University of Southern California Dissertations and Theses

Evaluating glacier movement fluctuations using remote sensing: a case study of the Baird, Patterson, LeConte, and Shakes Glaciers in central southeastern Alaska

(USC Thesis Other)

Evaluating glacier movement fluctuations using remote sensing: a case study of the Baird, Patterson, LeConte, and Shakes Glaciers in central southeastern Alaska

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content EVALUATING GLACIER MOVEMENT FLUCTUATIONS USING REMOTE SENSING: A
CASE STUDY OF THE BAIRD, PATTERSON, LECONTE, AND SHAKES GLACIERS IN
CENTRAL SOUTHEASTERN ALASKA

By
Robert Howard Davidson

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOGRAPHIC INFORMATION SCIENCE AND TECHNOLOGY)
March 2014
Copyright 2014 Robert Howard Davidson
ii

ACKNOWLEDGEMENTS
From conception to finalization of this thesis has been what seemed like a lifetime.
Excitement and despair seemed to change places on a weekly basis. There were many times
when I was elated to finally be working on this project; other times were spent wondering if I
would ever complete it. My chief supporter and encourager is my wife. To this day, I am not sure
how she and my two children put up with the long nights that I put in in writing, editing, and
finalizing this document. Besides my family, there are others that I should recognize; for without
their support and inspiration, you would not be reading this document.
I would like to thank my thesis chair person, Dr. Flora Paganelli, for her guidance and
assistance throughout this thesis. I would also like to thank Dr. Su Jin Lee and Dr. Lowell Stott
who also served on my thesis committee for their guidance. Many years ago, I had the privilege
of learning land surveying from Mr. Paul Bowen. Through his kind instruction and mentoring, I
developed a fascination with all things geospatial; especially surveying. It was in his class that I
first went to LeConte Glacier and conducted a ground-based survey of that glacier. Years later,
that experience would help me as a geospatial analyst/geodetic surveyor in the United States
Marine Corps. While serving in the Marine Corps, I developed a close friendship with Mr.
Michael Noderer who taught me most of what I know about remote sensing; especially
phenomenology.
Throughout my life, I have had the privilege of learning from some of the best people in
the “business”. It is to those people that I dedicate this thesis. From the bottom of my heart,
thank you.

iii

TABLE OF CONTENTS
Acknowledgements ......................................................................................................................... ii
List of Figures (Chapters 1-6) ......................................................................................................... v
List of Figures (Appendixes A-C) ................................................................................................. vi
List of Tables (Chapters 1-6) ........................................................................................................ vii
List of Tables (Appendixes A-C) ................................................................................................. viii
Abstract .......................................................................................................................................... ix
CHAPTER ONE: INTRODUCTION AND LITERATURE REVIEW ......................................... 1
1.1 Introduction ...................................................................................................................... 1
1.2 Review of remote sensing in glaciers studies................................................................... 3
1.3 Research question and objectives ................................................................................... 13
CHAPTER TWO: STUDY AREA ............................................................................................... 15
2.1 Study area physical and environmental description ....................................................... 15
CHAPTER THREE: DATA ......................................................................................................... 20
3.1 Study area characterization data ..................................................................................... 20
3.2 Global Land Survey (GLS) data..................................................................................... 26
3.3 Global Land Ice Measurement from Space (GLIMS) data ............................................ 32
CHAPTER FOUR: METHODOLOGY ....................................................................................... 35
4.1 Composite images and image processing....................................................................... 45
4.2 Glacier terminus delineation .......................................................................................... 50
CHAPTER FIVE: RESULTS ....................................................................................................... 62
5.1 Glacier movement qualification ..................................................................................... 66
5.2 Comparison of glaciers with similar terminal terrain conditions ................................... 79
5.3 Comparison of glaciers with dissimilar terminal terrain conditions .............................. 81
CHAPTER SIX: CONCLUSIONS AND SUGGESTED COMPLIMENTARY STUDIES ....... 87
6.1 Conclusions .................................................................................................................... 87
6.2 Lessons learned .............................................................................................................. 88
6.3 Suggested complimentary studies .................................................................................. 90
REFERENCES ............................................................................................................................. 93
APPENDIX A: GLOBAL LAND SURVEY (GLS) DATA SOURCING AND DOWNLOAD
..................................................................................................................................................... 102
APPENDIX B: GLOBAL LAND ICE MEASUREMENTS FROM SPACE (GLIMS) DATA
SOURCING AND DOWNLOAD .............................................................................................. 108
iv

APPENDIX C: GLACIER IMAGES USED TO CREATE THE GLACIER TERMINUSES
SHAPEFILES ............................................................................................................................. 112

v

LIST OF FIGURES (CHAPTERS 1-6)
Figure 1. Snow and ice discrimination with Landsat shortwave-infrared composite image (RGB:
4, 5, 7) graphic. ............................................................................................................................... 8
Figure 2. Central Southeast Alaska Glacier Study Project area orientation graphic. ................... 16
Figure 3. Central Southeast Alaska Glacier Study Project Advanced Spaceborne Thermal
Emission and Reflection Radiometer (ASTER) 30m digital elevation model (DEM). ................ 21
Figure 4. Central Southeast Alaska Glacier Study Project National Land Cover Data (NLCD
2001) graphic. ............................................................................................................................... 22
Figure 5. Alaska climate zones (traditional) graphic. ................................................................... 24
Figure 6. Alaska climate zones (revised) graphic. The project study area’s climate zone is further
defined as “Eastern Maritime”. Image source: Alaska History and Cultural Studies (2013). ...... 24
Figure 7. Alaska climate zones (expanded) graphic. .................................................................... 26
Figure 8. LeConte Glacier in GLS2010; RGB: 4, 5, 7. ................................................................ 30
Figure 9. Central Southeast Alaska Glacier Study Project: Global Land Ice Measurements from
Space data graphic. ....................................................................................................................... 33
Figure 10. Central Southeast Alaska Glacier Study Project slope graphic. .................................. 36
Figure 11. Central Southeast Alaska Glacier Study Project land use classification graphic. ....... 38
Figure 12. Glacier analysis process diagram. ............................................................................... 43
Figure 13. Landsat 7 ETM+ false-color shortwave composite image of Patterson Glacier (GLS
2005). ............................................................................................................................................ 45
Figure 14. Landsat 1 MS false-color near infrared composite image of Patterson Glacier (GLS
1975). ............................................................................................................................................ 46
Figure 15. “Composite Bands" tool dialog window completed for GLS2010 dataset using
Landsat 5 TM bands 4, 5, 7 (RGB). .............................................................................................. 47
Figure 16. ISO Cluster Unsupervised Classification for LeConte Glacier (GLS2010). ............... 49
Figure 17. "ISO Data Cluster Unsupervised Classification" tool dialog window completed for
GLS2000 dataset using Landsat 7 ETM+ bands 1-5 & 7. ............................................................ 50
Figure 18. Draw toolbar explained. .............................................................................................. 51
Figure 19. Draw toolbar continued. .............................................................................................. 52
Figure 20. Convert drawn graphics to features. ............................................................................ 52
Figure 21. Baird Glacier in GLS2010: the left graphic is Landsat 7 ETM+ near infrared (Band 4)
image and the right image is a natural color composite (RGB Bands 3, 2, 1). ............................. 53
Figure 22. Baird Glacier in GLS 2010: the left graphic is Landsat 7 ETM+ shortwave infrared
composite (RGB Bands 4, 5, 7) and the right image is an ISO Data Cluster Unsupervised
Classification (Bands 1, 2, 3, 4, 5, & 7). ....................................................................................... 54
Figure 23. Glacier edge deliniation for Baird Glacier (GLS2010). .............................................. 56
vi

Figure 24. Glacier valley buffers for Baird Glacier (GLS2010). .................................................. 56
Figure 25. Glacier centerline is completed for Baird Glacier (GLS2010). ................................... 57
Figure 26. Glacier centerline perpendicular is completed for Baird Glacier (GLS2010 ISO
Classification). .............................................................................................................................. 58
Figure 27. Baird Glacier movement measurement (partial). ........................................................ 59
Figure 28. Baird Glacier terminuses and perpendiculars for GLS2010, 2005, 2000, 1990, and
1975 datasets. ................................................................................................................................ 59
Figure 29. Patterson Glacier terminuses and perpendiculars for GLS2010, 2005, 2000, 1990, and
1975 datasets. ................................................................................................................................ 60
Figure 30. LeConte Glacier terminuses and perpendiculars for GLS2010, 2005, 2000, 1990, and
1975 datasets. ................................................................................................................................ 60
Figure 31. Shakes Glacier terminuses and perpendiculars for GLS2010, 2005, 2000, 1990, and
1975 datasets. ................................................................................................................................ 61
Figure 32. Glacier terminus results for the central southeast Alaska glacier GLS datasets. ........ 63
Figure 33. A summary of the movement distances for Baird, Patterson, LeConte, and Shakes
Glaciers during the periods of time covered by each GLS dataset. .............................................. 66
Figure 34. Slope at and within five-kilometers of Baird, Patterson, LeConte, and Shakes
Glaciers. ........................................................................................................................................ 68
Figure 35. Relationship between ice flow rates and temperatures. ............................................... 71
Figure 36. Mean yearly temperatures chart for 1973-2009. ......................................................... 71
Figure 37. The terminus conditions of Baird, Patterson, LeConte and Shakes Glaciers. ............. 74
Figure 38. This graph compares the movement of Shakes Glacier to the movement of Patterson
Glacier. .......................................................................................................................................... 81
Figure 39. Movement comparison for Baird Glacier versus Patterson Glacier, Baird Glacier
versus LeConte Glacier, and LeConte Glacier versus Patterson Glacier. ..................................... 85

LIST OF FIGURES (APPENDIXES A-C)
Figure 40. United States Geological Survey (USGS) Earth Explorer home page is the starting
point for downloading GLS datasets........................................................................................... 103
Figure 41. Define the area of interest for GLS image searches. ................................................. 104
Figure 42. Switch from AOD definition to dataset(s) selection. ................................................ 104
Figure 43. Specify the datasets for downloading. ....................................................................... 105
Figure 44. Earth Explorer search results page. ........................................................................... 105
Figure 45. GLS data download options. 11: Level 1 Product is selected. .................................. 106
Figure 46. Global Land Ice Measurements from Space (GLIMS) home page. .......................... 108
vii

Figure 47. GLIMS glacier database home page.......................................................................... 109
Figure 48. GLIMS glacier database graphical summary page for the entire world. ................... 109
Figure 49. GLIMS glacier database graphical summary for display window. ........................... 110
Figure 50. GLIMS data download page. ..................................................................................... 111
Figure 51. Baird Glacier in GLS2010 images used for glacier terminus delineation. ................ 113
Figure 52. Patterson Glacier in GLS2010 images used for glacier terminus delineation. .......... 114
Figure 53. LeConte Glacier in GLS2010 images used for glacier terminus delineation. ........... 115
Figure 54. Shakes Glacier in GLS2010 images used for glacier terminus delineation. ............. 116
Figure 55. Baird Glacier in GLS2005 images used for glacier terminus delineation. ................ 117
Figure 56. Patterson Glacier in GLS2005 images used for glacier terminus delineation. .......... 118
Figure 57. LeConte Glacier in GLS2005 images used for glacier terminus delineation. ........... 119
Figure 58. Shakes Glacier in GLS2005 images used for glacier terminus delineation. ............. 120
Figure 59. Baird Glacier in GLS2000 images used for glacier terminus delineation. ................ 121
Figure 60. Patterson Glacier in GLS2000 images used for glacier terminus delineation. .......... 122
Figure 61. LeConte Glacier in GLS2000 images used for glacier terminus delineation. ........... 123
Figure 62. Shakes Glacier in GLS2000 images used for glacier terminus delineation. ............. 124
Figure 63. Baird Glacier in GLS1990 images used for glacier terminus delineation. ................ 125
Figure 64. Patterson Glacier in GLS1990 images used for glacier terminus delineation. .......... 126
Figure 65. LeConte Glacier in GLS1990 images used for glacier terminus delineation. ........... 127
Figure 66. Shakes Glacier in GLS1990 images used for glacier terminus delineation. ............. 128
Figure 67. Baird Glacier in GLS1975 images used for glacier terminus delineation. ................ 129
Figure 68. Patterson Glacier in GLS1975 images used for glacier terminus delineation.. ......... 129
Figure 69. LeConte Glacier in GLS1975 images used for glacier terminus delineation. ........... 130
Figure 70. Shakes Glacier in GLS1975 images used for glacier terminus delineation. ............. 130

LIST OF TABLES (CHAPTERS 1-6)
Table 1. Landsat 1 Multispectral Scanner (MSS) spectral bands summary. .................................. 5
Table 2. Landsat 5 Thematic Mapper (TM) and Landsat 7 Enhanced Thematic Mapper Plus
(ETM+) spectral bands summary.................................................................................................... 6
Table 3. Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) 30m
digital elevation model (DEM) scenes which were used for this project. .................................... 20
Table 4. National Land Cover Data 2001 (NLCD 2001) scene that was used for this project. .... 22
viii

