. Guccione, H. W. Markewich, D. A. Wysocki, and L. B. Ward. 1996.
Loess stratigraphy of the Lower Mississippi Valley. Engineering Geology 45 (3): 167183.
Schaetzl, R. J. 2008. The distribution of silty soils in the Grayling Fingers region of Michigan:
Evidence for loess deposition onto frozen ground. Geomorphology 102 (1-2): 287-296.
Schaetzl, R. J. and J. Liebens. 1992. Spatial Analysis of Eolian Sediments in the Peshekee
Highlands, Marquette and Baraga Counties, Michigan. Presented at the annual meeting
of the Michigan Academy of Science, Arts and Letters, Mount Pleasant, MI.
Schaetzl, R. J. and J. Liebens. 1993. A Possible Eolian Deposit in Michigan's Upper Peninsula".
Presented at the 3rd International Association of Geomorphologists Conference,
Hamilton, Ontario, Canada.
Schaetzl, R. J. and S. A. Isard. 1996. Regional-Scale Relationships Between Climate and
Strength of Podzolization in the Great Lakes Region, North America. Catena 28: 47-69.
Schaetzl, R. J., and S. Anderson. 2005. Soils: genesis and geomorphology. New York:
Cambridge University Press.
Schaetzl, R. J., and J. Hook. 2008. Evidence for Loess in Northwest Lower Michigan:
Characterizing the Silty Mantle on the Buckley Flats Outwash Plain. Physical Geography
29 (2): 140-157.
Schaetzl, R. J., and W. L. Loope. 2008. Evidence for an eolian origin for the silt-enriched soil
mantles on the glaciated uplands of eastern Upper Michigan, USA. Geomorphology 100
(1): 285-295.
Schaetzl, R. J., J. T. Darden, and D. Brandt. (eds.). 2009. Michigan Geography and Geology.
Boston, MA: Pearson Custom Publishing.
Schwenner, C. 2007. Soil survey of Marquette County Area, Michigan. USDA, Natural
Resources Conservation Service. Washington, D.C.: U.S. Government Printing Office.
Scull, P., and R. J. Schaetzl. 2011. Using PCA to characterize and differentiate loess deposits in
Wisconsin and Upper Michigan, USA. Geomorphology 127 (1-14): 143-155.

144

Sims, P. K. 1992. Geologic map of Precambrian rocks, southern Lake Superior region,
Wisconsin and northern Michigan: U.S. Geological Survey Miscellaneous Investigations
Map I-2185, scale= 1:500,000.
Sims, P. K. 1993. Great Lake Textonic Zone in the Marquette area, Michigan-Implication for
Archean tectonics in north-central United States. In: Sims, P.K. (ed), Contributions to
Precambrian Geology of the Great Lakes Region. US Geological Survey Professional
Paper 1904E.
Smalley, I. J. 1975. Loess: Lithology and Genesis. Stroudsburg, PA: Dowden Hutchinson and
Ross.
Smalley, I. J. 1990. Possible Formation Mechanisms for the Modal Coarse-Silt Quartz Particles
in Loess Deposits. Quaternary International 7/8 (1): 23-27.
Smalley, I. J., and V. Smalley.1983. Loess material and loess deposits: Formation, distribution
and consequences. Eolian sediments and processes 38: 51–68.
Smith,G. D. 1942. Illinois Loess: Variations in Its Properties and Distribution, a Pedologic
Interpretation: University of Illinois, Agricultural Experiment Station.
Soil Survey Division Staff. 1993. Soil Survey Manual. USDA Handbook No. 18. US
Govt. Printing Office, Washington, DC.
Soil Survey Staff, Natural Resources Conservation Service, United States Department of
Agriculture. Official Soil Series Descriptions. Available online at
http://soils.usda.gov/technical/classification/osd/index.html. Accessed Feb. 21, 2010.
Sperazza, M., J. N. Moore, and M. S. Hendrix. 2004. High-Resolution Particle Size Analysis of
Naturally Occurring Very Fine-Grained Sediment through Laser Diffractometry. Journal
of Sedimentary Research 74 (5): 736-743.
Stanley, K. E. 2008. The characterization and paleoenvironmental significance of the north
central Wisconsin loess sheet, MS Thesis. Michigan State University, East Lansing.
Stanley, K. E., and R. J. Schaetzl. 2011. Characteristics and paleoenvironmental significance of a
thin, dual-sourced loess sheet, north-central Wisconsin. Aeolian Research 2 (1): 241-251.
Stoelting, P.K. 1989. Delimiting and subdividing terrain in the interlobate kettle moraine of
eastern Wisconsin. Wisconsin Geology. 5: 41-59.
Tsoar, H., and K. Pye. 1987. Dust Transport and the Question of Desert Formation.
Sedimentology 34 (1): 139-153.
Tyler, A. N., S. Carter, D. A. Davidson, D. J. Long, and R. Tipping. 2001. The extent and
significance of bioturbation on 137Cs distributions in upland soils. Catena 43 (2): 81-99.

145

Vanlier, K. E. 1963. Reconnaissance of the ground-water resources of Alger Co, Michigan
Geological Survey Division, Department of Conservation. Water Investigation. 1: 55.
Vilborg, L. 1955. The uplift of stones by frost. Geog. Annaler 37: 164-169.
Wascher, H. L., R. P. Humbert, and J. G. Cady. 1947. Loess in the southern Mississippi valley:
Identification and distribution of loess sheets. Soil Science Society of America 12 (1):
389-399.
Wikgren, K. 2007. Physiography prepared by Ken Wikgren, soil scientist, Natural Resources
Conservation Service. In: Schwenner, C. 2007. Soil survey of Marquette County Area,
Michigan. USDA, Natural Resources Conservation Service. Washington D.C.: U.S.
Government Printing Office.
Wikgren, K. 2011. Personal Communication.
Willman, H. B., and J. C. Frye. 1970. Pleistocene Stratigraphy of Illinois: Illinois State
Geological Survey Bulletin.

146