Table 5. Summary of the characteristics of the data collected at the Petersburg 1 meteorological
data collection point. ..................................................................................................................... 25
Table 6. Ancillary geospatial data that is used to create the various map graphics used in this
document. ...................................................................................................................................... 26
Table 7. Global Land Survey (GLS) sensor and imagery collection dates summary for Central
Southeast Alaska Glacier Study Project area. ............................................................................... 28
Table 8. World Reference System (WRS) image scene identification for central southeast Alaska
glacier study area imagery. ........................................................................................................... 28
Table 9. Cloud cover descriptive statistics for Landsat images collected during 2009. ............... 31
Table 10. Global Land Ice Measurement from Space database entries summary for Central
Southeast Alaska Glacier Study Project area. ............................................................................... 34
Table 11. Central Southeast Alaska Glacier Study Project percent slope computation summary.
....................................................................................................................................................... 37
Table 12. Central Southeast Alaska Glacier Study Project land use area by class computation
summary. ....................................................................................................................................... 39
Table 13. Baird, Patterson, and LeConte Glacier terminus distance summary during the various
time periods between GLS dataset collection events. ................................................................... 64
Table 14. Movement distance summary for each glacier from one GLS dataset collection event
to the next e.g. (GLS1975 to GLS1990, GLS1990 to GLS2000, and so forth). ........................... 64
Table 15. Summary of the average movement rates per year for Baird, Patterson, LeConte, and
Shakes Glaciers in each time period between GLS dataset collection events and average
movement for entire period covered by GLS1975 to GLS2010 datasets. .................................... 65
Table 16. Summary of glacier valley slopes for Baird, Patterson, LeConte, and Shakes Glaciers.
....................................................................................................................................................... 68
Table 17. Temperature trends for average monthly temperatures from 1973 to 2009. ................ 72
Table 18. Summary of glacier valley slopes for Baird, Patterson, LeConte, and Shakes Glaciers.
....................................................................................................................................................... 75
Table 19. Summary of glaciers in the North Cascades glacier study project. .............................. 76
Table 20. Movement summary for Shakes Glacier versus Patterson Glacier. .............................. 80
Table 21. Movement summary for Baird Glacier versus Patterson Glacier, Baird Glacier versus
LeConte Glacier, and LeConte Glacier versus Patterson Glacier. ................................................ 84

LIST OF TABLES (APPENDIXES A-C)
Table 22. Image details for Global Land Survey (GLS) datasets summary for Central Southeast
Alaska Glacier Study Project area. ............................................................................................. 107
Table 23. Glacier images page number summary. ...................................................................... 112
ix

ABSTRACT
Global Land Survey (GLS) data encompassing Landsat Multispectral Scanner (MSS),
Landsat 5’s Thematic Mapper (TM), and Landsat 7’s Enhanced Thematic Mapper Plus (ETM+)
were used to determine the terminus locations of Baird, Patterson, LeConte, and Shakes Glaciers
in Alaska in the time period 1975-2010. The sequences of the terminuses locations were
investigated to determine the movement rates of these glaciers with respect to specific physical
and environmental conditions.
GLS data from 1975, 1990, 2000, 2005, and 2010 in false-color composite images
enhancing ice-snow differentiation and Iterative Self-Organizing (ISO) Data Cluster
Unsupervised Classifications were used to 1) quantify the movement rates of Baird, Patterson,
LeConte, and Shakes Glaciers; 2) analyze the movement rates for glaciers with similar terminal
terrain conditions and; 3) analyze the movement rates for glaciers with dissimilar terminal terrain
conditions. From the established sequence of terminus locations, movement distances were
quantified between the glacier locations. Movement distances were then compared to see if any
correlation existed between glaciers with similar or dissimilar terminal terrain conditions. The
Global Land Ice Measurement from Space (GLIMS) data was used as a starting point from
which glacier movement was measured for Baird, Patterson, and LeConte Glaciers only as the
Shakes Glacier is currently not included in the GLIMS database.
The National Oceanographic and Atmospheric Administration (NOAA) temperature data
collected at the Petersburg, Alaska, meteorological station (from January 1, 1973 to December
31, 2009) were used to help in the understanding of the climatic condition in this area and
potential impact on glaciers terminus.
x

Results show that glaciers with similar terminal terrain conditions (Patterson and Shakes
Glaciers) and glaciers with dissimilar terminal terrain conditions (Baird, Patterson, and LeConte
Glaciers) did not exhibit similar movement rates. Glacier movement rates were greatest for
glaciers whose terminuses were in fresh water (Patterson and Shakes Glaciers), less for those
with terminuses in salt water (LeConte Glacier), and least for glaciers with terminuses on dry
land (Baird Glacier).Based upon these findings, the presence of water, especially fresh water, at
the terminal end of the Patterson and Shakes Glaciers had a greater effect on glacier movement
than slope. Possible explanations for this effect might include a heat sink effect or tidal motions
that hasten glacier disintegration in the ablation zone. In a heat sink scenario, the water bodies in
which the Patterson and Shakes Glaciers terminus are located could act as a thermal energy
transfer medium that increases glacier melting and subsequent retreat. On the other hand, tidal
motions could act as horizontal and vertical push/pull forces, which increase the fracturing rate,
calving, and subsequent retreat of glaciers terminus that are is salt water like the LeConte
Glacier.
Over the length of the study period, 1975 through 2010, there has been a 0.85°C increase
in annual air temperatures that, although may seem low, may prove important when determining
glacial mass balance rates. Further studies are necessary to test these hypotheses to determine if a
heat sink effect and tidal motions significantly affected the movement rates for the glaciers in
this study area.
An additional significant result of this study was the creation of shapefiles delineating the
positions of the Shakes Glaciers that are being submitted to the Global Land Ice Measurements
from Space (GLIMS) program for inclusion in their master worldwide glacier database.
1

CHAPTER ONE: INTRODUCTION AND LITERATURE REVIEW
1.1 Introduction
Worldwide, glaciers are estimated to cover about 10% of all land mass and hold 69% of all fresh
water on earth (United States Geological Survey, 2012). Glacier “health” is indicative of climate
“health” (Michna, 2012). This means that if the world’s climate is generally cooler, glaciers
should grow larger and advance forward. If glaciers are growing larger and advancing, then by
extension the climate should be getting cooler. It can also be said that glaciers, as the extensions
of ice caps and sheets, are indicative of overall “health” of the parent ice cap or sheet. As an ice
cap or sheet fluctuates in size, its associated glaciers should also fluctuate in size and move
accordingly.
Glaciers are an important source of dissolved organic matter (DOM), in the form of labile
carbon, for riverine and estuarine ecosystems (Hood et al., 2009). In a study of eleven coastal
watersheds along the Gulf of Alaska; Hood et al. (2009) found that glacial runoff is a significant
source of beneficial carbon for all ecosystems that are downstream of a glacier. Any reduction in
glacier ice mass is detrimental to the availability of DOM in the dependent ecosystems (Hood et
al. 2009). It is estimated that the Gulf of Alaska river drainage basins contain more than 10% of
all mountain glaciers on Earth and the annual runoff from these systems is the second largest for
the Pacific Ocean (Hood et al., 2009). The Gulf of Alaska and connected water bodies contain
several of the most productive salmon, ground fish, and shellfish fisheries in the world (Alaska
Department of Fish and Game, 2012). In 2011, Alaskan fisheries supported jobs for 78,500
people and generated 5.8 billion dollars from sales and services of which the Gulf of Alaska
fisheries provided a significant percentage (Alaska Department of Fish and Game, 2012). For
2

economic as well as ecologic reasons, precise glacial measurements is necessary to provide
scientific data that help the management of Alaskans’ fisheries that represent a reliable source of
food for native Alaskans and worldwide consumption.
The natural conditions that favor the formation of glaciers, like high elevations and
inhospitable climate and weather, make on-site glacier study difficult. While these variables
make direct glacier study problematic, they also directly impact glacier movement rates.
Waddington (2009) attributes glacier movement rates to several physical and environmental
factors; like glacial bed slope and warming temperatures. The warm, moist maritime climate
present in the study area, coupled with a relatively wide range of glacial bed slopes, resulted in a
variety of glacial movement rates for the glaciers studied for this project. Many glaciers require
substantial effort for researchers to approach, necessitating the use of aircraft in often hazardous
conditions. Remote sensing, as indirect method of detection of an object or phenomenon without
direct human contact (Noderer, 2007), whether by using an imaging sensor flown by aircraft or
satellite platform, provide the means for glacial monitoring. Satellite imagery such as Landsat
(Irons, 2013), RADARSAT (Canadian Space Agency, 2013), and ASTER (Jet Propulsion
Laboratory, 2013.) are successful in providing worldwide imaging capability that is repeatable
and consistent.
Glaciers are studied to increase our understanding of glacier processes. A great deal of
effort has been spent in documenting world glaciers by the development of databases of
worldwide glaciers, such as the Global Land Survey (GLS; http://gls.umd.edu/) and the Global
Land Ice Measurements from Space (GLIMS) project (Raup et al., 2007). It is currently
estimated that there are between 70,000 and 200,000 glaciers worldwide which, due to the
3

immense number of glaciers, necessitates automated or automatic methods to map them (Aher
and Dalvi, 2012). Remote sensing of glaciers has been expanded to include ice caps, fields, and
sheets; all of which are differentiated by size and shape. Ice sheets and caps are domed shaped,
icefields are flat; ice caps and fields are less than 50,000km
2
and ice sheets are greater than
50,000km
2
(Pidwirny and Jones, 2010).
Baird, Patterson, LeConte, and Shakes Glaciers originate from the Stikine Icefield. The
Stikine Icefield is a very large icefield, approximately 190km in length from its southern border
at the Stikine River to its northern border at the Taku River, straddling the southeastern Alaska
and British Columbia (Canada) border (Molnia, 2008). Rapper and Braithwaite (2006) concluded
that the melting rate for glaciers is different than the rate for ice caps or icefields; i.e. glaciers
melt faster. This is analogous to leaving a large block of ice and the contents of an ice cube tray
in the sun; the individual cubes will melt faster than the block of ice due to the significantly
larger percentage of surface area compared to volume. In order to accurately monitor icecap
mass change, it is necessary to consider glaciers and icecaps separately. After determining the
mass change for an icecap and its included glaciers separately, the results can be combined to
determine a final mass change figure that is indicative of the whole glacier-icecap system.
1.2 Review of remote sensing in glaciers studies
Remote sensing is considered reliable, safe, and a cost-effective method for assessing the
world’s glaciers. Several projects are gathering and processing data for monitoring purposes and
worldwide use that are based on precursory studies that set the path for suitable procedure and
the need for historical data archives. A literary search for remote sensing of glaciers returns an
overwhelming amount of material and data archives; such as the Global Land Survey (GLS) and
4

the Global Land Ice Measurements from Space (GLIMS) project. GLS and GLIMS are described
in detail based on the data type and main characteristics in support of regional glacier studies. It
is important to note that these initiatives took place as a response to standardize glaciers study
procedures and therefore several references are a collection of previous studies on which the
glaciers study community is built upon. The focus is on low-cost data and procedures for
regional glaciers monitoring such as in this study. However, it is important to mention that other
remote sensing techniques such as laser altimetry, also known as Light Detection and Ranging,
or LiDAR (National Oceanic and Atmospheric Administration, 2013), are used to measure the
surfaces of glaciers, although they are currently at large scale, e.g. Mendenhall and Taku
Glaciers by Hekkers (2010). LiDAR enable the rendering of very accurate three-dimensional
modeling of a glacier’s surface. However, this kind of technology is limited at this point to small
areas and is not cost-effective for larger studies and therefore was not considered in this study.
The Global Land Survey (GLS), a partnership between the U.S. Geological Survey
(USGS) and the National Aeronautics and Space Administration (NASA) in support of the U.S.
Climate Change Science Program (CCSP) and the NASA Land-cover and Land-use Change
(LCLUC) Program, builds on the existing geo-cover data sets developed for 1975, 1990, and
2000’s. Some 9,500 Landsat images, acquired in the period 2004-2007, are processed and made
available to the public. Given the failure of the Landsat-7 ETM+ Scan-Line-Corrector (SLC) in
2003, a combination of Landsat-7 gap-filled data and Landsat-5 data is used to create the
GLS2010 dataset (Haq et al., 2012). The GLS is a global dataset of 30-meter resolution satellite
imagery to support measurement of Earth's land cover and rates of land cover change during the
first decade of the 21st century (Haq et al., 2012).
5

Since GLS data is collected using Landsat multispectral remote sensing satellites, a short
review of the Landsat constellation is useful to highlight the capabilities, advantages, and
disadvantages of this data source. Landsat satellites orbit in a near-polar orbit; that is they do not
pass directly over the poles, rather they are slightly offset (Lillesand, Kiefer, and Chipman,
2008). Although Landsats 1-3 orbits at a different altitude than Landsats 5-8, all have an along
track swath width of 185km (United States Geological Survey, 2012). Because of the difference
in orbit altitude, Landsats 1-3 have a scene revisit time of 18 days and Landsats 5-8 have a scene
revisit time of 16 days (United States Geological Survey, 2012). It should be recognized that
Landsats 5 and 7 are eight days apart which reduces a scene revisit time to eight days when the
two satellites are used together (Noderer, 2007). Imagery characteristics of the Landsat 1’s
Multispectral Scanner (MSS) spectral bands are summarized in Table 1, and Landsat 5’s
Thematic Mapper (TM) spectral bands and Landsat 7’s Enhanced Thematic Mapper Plus
(ETM+) spectral bands in Table 2.
Table 1. Landsat 1 Multispectral Scanner (MSS) spectral bands summary. Landsat 1 lacks a visible blue
image band, which makes constructing a true color composite image difficult. It also lacks shortwave
infrared capability. It is important to note that although the band numbers between MSS and later sensors
do not correlate, the Electromagnetic Spectrum regions sensed are very similar.
Band
Number
Electromagnetic
Spectrum Region
Wavelength
(µm)
Spatial Resolution
(m)
Band 4 Visible Green 0.5-0.6 80
Band 5 Visible Red 0.6-0.7 80
Band 6 Near-Infrared 0.7-0.8 80
Band 7 Near-Infrared 0.8-1.1 80

6

Table 2. Landsat 5 Thematic Mapper (TM) and Landsat 7 Enhanced Thematic Mapper Plus (ETM+)
spectral bands summary. TM and ETM+ differ primarily in the panchromatic band’s spatial resolution.
For this study, the panchromatic band is not used for analysis; which makes the images obtained by TM
and ETM+ interchangeable.
Band Number
Electromagnetic
Spectrum Region
Wavelength
(µm)
Spatial
Resolution
(m)
Band 1 (TM & ETM+) Visible Blue 0.45-0.52 30
Band 2 (TM & ETM+) Visible Green 0.52-0.61 30
Band 3 (TM & ETM+) Visible Red 0.63-0.69 30
Band 4 (TM & ETM+) Near-Infrared 0.76-0.90 30
Band 5 (TM & ETM+) Shortwave-Infrared 1.55-1.75 30
Band 6 (TM & ETM+) Thermal-Infrared 10.40-12.50 120
Band 7 (TM & ETM+) Shortwave-Infrared 2.08-2.35 30
Band 8 (ETM+ Only) Panchromatic (Visible) 0.52-0.90 15
Landsat imagery has advantages and disadvantages depending upon what question the
imagery is used to answer. The Landsat program has been collecting imagery since 1972, which
is over forty years (Earth Resources Observation and Science Center, 2012). This longevity has
provided a data library with more than 3.3 million images as of February, 2012 (United States
Geological Survey, 2012). This very large image library that spans more than 40 years allows for
temporal analysis of earth’s features, like glaciers. The large image swath width of 185 km
reduces the number of images needed for most projects applied to glaciers studies (United States
Geological Survey, 2012). While a large image swath is an advantage in most situations, the
course multispectral pixel spatial resolution of 80m for Landsat 1 and 30m for Landsats 5 and 7
do not provide the resolution needed for analyzing small features, like parking lot congestion
7

(United States Geological Survey, 2012). However, the spatial resolution is sufficient for
analyzing features that cover large areas, like agriculture, forestry, or glaciers.
Landsat multiple spectral bands (four for Landsat 1 and seven for Landsats 5 & 7) allow a
better differentiation of features or phenomena. Ice and snow are often visually identical in the
visible region of the electromagnetic (EM) spectrum; however, they are sharply contrasting in
the shortwave-infrared regions making it possible to differentiate the many transitional areas that
are difficult to distinguish as either glacial ice or surrounding snow (Baolin et al., 2004; Noderer,
2007). Baolin et al. (2004) suggest that the most appropriate method is to perform either a
supervised classification or a semi unsupervised classification on every image scene to ensure
that the results are accurate (Baolin et al. 2004).
In a recent study by Haq et al. (2012) determined that false-color SWIR composite (RGB:
4, 5, 7) provided the most visual interpretability. This band combination accentuates differences
between snow and ice so that the extent of both features can be determined. When this image
composite is viewed, ice is a dark red color and snow is a very discernibly lighter shade of red.
Figure 1 illustrates the visual difference between areas of snow and ice when viewed in a
shortwave-infrared composite. Haq et al. (2012) also points out that the uneven terrain of glaciers
necessitates topographically correcting imagery radiance values. Shadowing effects caused by
terrain can negatively affect the brightness values of pixels so that material identification is
difficult. As a result, a normalized difference snow index function was used to identify image
pixels comprised of ice (Haq et al., 2012).
8

Figure 1. Snow and ice discrimination with Landsat shortwave-infrared composite image (RGB: 4, 5, 7)
graphic. In this false-color composite, snow is bright red and ice is dark red. The line between snow and
ice is visually discernable. Image source: Haq et al., 2012.
In a study by Baolin et al. (2004) glaciers position information is assessed by detecting
glacial marginal fluctuations using multi-temporal images. Digital image classification and
change detection yield root mean square errors (RMSE) of less than a pixel, which is sufficient
to produce measureable results. The results were checked with ground-truth data that is
generated by a visual inspection of the imagery and not by ground-based surveys (Baolin et al.
2004). The advantage to creating in-scene ground-truth data is that it can be done in a controlled
environment although it often requires significant experience in image interpretation techniques.
In a recent study by Haq et al. (2012) Landsat multispectral imagery from the early 1970s to
2010 is used to compare the size, shape, and location of the Gangotri Glacier. This type of
analysis reveals with a reasonable degree of accuracy how a feature changes over time. It should
be pointed out that Haq et al. (2012) did not use imagery with a particular anniversary date,
which can create problems when trying to compare results and determine the true movement of a
glacier.
9

The GLS data has been used for numerous studies that rely on temporal change detection
to monitor earth processes and human development activities. Lindquiest et al. (2008) uses
GLS2000 and GLS2005 data to monitor tropical forest cover change in the Democratic Republic
of the Congo. Likewise, Beuchle et al. (2011) relies on GLS1990 and GLS2000 to access the
deforestation of tropical forests. Gutman et al. (2008) predicts that GLS2005 data will be critical
in analyzing trends in:
 forest cover change to include deforestation and replanting efforts;
 agriculture expansion in arid regions that rely heavily on irrigation techniques;
 lingering flooding of low-lying areas and the negative impact these areas have on human
health;
 arctic terrain changes as areas with permafrost begin warming and transforming;
 increased human-footprint as evidenced by expanding urban areas and urbanization of
traditionally nonurban areas.
While GLS data has been used in numerous independent studies, it is also used as
foundational data for several national programs administered by the United States Government.
The US Climate Change Science Program has an executive mandate to provide a systematic
measurement of changes in global land cover (Justice, 2013). Wolfe et al. (2004) relies on
GLS1975, GLS1990, and GLS2000 data as a primary input for the Landsat Ecosystem
Disturbance Adaptive Processing System (LEDAPS). LEDAPS has the critical task of mapping
change detection for North American forests so that accurate carbon modeling can be developed.
Masek (2011) identifies LEDAPS as a cornerstone of the National Aeronautics and Space
Administration’s contribution to the North American Carbon Program (NACP) and an invaluable
10

tool for the United States Global Change Research Program (USGCRP) Carbon Cycle Science
Program. Although GLS heavily relies upon Landsat imagery, this reliance may shift to
alternative imagery sources. Gutman et al. (2008) concluded that with Landsats 5 and 7 reaching
the end of their useful life due to onboard fuel cell depletion, future GLS datasets will require
international cooperation to produce. This will shift the focus away from a US centric data
collection strategy to a worldwide responsibility to continue to monitor the changing world so
that human use of natural resources can be planned with sustainability as a cornerstone of any
development plan.
Another crucial initiative is The Global Land Ice Measurements from Space (GLIMS)
project, which is currently creating a unique glacier inventory storing critical information about
the extent and rates of change of the world's estimated 160,000 glaciers. GLIMS is an
international collaborative project, that includes more than sixty institutions world-wide, to
create a globally comprehensive inventory of land ice including: measurements of glacier area,
geometry, surface velocity, and snow line elevation, retreat, wasting, and thinning (Raup et al.
2007). To perform these analyses, the GLIMS project uses satellite data, primarily from the
Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) and the Landsat
Enhanced Thematic Mapper Plus (ETM+) as well as historical information derived from maps
and aerial photographs. Due to the very large number of glaciers worldwide, no single analysis
center or group is responsible for all glaciers; rather, a series of regional centers are responsible
for their region of the world (Raup et al. 2007).
Several studies have been conducted as a way to test the GLIMS database. Haritashya et
al. (2009) evaluated several imagery sources, including Landsat, to perform a study over a
11

twenty-seven year period for the Wakhan Corridor of Afghanistan which concluded that many
glaciers in the region have retreated from historic positions. Bishop et al. (2004) acknowledge
that assessing glacier mass-balance with space-borne remote sensing is very challenging and the
work associated with GLIMS greatly assists in developing new methods for studying glaciers
and the complex relationship between glaciers and their environments. GLIMS-based research is
critical to developing new methods for mapping glaciers from spatial data where glacier features
might be obscured. This research is critical since it is very difficult to develop and maintain a
world glacier inventory if glaciers have to be manually differentiated from their surroundings
(Bishop et al., 2004). Arendt et al. (2012) describe the Randolph Glacier Inventory (RGI) as a
global catalogue of glaciers that is intended to supplement the GLIMS database. The RGI used
satellite imagery and other data to catalogue worldwide glaciers (Arendt et al., 2012). Arendt et
al. (2012) continues to say that GLIMS data provided valuable information for glaciers in several
regions of the world. In addition to GLIMS, data was ingested into RGI from the World Glacier
Inventory (WGI) (Arendt et al., 2012). As of 2012; the WGI contains entries for more than
130,000 glaciers worldwide (National Snow and Ice Data Center, 2013). As of 2007, the
GLIMS database contained more than 52,000 glaciers (GLIMS: Global Land Ice Measurements
from Space, 2007). These values are significant considering that there is an estimated 70,000 to
200,000 glaciers worldwide (Aher et al. 2012).
The GLIMS is not a wholly original effort as it takes the World Glacier Inventory (WGI)
model and expands the parameters from a single point, which represents a glacier, to an outline
shapefile that accurately describes the face of a glacier (Raup et al., 2007). Raup et al. (2007)
continues to describe the GLIMS as a complimentary system that allows forward and backwards
12

compatibility with WGI. This is an important lesson that geospatial database designers should be
cognizant of in an era of declining resources and increasing expectations; leverage the work of
others as often as possible. GLIMS provide online tools that allow researchers to access,
download, and analyze glacier data and imagery for free. This level of transparency is critical to
the longevity of GLIMS as it allows anyone to access the data and encourages diverse users to
submit data to increase the coverage of GLIMS. Raup et al. (2007) describes the GLIMS as a
capable toolset for recording, measuring, and assessing glaciers.
The GLIMS is more than a data repository as it contains also online tools, such as
standard indices, supervised classification, and geomorphology-based methods, which allow
researchers with internet access to analyze remotely sensed imagery to access glacier health
(Raup et al. 2007). The online tools allow analysts to digitize the extent of a glacier after they
perform image enhancement or feature extraction processes (Raup et al. 2007); results can be
submitted to GLIMS for inclusion in the database. This approach allows the GLIMS to leverage
the talents of potentially millions of analysts for free. This is similar to the approach that the
Search for Extraterrestrial Intelligence (SETI) program uses where it allows nonscientists to
process radio telescope data with their personal computers and submit results back to the SETI
program administrators (Space Sciences Laboratory, 2013).
Although many people contribute to GLIMS, the task of continuously updating GLIMS is
daunting. Current estimates of 70,000 to 200,000 glaciers worldwide necessitate automated or
automatic methods for monitoring their health (Aher et al. 2012). Considering the exponential
increase in computer processing power, software functionality, and decades of multispectral
imagery, scientists should have the tools and techniques necessary to monitor every glacier
13

(National Aeronautics and Space Administration, 2012). However, the reality is that many of the
automated tools are problematic and require much effort to ensure that the results are accurate
and reliable.
It is useful to mention that Quincey and Luckman (2009) evaluated the utility of multiple
types of alternative remote sensing data: synthetic aperture radar (SAR) interferometry, feature
tracking, scatterometry, altimetry, and gravimetry. Traditionally, researchers have been limited
to optical sensors to collect remote sensing data for studying ice sheets and glaciers. However,
advances in radar altimetry, gravimetric, and microwave technologies allow researchers to
analyze glacial movement, melting, swelling, and contracting (Quincy et al. 2009). This
approach may provide a three-dimensional profile of a glacier to better understand if a particular
glacier is shrinking or enlarging. This multitude of new data sources will continue to improve the
understanding of glacial processes.
1.3 Research question and objectives
The ability to characterize the movement rates of Baird, Patterson, LeConte, and Shakes
Glaciers using low resolution and cost effective remote sensing imaging data was the main
research question in this study. In addition comparative analysis of the movements rates of the
glaciers with respect to specific physical and environmental conditions were conducted. The
investigation used GLS false color composite images enhancing ice-snow differentiation and
Iterative Self-Organizing (ISO) Data Cluster Unsupervised Classification to achieve three
measureable objectives:
1) movement of Baird, Patterson, LeConte, and Shakes Glaciers,
2) movement rates for glaciers that have similar terminal terrain conditions,
14

3) movement rates for glaciers with dissimilar terminal terrain conditions.
These measurements were compared against the GLIMS database to assess the relative glacier
movements and their behavior with respect to specific glacier physical and environmental
conditions.

15

CHAPTER TWO: STUDY AREA
2.1 Study area physical and environmental description
The project study area is located in the Southeastern region of Alaska and within the Alexander
Archipelago. The Alexander Archipelago extends west of the British Columbia provincial border
line and constitutes most of the land area of the Tongass National Forest (Figure 2); hereafter
referred to as simply the Tongass. The Tongass is an expansive temperate rain forest with sparse
human habitation. The closest town to the study area is Petersburg, Alaska; which is located
several kilometers west of the study area. The two closest urban areas are Anchorage, Alaska
(1,100km to the north) and Seattle, Washington (1,250km to the south). Within the study area is
Baird, Patterson, LeConte, and Shakes Glaciers.
16

Figure 2. Central Southeast Alaska Glacier Study Project area orientation graphic. The approximate
geographic center of the study area is at 57.232°N and 132.503°W. The approximate distance between the
terminuses of the northernmost (Baird) and southernmost glaciers (Shakes) is 51km. Baird Glacier
discharge into Thomas Bay. Patterson Glacier also discharges into Thomas Bay via the Patterson River.
LeConte Glacier, the only tidewater glacier in the study, discharges into LeConte Bay. Shakes Glacier,
via Shakes Slough and the Stikine River, discharges into Frederick Sound. The community of Petersburg,
located on Mitkof Island, is nearby.
Approximately 25 km northeast of Petersburg, Alaska is Thomas Bay. At the head of
Thomas Bay are Baird and Patterson Glaciers. Both Baird and Patterson Glaciers discharges melt
water into Thomas Bay; Baird Glacier discharges directly and Patterson Glacier discharges via
the Patterson River. Approximately 30km east of Petersburg, Alaska is LeConte Bay. LeConte
Bay is headed by LeConte Glacier. Near LeConte Glacier is Shakes Glacier, which discharges
17

into the marine environment via a connected slough and river. On many occasions, I have been
to LeConte Glacier, either by boat or helicopter. I have also been to Shakes Glacier by boat
several times. Because of my many visits, I am intimately familiar with the area and this
familiarization will be very beneficial for image classification processes.
Tongass
The Tongass extends from 54.5°N to 60.0°N; about 800km. It has an area of
approximately 68,790km
2
, which makes it the largest national forest in the United States and the
largest intact temperate rainforest in the world (Cape Decision Lighthouse Society, 2013; United
States Department of Agriculture, Forest Service, 2013). The Tongass has diverse topology,
vegetation, and seasonal climate variations and according to the 2010 Census, is home to
approximately 71,664 people (State of Alaska Department of Labor and Workforce
Development, 2013). The highest point in the Tongass is Kates Needle, a peak on the United
States and Canada border with an elevation of over 3,000m (Schweiker & Olson, 2012). The
lowest elevation is sea level. Slopes range from zero slope (flat) to no slope (vertical cliff).
Vegetation also ranges from the very large, like Sitka spruce, to the very small, like mosses and
lichen. A very diverse range of species inhabit the lands and waters; moose, deer, humpback
whales, salmon, ducks, geese, and bald eagles are numerous.
The five largest communities, which have population concentrations of over 2,000
people, are Juneau, Sitka, Ketchikan, Petersburg, and Wrangell; the combined population in
2010 was 53,523 people which represent approximately 75% of the total population of the
Tongass (United States Census Bureau, 2013). Those same communities have an estimated
combined area of 113km
2
as measured on IKONOS imagery, which is 0.2% of the entire area of
18

the Tongass (Statewide Digital Mapping Initiative, 2012; United States Department of
Agriculture, Forest Service Geospatial Service and Technology Center, 2013). As a comparison,
the combined land area of Massachusetts, Vermont, and New Hampshire is approximate to the
size of the Tongass, but the 2010 census population of these three states is 8,489,400 people, or
12,792% of the population in the Tongass (United States Census Bureau, 2013; Environmental
Systems Research Institute, 2013). The study area varies greatly in 1) topology, 2) land cover,
and 3) climate. Topology, land cover, and climate data (Chapter 3-Data, and 4-Methodology)
were used to characterize the study area and were assessed quantitatively and qualitatively
(Chapter 5-Results) to determine their impact on glacier movement within the study area.
Survey history of Baird, Patterson, LeConte, and Shakes Glaciers
Baird, Patterson, LeConte, and Shakes Glaciers are the subject of many geophysical
studies, surveys, and imagery collection events (Molnia, 2008). Many of the early surveys date
back to the 1800s; which corresponds with the necessity to map and survey the Alaska territory
after it had been purchased from Russia in 1867 (Billington, 2013). Molnia (2008) chronicles the
earliest survey or imaging of these glaciers as follows:
 Baird Glacier: United States Coast and Geodetic Survey, 1887
 Patterson Glacier: United States Coast and Geodetic Survey, 1879
 LeConte Glacier: United States Coast and Geodetic Survey, 1887
 Shakes Glacier: Aerial Survey of Alaska, 1948
In the time since these glaciers were first surveyed or imaged, they have continued to be studied
with increasingly sophisticated instruments and/or collection methods (Molnia, 2008):
19

 land-based and manual bathymetric surveys
 aircraft and panchromatic film cameras
 satellite-based imagery
 airborne light detection and ranging (LiDAR) surveys
 global positioning system (GPS) surveys
 remotely operated underwater vehicles
As the technology progressed, it was applied to glacier surveys. It is logical to conclude that as
gravimetric, thermal, or hyperspectral imaging becomes more prevalent and available, it will be
applied to studying glaciers; especially Baird, Patterson, LeConte, Shakes and the other 100,000
estimated glaciers that are located in Alaska (Molnia, 2008).

20

CHAPTER THREE: DATA
Data used for the physical characterization of the study was publicly available and mainly
consisted of raster data and ancillary GIS datasets. The analysis of the glaciers terminus was
conducted with publicly available raster data from the Global Land Survey (GLS) and Global
Land Ice Measurement from Space (GLIMS) repository data sources.
3.1 Study area characterization data
Topology
Advanced Spaceborne Thermal Emission and Reflection Radiometer
(http://asterweb.jpl.nasa.gov/) Global Digital Elevation Model Version 2
(http://asterweb.jpl.nasa.gov/) slope values for the study area were calculated (United States
Geological Survey, 2013). The study area is on the border between two ASTER scenes (Table 3)
as shown in Figure 3.
Table 3. Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) 30m digital
elevation model (DEM) scenes which were used for this project. The project required two separate scenes
(ASTGTM2_N56W133 and ASTGTM2_N57W133) to cover it entirely.
Sensor Raster Scenes Numbers Source
Image
Collection
Date Data Type
Project
Use
Advanced
Spaceborne
Thermal Emission
and Reflection
Radiometer 30m
Digital elevation
model
ASTGTM2_N56W133
ASTGTM2_N57W133
Alaska
Statewide
Digital
Mapping
Initiative
January
2000
Elevation
data
Project
study
area
slope
graphic

21

Figure 3. Central Southeast Alaska Glacier Study Project Advanced Spaceborne Thermal Emission and
Reflection Radiometer (ASTER) 30m digital elevation model (DEM). The thick-brown diagonal line
illustrates the border between ASTER scenes: ASTGTM2_N56W133 and ASTGTH2_N57W133. In this
graphic, elevation is shown as gray-scale; the lowest elevation is 0m and the highest elevation is 2,893m.
The total area encompassed for this graphic was 1,917.7km
2
.
Land cover
The land cover classification for the study area was derived from the National Land
Cover Dataset, 2001 (United States Geological Survey, 2013). National Land Cover Data 2001
(NCLD 2001) data was derived from Landsat imagery and therefore retains the source data’s
30m spatial resolution (Table 4). In Figure 4 is shown the original NLCD coverage before a
reclassification process was applied for the characterization of the study area.
22

Table 4. National Land Cover Data 2001 (NLCD 2001) scene that was used for this project.
Dataset Type Source Date Features Use
National
Land Cover
Data 2001
Raster
United States
Geological Survey
March
2008
Land cover
Project study area land
cover classification
graphic

Figure 4. Central Southeast Alaska Glacier Study Project National Land Cover Data (NLCD 2001)
graphic. The study area was classified into 12 distinct categories. The “Not Classified” area is located in
Canada. Because NLCD 2001 is a US only dataset, areas located in Canada are not classified. The total
area encompassed for this classification is 1,916.4km
2
.
23

Climate and weather
Traditionally, climatologists classify Alaska into several distinct climate zones: arctic,
continental, and maritime; which is shown in Figure 5 (Alaska Climate Research Center, 2010).
In this classification, the Tongass and this project’s study area were located in a maritime climate
zone. The Western Regional Climate Center (2013) also characterizes the climate of the Tongass
as maritime in nature with annual precipitation amounts of up to 508cm and average temperature
from the -6.6s to the 15.5s (°C) depending upon season. The Alaska History and Cultural Studies
(2013), as shown in Figure 6, further distinguish the climate of the Tongass as Eastern Maritime.
In a climate division study of Alaska, Bieniek, Bhatt, Thoman, Angeloff, Partain, Papineau,
Fritsch, Holloway, Walsh, Daly, Shulski, Hufford, Hill, Calos, and Gens (2012) consider
localized temperature and precipitation amounts to further subdivide the major climate zones of
Alaska into smaller, more homogenized regions. Bienieket al. (2012), as shown in Figure 7,
concludes that the Tongass can be subdivided into four smaller climate zones: North Panhandle,
Northeast Gulf, Central Panhandle, and South Panhandle. In this climate classification, the
project study area was located within the proposed Central Panhandle climate region.
24

Figure 5. Alaska climate zones (traditional) graphic. The project study area is located in a “Maritime”
climate zone (lower right corner of the image). Image source: Alaska Climate Research Center (2010).

Figure 6. Alaska climate zones (revised) graphic. The project study area’s climate zone is further defined
as “Eastern Maritime”. Image source: Alaska History and Cultural Studies (2013).
Although climate variables were not extensively considered for the analysis of glacier
movements in this study, the National Oceanographic and Atmospheric Administration (NOAA)
temperature data collected at the Petersburg, Alaska meteorological station from January 1, 1973
25

to December 31, 2009 was used to derive a general climate trend. Petersburg, Alaska is the
closest meteorological data collection point to the study area. It is understood that atmospheric
conditions in Petersburg, Alaska, are only close approximations for the conditions near Baird,
Patterson, LeConte, and Shakes Glaciers. A summary of the characteristics of the NOAA
meteorological data is provided in Table 5.
Table 5. Summary of the characteristics of the data collected at the Petersburg 1 meteorological data
collection point.
NOAA
Meteorological
Station ID
Data
Source
Data
Type
Data Collection
Period Start
Date
Data Collection
Period End
Date
Observation
Frequency
Number of
Possible
Observations *
Number of
Actual
Observations *
Petersburg 1 NOAA Text January 1, 1973
December 31,
2009
Monthly
Average
444 360
*
Due to the lack of available data for many of the observation collection events, especially 1978-1980 and 1996-
2000, the number of possible observations is different than the number of actual observations.
The prevalent weather conditions of a maritime climate zone is rain; often hundreds of
centimeters annually. Rain, and the clouds that produce rain, often obscure the surface of the
earth from remote sensing satellites and aircraft. This can create a serious problem in acquiring
useable data at a specific point in time. For this project, thousands of images were reviewed to
select the final images that were used to complete this project.
26

Figure 7. Alaska climate zones (expanded) graphic. The project study area’s climate is further refined as
“Central Panhandle”. Image source: Bienieket al. (2012).
Ancillary data
These data encompass GIS data layers (Table 6) used as spatial context for the study area.

Table 6. Ancillary geospatial data that is used to create the various map graphics used in this document.
Dataset Type Source Date Features Use
Alaska coastline Vector
Alaska State Geo-
Spatial Data
Clearinghouse
February
1998
Coastal
shoreline
Project study
area graphics
Alaska hydrography Vector
Alaska State Geo-
Spatial Data
Clearinghouse
January
2007
Linear
hydrography
features
Project study
area graphics
3.2 Global Land Survey (GLS) data
Global Land Survey (GLS) data is a partnership between the United States Geological
Survey (USGS) and the National Aeronautics and Space Administration (NASA) to create a
global imagery mosaic for regular anniversary dates: 1975, 1990, 2000, 2005, and 2010 (Earth
Resources Observation and Science Center, 2012). The data is collected using the latest available
Landsat sensor for each collection period and provides the necessary imagery for accomplishing
27

this study objectives. Table 7 identifies which sensor was used to collect imagery for each
dataset and the range of image collection dates that each dataset encompasses. For this study the
GLS1975, GLS1990, GLS2000, GLS2005, and GLS2010 datasets containing all the imaging
bands were downloaded from the United States Geological Survey Earth Explorer (2013) portal.
All datasets were preprocessed at the data source to Level 1 standards. A Level 1 product
corrects for either sensor detector variations, image geometry, or both (Piwowar, 2001). The
GLS data acquisition is covered step-by-step in Appendix A.
For this study the panchromatic band was not used, making the images obtained by
Landsat TM and ETM+ interchangeable. Table 7 summarizes the collection dates and collection
sensor for each GLS dataset.

28

Table 7. Global Land Survey (GLS) sensor and imagery collection dates summary for Central Southeast
Alaska Glacier Study Project area. Imagery from Landsats 1, 5, and 7 are used in this study.
Dataset Sensor Collection Dates
GLS1975 Landsat 1-5 1972-1987
GLS1990 Landsat 4
*
-5 1987-1997
GLS2000 Landsat 7 ETM+ 1999-2003
GLS2005 Landsat 5 TM, Landsat 7 ETM+, EO-1 Ali 2003-2008
GLS2010 Landsat 5 TM, Landsat 7 ETM+ 2008-2011
*
Landsat 4 was not used in this work
Landsat imagery is collected in a grid pattern and each image scene has a unique row and
path identification number that is referred to as the Worldwide Reference System (WRS) (Irons,
2013). Because Landsats 1-3 are flown at a different altitude than Landsats 4-7, there is a
difference in the WRS identification number. The WRS for Landsats 1-3 is referred to as WRS-1
and the WRS for Landsats 4-7 is referred to as WRS-2 (United States Geological Survey, 2013).
Table 8 summarizes the WRS identification numbers for the GLS data used for this study.
Table 8. World Reference System (WRS) image scene identification for central southeast Alaska glacier
study area imagery.
GLS Dataset Imagery Source Date Landsat Satellite WRS Version Path ID Row ID
GLS1975 September 3, 1974 Landsat 1 WRS-1 60 20
GLS1990 September 9, 1989 Landsat 5 WRS-2 56 20
GLS2000 August 12, 1999 Landsat 7 WRS-2 56 20
GLS2005 August 12, 2005 Landsat 7 WRS-2 56 20
GLS2010 July 30, 2009 Landsat 5 WRS-2 56 20

29

There are three primary reasons why GLS data was essential to this study:
1) multispectral capability;
2) temporal correlation;
3) consistent and predictable data quality.
Multispectral capability
The Landsats 5 & 7 multispectral range in the near-infrared (NIR) and shortwave-infrared
(SWIR) provides the best spectral differentiation and interpretability of ice and snow; in
particular, the use of SWIR (RGB: 4, 5, 7) false-color composite has proven to be very effective
in differentiating ice and snow (Haq et al., 2012). Figure 8 provides a graphical illustration of the
stark difference between snow and ice when this particular image composite is used.
Unfortunately, this composite could not be created for the GLS1975 data. GLS1975 dataset is
collected using Landsat 1 MSS sensors which only collected imagery in the visible green, visible
red, and near infrared regions of the EM spectrum (refer to Table 1).
30

Figure 8. LeConte Glacier in GLS2010; RGB: 4, 5, 7. In this image, snow is bright red and ice is dark
red; water is black; bare ground is cyan; and vegetation is shades of green. This band combination clearly
distinguishes between ice and snow.
Temporal correlation
The Landsat program began in the 1970s. Although each consecutive sensor includes
new capabilities, legacy capabilities are also retained. This means that while Landsat 7 ETM+
sensor has an improved panchromatic band (15m spatial resolution), the visible, near-infrared,
and shortwave-infrared bands retain the same collection parameters (spectral sensitivities and
ranges) as several previous Landsat sensors, like Landsat 5 TM. In addition to retaining the same
collection parameters, collection areas (both size and location) are mostly identical. This sensor
generational redundancy has resulted in a very large image library that can be used
interchangeably. Using this multi-decadal image library, allows for cost-effective change
detection of natural and man-made features; such as glaciers, deforestation, and urban sprawl.

LeConte Glacier
31

Data quality
The atmospheric conditions in the project study area, which is located in the Tongass, are
predominantly misty, rainy, and cloud covered (refer to Section 2.1). In 2009, the year that the
project area was imaged for GLS2010 dataset, twenty-eight images are collected by a
combination of Landsats 5 and 7. These images ranged in cloud cover percentages of 1 to 100.
Statistical analysis of these cloud cover values was summarized in Table 9.
Table 9. Cloud cover descriptive statistics for Landsat images collected during 2009. During 2009 (the
year that GLS 2010 data was collected), a total of 28 Landsat 5 TM and Landsat 7 ETM+ images are
collected. The least amount of cloud cover present in an image is 1%, the most cloud cover is 100%. The
mean cloud cover is 63.68%.
Descriptive
Statistics
Value (% Cloud
Cover)
Descriptive
Statistics
Value (% Cloud
Cover)
Mean (µ) 63.68 Standard Deviation 33.82
Median 70.00 -1 Deviation 29.86
Mode 87.00 +1 Deviation 97.50
Standard Error 6.39
Klibanoff et al. (2005) asserts that 68.27% of all values in a population are within one standard
deviation of the mean. This study required images with low cloud cover values and the analyzed
population 17 (67.67% of the population) is within one standard deviation; in statistical terms,
the negative outliers are the most desirable. Although the results of the statistical analysis favors
Kilbanoff et al. (2005), they do not favor remote sensing; cloudy images are usually useless for
most applications. Of the 28 images that were considered for 2009, there were six negative
outliers (21.43% of the population) with less than 29.86% cloud cover. The unfortunate reality
was that due to the infrequent cloud-free days in the Tongass, the study area was not conducive
to satellite imaging.
32

GLS data is proving very useful for studying a variety of phenomena. More specific to
this study, it has been used to previously describe the extent of Baird, Patterson, and LeConte
Glaciers. In 2006, Beedle (2013) used GLS2000 data to determine the extents of Patterson and
LeConte Glaciers. These extents (shapefiles) are currently included in the GLIMS database. As
various processes were run during the course of this study, the shapefiles of those glaciers was
used to verify results and determine validity.
3.3 Global Land Ice Measurement from Space (GLIMS) data
The Global Land Ice Measurements from Space (GLIMS) database and data access web
portal is administered by the National Snow and Ice Data Center (NSIDC) in Boulder, Colorado
(Raup et al. 2007). Due to the large variety of national and international projects currently
managed by NSIDC, leveraging the capabilities and resources of NSIDC adds professional
credibility to the GLIMS project (National Snow and Ice Data Center, 2013).
GLIMS data is stored as geographic shapefiles. The GLIMS web portal provides tools to
geographically search for data. This search uses industry standard area of interest (AOI) type
tools to select an area which is intersected with the GLIMS database to extract available data for
download in shapefile format. The GLIMS data acquisition is covered step-by-step in Appendix
B. In Figure 9 the GLIMS shapefiles for Baird Glacier, Patterson Glacier, and LeConte Glacier is
shown. While Shakes Glacier is not currently in the GLIMS database, it was shown on Figure 9
in relation to the other glaciers.
One advantage of using a geographically referenced shapefile format is that it displays
correctly with other georeferenced data, such as GLS data, in ArcGIS v10.2. Table 10
33

summarizes the GLIMS database entries for Baird Glacier, Patterson Glacier, and LeConte
Glaciers.

Figure 9. Central Southeast Alaska Glacier Study Project: Global Land Ice Measurements from Space
data graphic. In this graphic, the extent of Baird Glacier is shown in salmon, Patterson Glacier is shown in
light gray, and LeConte Glacier is shown in light pink. Shakes Glacier, shown in tan, does not currently
have an entry in the GLIMS database. It is shown only to provide spatial orientation.

34

Table 10. Global Land Ice Measurement from Space database entries summary for Central Southeast
Alaska Glacier Study Project area. Baird Glacier is the most recently analyzed glacier with imagery from
2005. Both LeConte and Patterson Glaciers were last analyzed in 1999 – almost 15 years ago.
Glacier Name
Date Last
Analyzed
Source Imagery
Collection Date
Source Imagery
for GLS Dataset
Baird Glacier January 1, 2007 August 13, 2005 GLS2005
LeConte Glacier April 6, 2006 August 12, 1999 GLS2000
Patterson Glacier April 10, 2006 August 12, 1999 GLS2000
Global Land Ice Measurements from Space (GLIMS) data was used for the imagery analysis
process check as well as a metric from which to quantify glacier movement within the project
study area.

35

CHAPTER FOUR: METHODOLOGY
Study area physical characterization: topology and land cover
Topology
Using ArcGIS version 10.2 and 1-arc second ASTER DEM mosaic a slope map was
derived and thematically classified by ranges of slope percentages. From zero to 100 percent, the
slope values were placed in 10 equal interval bins of 10 percent each; i.e. 0 to 10 percent was the
first bin, from greater than 10 to 20 percent was in the second bin, and so forth. Slope values
over 100 percent were grouped into a single bin. For reference, 100 percent slope is equivalent to
45 degrees slope. Note, by commonly accepted mathematical definition, a horizontal line has
zero slope; a vertical line has no slope. The results are shown as Figure 10.

36

Figure 10. Central Southeast Alaska Glacier Study Project slope graphic. Slope is categorized from least
to greatest; green areas have the least amount of slope and red areas have the greatest amount. The valleys
surrounding the four glaciers had 100% or greater slope; the total area encompassed for this classification
was 1,917.7km
2
.
The examination of the slope map outlined that approximately half (47.3%) of the project
study area had slope values of less than 30 percent and slopes of 10 percent or less were the most
common class. Table 11 summarizes the total number of pixels and the area that they represent
within the study area. It should be noted that glacier features in this study area usually had low
slope values, often less than 10 percent.
37

Table 11. Central Southeast Alaska Glacier Study Project percent slope computation summary. Areas
with 0% to 10% slope were the most common; areas >70% to 80% were the least common. Slopes values
of less than 30% accounted for almost half (47.2%) of the project study area.
Class
Number of
Pixels
Area of
Pixels (km
2
)
Percent of
Total Area
0% to 10% 736,989 436.9 22.8
>10% to 20% 464,734 275.5 14.4
>20% to 30% 326,040 193.3 10.1
>30% to 40% 287,611 170.5 8.9
>40% to 50% 284,523 168.7 8.8
>50% to 60% 266,046 157.7 8.2
>60% to 70% 222,245 131.8 6.9
>70% to 80% 171,726 101.8 5.3
>80% to 90% 125,778 74.6 3.9
>90% to 100% 91,199 54.1 2.8
>100% 257,699 152.8 8.0
Total 3,234,590 1,917.7 100.0
Land Cover
Using ArcGIS 10.2, the NLCD 2001 was clipped to the study area (as shown in Figure
11). Similar feature classes were combined into a generic feature class; e.g. deciduous,
evergreen, and mixed forest classes were combined into a single class: forest. In that case, image
simplification improved visual interpretability. The results are shown as Figure 11. Like the
slope map, feature class pixel counts, areas, and percentages of study area were determined.
Based upon these calculations, several conclusions were inferred from the land cover map.

38

Figure 11. Central Southeast Alaska Glacier Study Project land use classification graphic. Vegetated
areas dominated the western edge of the study area. The large non-vegetated area in the center and eastern
edge of the study area was the Stikine Icefield. The total area encompassed for this classification is
1,916.4km
2
.
Table 12 summarizes the land cover classification pixel analysis results. The most
prevalent class was Perennial Ice/Snow; which covered 852.0km
2
(44.46%) of the study area.
Vegetated land, a combination of forest, shrubs, and wetlands, accounted for 591.0km
2
, or 30.84
percent of the study area.

39

Table 12. Central Southeast Alaska Glacier Study Project land use area by class computation summary.
Areas with “Perennial Ice/Snow” were the most common; “Developed” areas were the least common.
“Not Classified” areas are those located in Canada. Since NLCD 2001 is a US dataset, Canada is not
covered. The total area encompassed for this classification is 1,916.4km
2
.
Class
Number of
Pixels
Area of Pixels
(km
2
)
Percent of
Total Area
Open Water 150,301 135.3 7.06
Perennial Ice/Snow 946,661 852.0 44.46
Developed 484 0.4 0.02
Forest 366,478 329.8 17.21
Shrub 282,598 254.3 13.27
Wetlands 7,627 6.9 0.36
Barren Land 230,586 207.5 10.83
Not Classified 144,562 130.1 6.79
Total 2,129,297 1,916.4 100
Quantify the movement of Baird, Patterson, LeConte, and Shakes Glaciers
Several shortcomings were noticed in reviewing available literature. Change detection,
especially the multi-decadal projects, attempted to compare imagery collected from different
months. For example, imagery collected in March was compared with imagery collected in
November. Change between the two images was concluded to be indicative of net glacier
movement. At first glance, this approach seemed correct. However, March imagery reflected a
glacier that has just come out of the coldest months of the year and has experienced its greatest
potential movement. Conversely, November imagery showed a glacier that came out of the
warmest months of the year and has exhibited the greatest potential for retreating by melting or
40

similar processes. In either situation, it was erroneous to compare glacier activity in March
imagery to November imagery and draw a definitive conclusion about whether a glacier had
advanced, remained the same, or retreated. The second discrepancy noticed was the creation of
ground-truth data from in-scene examination. This approach did not yield valid ground-truth
data. As the name implies, ground-truth means that field surveys were conducted or observations
recorded. Because, it was difficult to ascertain the imagery interpretation experience (e.g. size,
shape, shadow) of the analyst that produced the in scene ground-truth data, the validity of the
results should be accessed.
Mitigation of the first problem was addressed by selecting imagery with similar month
and day anniversary dates; for example, May 7, 2003 was compared to May 3, 2008. The Global
Land Survey (GLS) datasets contained imagery with similar collection anniversary dates; e.g. all
dataset imagery anniversary dates are within a month or two of each other. The potential
problems of in-scene ground-truth data were addressed by using multiple classified images and
false-color composite images to ensure the terminus of each glacier was accurately delineated. In
addition, GLIMS data for Baird, Patterson, and LeConte Glaciers was used as a benchmark from
which to measure glacier movements. Relying on multiple images with similar collection
anniversary dates to delineate glacier terminus locations and using the Global Land Ice
Measurement from Space (GLIMS) data as a benchmark made this glacier study project different
than previous studies.
The GLS 1975, 1990, 2000, 2005, and 2010 data was used to determine the location of
the terminuses of Baird, Patterson, LeConte, and Shakes Glaciers. The source imagery for these
five datasets is Landsat and was collected on: September 3, 1974; September 9, 1989; August 12,
41

1999; August 12, 2005; and July 30, 2009 respectively. After determining the glaciers’ terminus
locations, these locations, with the exception of Shakes Glacier, were compared to the Global
Land Ice Measurement from Space (GLIMS) database; Shakes Glacier was not listed in GLIMS.
In GLIMS, Baird Glacier was last analyzed on August 13, 2005 and Patterson and LeConte
Glaciers on August 12, 1999. Comparison of terminus locations in the GLS data to the GLIMS
allowed for the computation of movement rates. Since Shakes Glacier was not in the GLIMS
database, movement rates were calculated using only the GLS data. Completion of this objective
allowed objectives two and three to be completed.
Analyze the movement rates for glaciers that have similar terminal terrain conditions
The second goal was to determine if similar terrain conditions where glacier terminuses
are located induce similar glacial movement rates. During the study period (1974-2009), both
Patterson Glacier and Shakes Glaciers’ terminuses were located in freshwater lakes, Patterson
Lake and Shakes Lake respectively. Movement rates for Patterson Glacier and Shakes Glacier
were compared to determine if a significant difference in movement rates exists.
This assessment was a multi-step process. First, using Patterson and Shakes Glacier
terminus locations in the various GLS datasets, which were determined for objective one,
movement distances were measured between terminus locations. Second, after the movement
distances were calculated for Patterson and Shakes Glaciers, the distances were compared to
each other; e.g. the movement distance between the GLS2010 and GLS2005. Analysis of this
data was conducted to determine if the movement rates for Patterson and Shakes Glaciers were
similar.
42

Analyze the movement rates for glaciers with dissimilar terminal terrain conditions
The third goal was to determine if dissimilar terrain conditions of the glacier terminuses
locations affected the movement rates. Baird Glacier’s terminus was located on land, Patterson
Glacier and Shakes Glaciers terminuses were located in fresh water lakes, and LeConte Glacier’s
terminus was located in LeConte Bay, which is a marine bay. While both Patterson and Shakes
Glaciers had similar terminal terrain conditions, Patterson Glacier was selected for this analysis
because its more conservative movement rate was closer to the movement rates of Baird and
LeConte Glaciers. Shakes Glacier retreat was substantially greater than the other glaciers in this
study.
This assessment process was identical to assessing the movement rates of glaciers with
similar terminal terrain conditions. For this analysis, Baird Glacier’s movement rate was
compared to Patterson and LeConte Glaciers. Also, the movement rate for LeConte and
Patterson Glaciers was compared.
Visual inspection of the GLS imagery that was used for this study reveals the difficulty in
delineating the terminuses of Baird, Patterson, LeConte, and Shakes Glaciers. Because of this
challenge, manual digitization from the unprocessed GLS data was problematic. Instead,
unsupervised image classification methods offered the best possibility of accurately separating
glacier terminuses from their surroundings (Noderer, 2010). Based on the literature review,
personal experience, and a thorough exploration of the GLS data, the false-color SWIR
composite image (Baolin, Zhang, and Chenghu. 2004), false-color NIR composite image, natural
color composite image, Landsat NIR and SWIR image bands, and Iterative Self-Organizing
(ISO) Data Cluster Unsupervised Classification (Raup et al., 2007), were used to delineate the
43

glacier terminuses. This process is covered in detail in Section 4.1. After assembling the various
image composites, a heads-up digitization (Raup et al., 2007) of each glacier terminus was
performed; the digitized terminus was recorded as a georeferenced shapefile. This process is
covered in detail in Section 4.2. Figure 12 shows a schematic approach of the methodology.

Figure 12. Glacier analysis process diagram. Reading from left to right: research questions were
formalized; required data was identified, and acquired; color composites were assembled, ISO
unsupervised classifications were performed, temperature data was compiled, National Land Cover Data
(NLCD) reclassification was performed, and ASTER DEM data was mosaicked; glacier outlines,
terminuses, and centerlines were determined, ASTER DEM slope calculations and the physical
characteristics of study area were determined; slope, climate graphs, classification, and glacier movement
maps were produced; and the physical characteristics of the study area and glacier movements were
synthesized.
To quantify the movement of Baird, Patterson, LeConte, and Shakes Glaciers, a
measurement baseline was needed. A baseline provided a geographic reference point along
which all glacier movement was measured; whether a glacier move forward or backward and by
how much. From Table 10 (section 3.2) note that Baird Glacier used GLS2005 data and LeConte
Glacier and Patterson Glaciers both used GLS2000 data to create the most recent shapefile in
44

GLIMS. In this study, the original imagery GLS datasets, image composites, band images, image
classifications, and GLIMS shapefiles were used. The image processing that was performed on
the original GLS datasets was validated using the GLIMS shapefiles.
For each of the five GLS datasets, GLS1975, GLS1990, GLS2000, GLS2005, and
GLS2010, unsupervised image classification was used to determine the terminus of Baird,
Patterson, LeConte, and Shakes Glaciers. This process was often hit-or-miss and requires much
refinement before adequate results were achieved. Because of this difficulty, image classification
was performed using GLS2005 and GLS2000 data first. The GLIMS shapefiles for Baird,
Patterson, and LeConte Glaciers were then used to check the image classification results. Results
that appeared incorrect served as a basis for refining the image classification procedures. Once
all image classification results were deemed satisfactory, then the glacier terminuses were
delineated in each GLS dataset.
Completion of the image classification of all five GLS datasets resulted in a total of 20
shapefiles, four different glaciers at five points in time. The GLIMS shapefiles were then
compared to the terminus locations of Baird, Patterson, and LeConte Glaciers in the
GLS2000/2005 datasets. Based upon positional differences of each glacier’s terminus in the
various GLS datasets, movement rates were computed. The GLIMS dataset has been essential to
this study as it was used in the validation process of the image classification results and in
quantifying glacier movements over the last thirty-four years.
The Environmental Systems Research Institute (Esri) ArcGIS Desktop v10.2 software
(ArcGIS, 2013) was used for all data processing and map production aspects of this study.
45

ArcGIS v10.2 possesses the functionality to carry on the image classification, feature extraction,
feature digitization, and map production capabilities that were needed for this study.
4.1 Composite images and image processing
False-color composite image assembly
The best spectral separation of snow and ice in multispectral imagery was successfully
achieved by Haq (2012) using shortwave infrared (SWIR) spectral bands, in particular a false-
color composite of bands 4, 5, 7 (RGB) images. The same approach was used in this study and
false-color composite of band 4, 5, 7 (RGB) images were created for GLS2010, GLS2005,
GLS2000, and GLS1990 datasets. An example for Patterson Glacier in GLS2005 is shown in
Figure 13.

Figure 13. Landsat 7 ETM+ false-color shortwave composite image of Patterson Glacier (GLS 2005).
This image is a near infrared, shortwave, shortwave composite (RGB: 4, 5, 7). In this image, ice is dark
red, vegetation is shades of orange, water is black, and bare ground is shades of cyan/gray.
The GLS1975 data lacks shortwave-infrared capabilities, therefore a false-color
composite of bands 7, 5, 4 (RGB) was used instead (Noderer, 2007). In this instance, Band 7 is a
Patterson Glacier
46

near-infrared band and not a shortwave infrared band (United States Geological Survey, 2013).
An example for Patterson Glacier in GLS1975 is shown in Figure 14.

Figure 14. Landsat 1 MS false-color near infrared composite image of Patterson Glacier (GLS 1975).
This image is a near infrared, visible red, visible green composite (RGB: 7, 5, 4). In this image, ice is
cyan, vegetation is shades of red, water is dark blue/cyan, and bare ground is shades of gray.
The bands combination and false-color image assembly were processed in ArcGIS v10.2
using the Composite Bands tool, of the Raster Processing toolset. The tool and procedure is
shown in Figure 15, which illustrates the process for the GLS2010 dataset. For the GLS2005,
GLS2000, GLS1990, and GLS1975 datasets, the procedure was the same. The analysis of the
false-color composite images is presented in the results section.
Patterson Glacier
47

Figure 15. “Composite Bands" tool dialog window completed for GLS2010 dataset using Landsat 5 TM
bands 4, 5, 7 (RGB). 1: file location of the raster images; 2: list of the selected images. It is important to
note that the images were listed in the order that they will be combined to form a RGB image; 3: save
output image; 4: execute the process that creates the composite band image.
Iterative Self-Organizing (ISO) Data Cluster Unsupervised Classification
An Iterative Self-Organizing (ISO) Data Cluster Unsupervised Classification is a type of
unsupervised classification. It is unsupervised because the analyst does not interactively define
training samples. The result of an ISO unsupervised classification is an image that is classified
into classes with similar spectral properties; e.g. water, ice, etc. An additional benefit that was
realized for this project was that the separation between glacier terminuses and surrounding
features were better accentuated. The result of an ISO unsupervised classification is a gray scale
image where each shade represents a class of pixels with similar spectral properties (Noderer,
2007). The number of resulting classes is dependent upon the variability in the input image and
the number of desired output classes, which is defined a priory. This classification method is
superior to the minimum distance method since it considers pixels in multidimensional data
48

space where more input image bands results in a more complex data space (Campbell et al.,
2011). This means that as the number of input images increases, the classification results should
correspondingly improve.
An ISO unsupervised classification process produces a thematically classified image.
Pixels with similar brightness values are grouped together. It can be expected that since similar
features are spectrally similar, they will be grouped together. However, the process simply
groups pixels; it is up to an analyst to determine which features are represented by a particular
grouping, e.g. water, ice, forest, etc.
The fidelity, or pureness, of each output class depends upon three factors. First, sufficient
number of classes must be created; too few classes and each class will contain pixels that
represent multiple classes. Likewise, too many classes introduce class redundancy. Second, more
images should yield better results since each pixel is compared across n-dimensionality (where n
represents the number of images). Third, the degree of image feature homogeny affects the
classification results.
Each of the previously discussed three factors will have different levels of impact
depending upon each image scene. Feature homogeneity, the number of input images, and the
number of output classes vary considerably. For this analysis, between four and six input images
were used depending upon the GLS dataset. Initial output classes were set for each dataset
ranging between 20 and 40. Often, more classes produced worse results and the process had to be
fine-tuned to find the optimal number of classes. Output classes were combined to produce the
final output classes: ice, water, rock, sediment, and vegetation. Figure 16 shows a final
classification for LeConte Glacier using GLS2010 data.
49

Figure 16. ISO Cluster Unsupervised Classification for LeConte Glacier (GLS2010). In this
classification, ice is light blue, rock is gray, vegetation is green, and water is dark blue.
The ISO unsupervised classification process was performed in ArcGIS v10.2 using the
ISO Data Cluster Unsupervised Classification tool, which is part of the Classification toolset
(ArcGIS, 2012). The tool and procedure is shown in Figure 17, which illustrates the process that
was used for the GLS2000 dataset. For the GLS2010, GLS2005, GLS1990, and GLS1975
datasets, the procedure was basically the same. Differences between GLS datasets and glaciers
were reflected in the number of classes that were specified; which ranged between 20 and 40.
LeConte Glacier
50

Figure 17. "ISO Data Cluster Unsupervised Classification" tool dialog window completed for GLS2000
dataset using Landsat 7 ETM+ bands 1-5 & 7. 1: file location of the raster images; 2: list of selected
images; 3: number of classes for the image pixels to be placed in; 4: save output image; 5: minimum class
size (number of pixels); 6: sample interval; 7: execute the process that creates and ISO Data Cluster
Unsupervised Classified image.
4.2 Glacier terminus delineation
Glacier terminus location shapefile creation
The terminuses of the Baird, Patterson, LeConte, and Shakes Glaciers were delineated
using the GLS dataset images (either individual images and/or image composites) and results
from ISO unsupervised classifications. Visual inspection of each image clearly identified the
general location of each glacier terminus. While this might have been suitable for some
applications, it does not lend itself to accurate measurements. In order to determine the glacier
terminuses with the available imagery, pixels were individually verified (by examining pixel
brightness values) whether it was ice or something else. In the case of Baird Glacier in GLS2010,
Band 4 (NIR), true color composite image, false-color SWIR composite image, and the results
51

from the ISO unsupervised classification were used to delineate the glacier’s terminus. The
resulting terminus for Baird Glacier in GLS2010 is shown in Figures 21 and 22. The images that
were used for each glacier are collected in Appendix C.
The creation of a shapefile was a two-step process. First, a drawing was created that
represented a glacier terminus. Pixels that were confirmed to be ice became vertices that were
connected with a line feature which was subsequently stored as a geographically referenced
shapefile. Second, the drawing was converted to a geographically referenced shapefile. From a
software functionality standpoint, this approach was more efficient for creating many shapefiles
which only contain a single feature with minimal attribution. For this study, each glacier
terminus; e.g. Baird Glacier in GLS2010, was stored as a single feature shapefile and
differentiation was accomplished with unique shapefile file names. ArcGIS v10.2 was used to
create the shapefiles whose process is summarized in Figures 18-20.

Figure 18. Draw toolbar explained. 1: allows the user to select the type of feature that they wish to draw;
2: is the “Line” tool. Because the glacier terminuses are linear features, the line tool is the best choice.

52

Figure 19. Draw toolbar continued. 3: contains additional drawing options; 4: allows the user to convert
drawn graphics to features (shapefiles).

Figure 20. Convert drawn graphics to features. 5: allows the user to specify the type of features that will
be contained in the output shapefile; lines, points, or polygons. Since line graphics were drawn, the output
features type defaults to line graphics; 6: allows ensures that the output shapefile will possess the same
coordinate system as the data upon which the graphic was drawn; 7: allows the user to specify an output
image and save location; 8: is checked so that the graphics will be deleted after the shapefile is created; 9:
begins the conversion process. After the graphics have been converted to shapefile, they will be loaded
into the map document (user is prompted to whether or not they wish this to happen).
53

After the individual glacier terminus shapefiles were created (a total of 20 shapefiles),
each shapefile was then attributed to include “Glacier Name” and “GLS Dataset”. This was
necessary so that the glacier terminuses could be differentiated in the final glacier shapefiles.
Upon completion of the shapefile attribution, the shapefiles were combined to produce a final
shapefile for each glacier that contains the glacier’s terminus locations in the GLS2010,
GLS2005, GLS2000, GLS1990, and GLS1975 datasets (a total of four final shapefiles; one for
each glacier).

Figure 21. Baird Glacier in GLS2010: the left graphic is Landsat 7 ETM+ near infrared (Band 4) image
and the right image is a natural color composite (RGB Bands 3, 2, 1). In both images, the glacier terminus
was shown as a yellow line. The stair-stepped appearance of the terminus was due to the image’s 30m
pixels.

Near Infrared (Band 4) True Color Comp. (RGB: 3, 2,
1)
Baird Glacier
(2010)
Baird Glacier (2010)
54

Figure 22. Baird Glacier in GLS 2010: the left graphic is Landsat 7 ETM+ shortwave infrared composite
(RGB Bands 4, 5, 7) and the right image is an ISO Data Cluster Unsupervised Classification (Bands 1, 2,
3, 4, 5, & 7). In both images, the glacier terminus was shown as a yellow line. The stair-stepped
appearance of the terminus was due to the image’s 30m pixels.
Glacier movement quantification
Glacier terminus location shapefiles that were created from analyzing Global Land
Survey (GLS) datasets were measured against the Global Land Ice Measurements from Space
(GLIMS) shapefiles. The positions of Baird, Patterson, and LeConte Glaciers in the GLIMS
shapefiles was used as a starting point, or zero position, from which each glacier’s terminus
position in the various GLS datasets was measured. These measurements indicated the
movement directions and distances moved. Shakes Glacier was considered differently since it
lacks an entry in the GLIMS database. For Shakes Glacier, its movement direction and distance
was compared among the various GLS datasets. In addition, distances were measured for each
glacier between its terminus positions in the various GLS datasets; e.g. GLS1975 to GLS1990,
GLS1990 to GLS2000, and so forth.
Shortwave Comp. (RGB: 4, 5, 7) ISO Unsupervised Classification
Baird Glacier (2010) Baird Glacier (2010)
55

In order to measure glacier movement over the length of the study period, approximately
1975 to 2010, a measurement method was needed. After much thought and experimentation, the
following method was developed to measure glacier movement. The four-step method accounted
for the shape of each glacier and the fluid nature of its movement. An explanation of each step is
provided and an illustration is captured in Figures 23-26.
The first step was to determine the centerline of each glacier. Initially, this step was the
most difficult to execute using the tools in ArcGIS v10.2. A line shapefile for each glacier was
created where the edges of each glacier is annotated. Starting from a point several kilometers
beyond the farthest advance of each glacier, the two edges of each glacier were digitized
“upstream” for several kilometers beyond the most retreated glacier terminus position. Figure 23
shows the results of this process for Baird Glacier in GLS2010. Next, a multi-ring buffer process
was run against each shapefile. Depending upon the distance between the edges of each glacier,
the buffer distances could be considerable (1,500m-2,500m). At the center point between the
sides of each glacier, the buffers would intersect, as shown in Figure 24. Finally, by connecting
the “dots” created by the intersecting buffers, the centerline for each glacier was determined; as
shown in Figure 25. The process was repeated for Patterson, LeConte, and Shakes Glaciers.

56

Figure 23. Glacier edge deliniation for Baird Glacier (GLS2010). The glacier’s edges are shown in pink
and the terminus in yellow. The glacier edges were digitized from a point several kilometers
“downstream” of the terminus to a point several kilometers “upstream of the terminus. This provided an
accurate understanding of the fluid nature of the glacier so that an accurate glacier centerline could be
determined.

Figure 24. Glacier valley buffers for Baird Glacier (GLS2010). The glacier’s edges are shown in purple
and the buffers in black. The left image shows the buffers for all of Baird Glacier; while the image on the
right shows an enlargement so that the buffers may be better seen.

Baird Glacier
Baird Glacier Baird Glacier
57

Figure 25. Glacier centerline is completed for Baird Glacier (GLS2010). The glacier’s edges are shown
in purple, the centerline is pink, the terminus is yellow, and buffer rings are black. The left image shows
the centerline for all of Baird Glacier; while the image on the right shows an enlargement so that the
centerline can be better seen.
The second step established a measuring point on each glacier centerline from which
quantitative glacier movement analysis could be performed. In the previous step, a glacier
centerline was established for Baird Glacier in GLS2010; which is labeled “1” on Figure 26.
That centerline was critical to this step. Perpendicular lines were drawn to the glacial valley
center line to identify the point on a glacier’s terminus shapefile that was the most advanced
point of tangency; which is labeled “2” in Figure 26. The perpendicular, intersecting both the
point of tangency and the glacial valley center line, is labeled “3” on Figure 26; the point where
the perpendicular intersects the glacier’s centerline is labeled “4”, which is the point of
measurement for the terminus movement. For Baird, Patterson, and LeConte Glaciers there were
a total of six perpendiculars identifying five terminus positions, one for each of the five GLS
datasets, and one for the terminus position in the GLIMS dataset. For Shakes Glacier, there were
no GLIMS dataset; so there were only five perpendiculars that correspond to the terminus
Baird Glacier Baird Glacier
58

positions in the GLS datasets. Figures 28-31 illustrated the completed glacier valley center lines
and perpendiculars for Baird, Patterson, LeConte, and Shakes Glaciers.

Figure 26. Glacier centerline perpendicular is completed for Baird Glacier (GLS2010 ISO Classification).
The glacier’s centerline is shown in purple, the perpendicular is black, and the terminus is yellow. The left
image is an overview of the centerline perpendicular; while the image on the right shows an enlargement
and detail of the glaciers’ terminus measurement point (4).
With the origin points determined along each glacier centerline, movement and direction
was then determined. Movement distance measurement was accomplished by simply conforming
to the shape of each centerline and measuring between origin points. This method ensured that
all measurements for Baird, Patterson, LeConte, and Shakes Glaciers were uniform and no bias
was introduced favoring one glacier over another. Direction of movement (advance or retreat)
was determined by comparing a glacier terminus position in one GLS dataset to its position in
another GLS dataset. A partial example for Baird Glacier is shown in Figure 27.

Baird Glacier Baird Glacier
1
2
3
4
59

Baird Glacier

Figure 27. Baird Glacier movement measurement (partial). The glacier centerline is shown in red, the
perpendiculars are magenta, GLS2010 terminus is yellow, and GLS1990 terminus is brown. The distance,
as measured along the glacier centerline, between GLS1975 origin and GLS2010 origin is 356m. The
image scale is 1:25,000.
Baird Glacier

Figure 28. Baird Glacier terminuses and perpendiculars for GLS2010, 2005, 2000, 1990, and 1975
datasets. The image on the left is an overview and the image on the right is an enlargement. Terminuses
are shown as solid lines and perpindiculars are dashed lines.

60

Patterson Glacier

Figure 29. Patterson Glacier terminuses and perpendiculars for GLS2010, 2005, 2000, 1990, and 1975
datasets. The image on the left is an overview and the image on the right is an enlargement. Terminuses
are shown as solid lines and perpindiculars are dashed lines.
LeConte Glacier

Figure 30. LeConte Glacier terminuses and perpendiculars for GLS2010, 2005, 2000, 1990, and 1975
datasets. The image on the left is an overview and the image on the right is an enlargement. Terminuses
are shown as solid lines and perpindiculars are dashed lines.
61

Shakes Glacier

Figure 31. Shakes Glacier terminuses and perpendiculars for GLS2010, 2005, 2000, 1990, and 1975
datasets. The image on the left is an overview and the image on the right is an enlargement. Terminuses
are shown as solid lines and perpindiculars are dashed lines.
Comparison of glacier movement for glaciers with similar terminus conditions
This analysis was conducted by comparing the determined movement rates for the Shakes
Glacier and Patterson Glaciers, both of these glaciers’ terminuses were in freshwater lakes. The
results of this analysis are presented in Chapter 5.
Comparison of glacier movement for glaciers with dissimilar terminus conditions
This analysis was conducted by comparing the determined movement rates for the Baird,
Patterson, and LeConte Glaciers. Baird Glacier’s terminus was on dry land, Patterson Glacier’s
terminus was in fresh water (lake), and LeConte Glacier’s terminus was in salt water (marine
bay). The results of this analysis are presented in Chapter 5.
62

CHAPTER FIVE: RESULTS
The final combined shapefile terminus results for Baird, Patterson, LeConte, and Shakes Glaciers
are shown in Figure 32. The glacier terminuses were “stair stepped” in appearance due to the
usage of 30m spatial resolution Landsat imagery. No feature smoothing algorithm was applied to
the terminus shapefiles.
The movement values and movement direction for Baird, Patterson, and LeConte
Glaciers using the GLIMS dataset as a baseline are summarized in Table 13, while the movement
distances for Baird, Patterson, LeConte, and Shakes Glaciers for the periods of time between
GLS dataset collection events are summarized in Table 14 and Figure 33. Table 15 provides the
average glacier movement for the periods of time between GLS dataset collection events as well
as providing an average glacier movement per year for the entire breadth of the GLS datasets;
which is approximately 35 years (1974-2009). The values in Table 15 were determined by
dividing the movement values in Table 14 by the length of time that passed between the GLS
dataset collection dates. For GLS1975 to GLS1990, the length of time is 15 years; for GLS1990
to GLS2000, the length of time is 10 years; for GLS2000 to GLS2005, the length of time is 6
years; and for GLS2005 to GLS2010 the length of time is 4 years. Refer to Table 10 for the exact
collection dates for the GLS datasets.
63

Baird Glacier Patterson Glacier

LeConte Glacier Shakes Glacier

Figure 32. Glacier terminus results for the central southeast Alaska glacier GLS datasets. The top left
image is Baird Glacier, the top right image is Patterson Glacier, the bottom left image is LeConte Glacier,
and the bottom right image is Shakes Glacier. The background images are the ISO Cluster Unsupervised
Classification results for the GLIMS 2010 dataset.

64

Table 13. Baird, Patterson, and LeConte Glacier terminus distance summary during the various time
periods between GLS dataset collection events. Using the GLIMS perpendicular’s intersection with the
centerline as a starting point, distances to the intersection of each dataset’s perpendicular and the
centerline were measured. A positive value indicates that a glacier’s terminus is in front (downstream) of
the GLIMS data; a negative value indicates that a glacier terminus is behind (upstream) of the GLIMS
data.
GLS Dataset Glacier Position from GLIMS Baseline (m)
Baird Patterson LeConte
GLS1975 -242 1460 1409
GLS1990 -141 932 1242
GLS2000 -168 164 295
GLS2005 -267 -567 136
GLS2010 -597 -645 380

Table 14. Movement distance summary for each glacier from one GLS dataset collection event to the
next e.g. (GLS1975 to GLS1990, GLS1990 to GLS2000, and so forth). A total amount of movement over
the 35 years of data coverage is also provided at the bottom of the table. Distances were measured in
meters. From approximately 1975 to 2010, Shakes Glacier experienced the greatest total movement
amount; Baird Glacier experienced the least.
GLS Dataset Temporal
Range
Glacier Movement per GLS Datasets (m)
Baird Patterson LeConte Shakes
At GLS1975 0 0 0 0
GLS1975 to GLS1990 101 528 167 1942
GLS1990 to GLS2000 27 768 947 685
GLS2000 to GLS2005 99 731 159 415
GLS2005 to GLS2010 330 78 244 481
Total 557 2105 1517 3523

65

Table 15. Summary of the average movement rates per year for Baird, Patterson, LeConte, and Shakes
Glaciers in each time period between GLS dataset collection events and average movement for entire
period covered by GLS1975 to GLS2010 datasets. Distances were measured in meters. From
approximately 1975 to 2010, Shakes Glacier experienced the greatest average movement; Baird Glacier
experienced the least.
GLS Dataset Temporal
Range
Average Glacier Movement per Year (m)
Baird Patterson LeConte Shakes
At GLS1975 0 0 0 0
GLS1975 to GLS1990 7 35 11 129
GLS1990 to GLS2000 3 77 95 69
GLS2000 to GLS2005 17 122 27 69
GLS2005 to GLS2010 83 20 61 120
Glacier Movement per
Year for GLS1975 to
GLS2010
16 60 43 101
66

Figure 33. A summary of the movement distances for Baird, Patterson, LeConte, and Shakes Glaciers
during the periods of time covered by each GLS dataset. The total movement distances for each glacier in
each GLS dataset are shown as solid lines. The average movement distance per year covered by each
dataset is shown as dotted lines.
5.1 Glacier movement qualification
Patterson and LeConte Glaciers steadily retreated during the time spanned by the
GLS1975 to GLS2005 datasets. However, in the GLS2010 datasets, Patterson Glacier retreated
and LeConte Glacier advanced from previous positions. One potential conclusion for the retreat
during 1975 to 2005 is that these glaciers attempted to equalize to changing climate conditions
and began retreating, as evidenced by the steady retreat exhibited in the GLS1975 to GLS2005
time period. However, they contracted too quickly and had begun to surge forward in an attempt
to equalize with the environment. In the GLS2010 time period, LeConte Glacier had advanced,
but Patterson Glacier continued to retreat. This oscillation movement is very similar to a weight
67

hanging at the end of a rubber band that is released and moves up and down until equilibrium
between the rubber band’s elasticity and the pull of gravity can be achieved. In this example,
equilibrium is achieved when the glaciers melt rate equaled it growth rate.
There are many variables that must be considered to understand why a particular glacier
flows in the manner that it exhibits. Waddington (2009) determined that glaciers flow because of
two simplified principles: 1) ice deformation and 2) glacier substrate allows glacier to move over
it. Waddington (2009) further explains that ice flow speed is determined by four factors: 1) ice
thickness, 2) slope, 3) ice properties, and 4) bed properties. In addition, any moveable object will
tend to move from an area of higher elevation to an area of lower elevation, if possible. This is
typified by water moving from an area of higher elevation to an area of lower elevation, e.g.
water normally runs downhill.
Glacier slope
In the case of glaciers, the high point from which they flow is a much larger collection
area, like an icefield or ice cap. For Baird, Patterson, LeConte, and Shakes Glaciers, that icefield
is the Stikine Icefield (Molina, 2008). To illustrate the immutable tendency of glaciers to flow
from areas of higher elevation to areas of lower elevation, glacier surface elevations were
determined for Baird, Patterson, LeConte, and Shakes Glaciers at the terminuses and also 5km
“upstream” from the terminus. The resulting elevation values are listed in Table 16. As discussed
in Section 4, the slope of the study area ranged from 0% to greater than 100%. While the slope
values in the study area varied considerably, the slopes of the glaciers were comparatively slight;
ranging from 3.02% to 9.20%. The glacier slopes and average movement rates are summarized
in Table 15 and shown in Figure 34.
68

Table 16. Summary of glacier valley slopes for Baird, Patterson, LeConte, and Shakes Glaciers.
Elevations were determined at terminuses and five-kilometers upstream from each terminus. Values were
recorded in meters and were measured on the surface of the glacier. While this was not ideal, the lack of
glacier bed elevation data at the point five-kilometers upstream necessitated surface measurements. In
order to remain consistent, terminus and 5km upstream elevation values were measured on glacier
surfaces. Technology, like ground penetrating radar, can provide accurate ice thicknesses which can be
used to calculate glacial bedrock profiles and elevations (Geophysical Survey Systems, Inc., 2012).
Glacier
Elevation at
Terminus (m)
Elevation 5km
Upstream (m)
Slope
(%)
Average Movement
Rates Per Year (m)
Baird 27 178 3.02 16
Patterson 49 424 7.50 60
LeConte 46 506 9.20 43
Shakes 11 201 3.80 101

Figure 34. Slope at and within five-kilometers of Baird, Patterson, LeConte, and Shakes Glaciers. From 0
to 100 percent, slope values are divided into 10 bins; slopes greater than 100 percent are placed in a single
bin. Glacier valley slopes were typically less than 10% and the surrounding terrain is much steeper.
69

Looking father “upstream” to the Stikine Icefield, which is the source of Baird, Patterson,
LeConte, and Shakes Glaciers, elevations commonly exceeded 1,500m. Logic would indicate
that glaciers with the greatest slope values should experience the greatest amount of forward
movement (due to gravity assisting movement) and the least amount of backwards movement
(also due to gravity retarding movement). Comparing the slope values to the average movement
rates per year for GLS1975 to GLS2010 (Table 15) is contrary to this belief.
Baird Glacier had the least slope (3.02%) and the lowest average movement rate per year
over the span of the GLS datasets (16m). Shakes Glacier also had low slope (3.80%), but
experienced the greatest average movement rate per year over the span of the GLS datasets
(101m). Patterson Glacier had greater slope (7.50%) and experienced the second greatest average
movement rate per year over the span of the GLS datasets (60m). LeConte Glacier had the
greatest slope (9.20%), but had experienced low average movement rate per year over the span
of the GLS datasets (43m). The relationship between slope and glacier movement rates were
typified in Baird and Patterson Glaciers, but LeConte and Shakes Glaciers behaved differently.
An appropriate conclusion is that other factors affect glacier movement rates, not just
slope. Unfortunately, this data did not indicate what those factors may be, or the effect that they
had on glacier movement rates for Baird, Patterson, LeConte, or Shakes Glaciers. Further study
is warranted to identify these additional factors and the role they play on glacier movement rates.
Glacier bed properties
In addition to slope, Waddington (2009) attributed glacier flow speed to glacial bed
properties. Analysis of land cover in the study area showed the areas at and near Baird,
70

Patterson, LeConte, and Shakes Glaciers are characterized by bare rock voided of any trees or
vegetation. The lack of any vegetative cover reduced any impedance to the Glaciers’ movement.
Additionally, Glaciers in the study area had a historical record of advance-retreat-advance;
LeConte Glacier has advanced six times since 1,600 years before present (Molina, 2008). Each
time that this had occurred, the glacial bed is progressively polished and the friction between ice
and rock is incrementally reduced (Michna, 2012).
Temperature dependence of ice flow
Waddington (2009) also states that there is a direct relationship between glacier flow
speed and temperature; the warmer it is, the faster a glacier can flow. Figure 35 (Waddington,
2009), identifies the relationship between glacier flow speed and temperature. The temperature
values are given in Kelvin (K); for reference water freezes at 273.15K and boils at 373.15K.
As temperatures increase, glacier deformation increases as does glacier flow speed (Waddington,
2009). Analyzing National Oceanographic and Atmospheric Administration (NOAA)
temperature data that was collected at the Petersburg, Alaska meteorological station from
January 1, 1973 to December 31, 2009 revealed a general warming trend. Petersburg, Alaska was
the closest meteorological data collection point to the study area. It is understood that
atmospheric conditions in Petersburg, Alaska, were only close approximations for the conditions
near Baird, Patterson, LeConte, and Shakes Glaciers; i.e. large masses of ice have a noticeable
effect on temperature and precipitation. As the study area is located in the Tongass, precipitation
was significant (refer to Section 2.1). From January 1, 1973 to December 31, 2009, mean yearly
temperatures have been increasing as shown by the positive trend line in Figure 36.
71

Figure 35. Relationship between ice flow rates and temperatures. As the temperature increases towards
the melting point (which occurs at 273.15K), the ice deformation rate increases. Ice deformation is the
ability of ice to change shape without breaking (Waddington, 2009).

Figure 36. Mean yearly temperatures chart for 1973-2009. Years that were not represented by a data
point were those years for which no data was available. The data exhibited a positive trend; suggesting
that the average yearly temperature was increasing. The trend line equation is: y = 0.0223x + 278.22.
260
265
270
275
280
285
290
Mean Yearly Temperature
Mean Yearly Temperature (K) Mean Temperature Trend
Mean Yearly Temperature (K) Mean Temperature Trend Freezing Point (273.15K)
Temperature (K)
72

The increase in mean yearly temperatures can be further analyzed by examining the
average monthly temperature for the same period of time. It was discovered that warming trends
were experienced in the months of January, May through September, and November through
December. Conversely, cooling trends were experienced in February through April, and October.
Eight out of twelve months were getting warmer, while the other four months were getting
colder. These trends are summarized in Table 17. It could be inferred from this data that the
summers and several traditionally winter months were getting warmer. This general warming
trend is detrimental to ice formation and should be leading to greater glacier melting. With the
exception of Baird Glacier in the GLS1975 to GLS1990 and LeConte Glacier in GLS2005 to
GLS2010 datasets, Baird, Patterson, LeConte, and Shakes Glaciers have retreated between 1975
and 2010.
Table 17. Temperature trends for average monthly temperatures from 1973 to 2009. A positive trend
indicates that the particular month had gotten progressively warmer from 1973-2009; a negative trend
indicates that the month had gotten colder for the same period of time. From 1973-2009, May through
September and November through January have showed a positive trend; while February through April
and also October have exhibited a negative trend.
Month Season Temperature Trend Month Season Temperature Trend
March Spring Negative September Fall Positive
April Spring Negative October Fall Negative
May Spring Positive November Fall Positive
June Summer Positive December Winter Positive
July Summer Positive January Winter Positive
August Summer Positive February Winter Negative

73

In this study, it was originally suspected that slope would be a key factor in glacier
movement, e.g. greater slope would yield greater glacier movement rates. However, the results
fail to conclusively support this assumption. While it has been well established that glaciers are
indicators of climate change, both cooling and warming, further investigation reveals that
glaciers are often slow to respond to climate change and many factors influence a particular
glaciers susceptibility to climate change. To further confuse the subject, each glacier’s tolerance
to the degree of climate change varies, for example one glacier may require a hypothetical net
mass balance change of only 1°C, while another glacier requires a hypothetical net mass balance
change of 3.5°C; Davies (2014) was able to force a glacier mass balance change with an increase
of only 0.5°C.
Terminus conditions and response to climate change
Referring to Table 18, Baird Glacier’s slope is 3.02%, its average yearly movement rate
was 16m, and its terminus ended on land; Patterson Glacier’s slope is 7.50%, its average yearly
movement rate is 60m, and its terminus ended in a fresh water (lake); LeConte Glacier’s slope is
9.20%, its average yearly movement rate is 43m, and its terminus ended in salt water (marine
bay); and Shakes Glacier’s slope is 3.80%, its average yearly movement rate is 101m, and its
terminus ended in a fresh water (lake). If steeper slopes are expected to produce greater
movement rates, the results fail to conclusively support this premise. This is especially true for
the Shakes Glacier which has slight slope, but the greatest average yearly movement rate.
Because of the lack of conformity to this expectation, other factors besides slope must be
examined. For example, is there a time delay between glacial mass accumulation and ablation; if
so, is that delay different for glaciers with different terminus conditions (land, freshwater, or
74

saltwater)? The terminus conditions of Baird, Patterson, LeConte and Shakes Glaciers are shown
in Figure 37 and summary of conditions and rate of movement in Table 18.

Figure 37. The terminus conditions of Baird, Patterson, LeConte and Shakes Glaciers. Baird Glacier’s
terminus ends on dry land, LeConte Glacier’s terminus is located in salt water, and Patterson and Shakes
Glaciers’ terminus are located in fresh water lakes.
ISO Unsupervised Classification
Baird Glacier (2010)
ISO Unsupervised Classification
LeConte Glacier (2010)
ISO Unsupervised Classification
Patterson Glacier (2010)
ISO Unsupervised Classification
Shakes Glacier (2010)
Sal t
W a t er
Fr esh
W a t er
Fr esh
W a t er
Land
75

Table 18. Summary of glacier valley slopes for Baird, Patterson, LeConte, and Shakes Glaciers.
Glacier Terminus Ends In/On Slope (%) Average Movement Rates Per Year (m)
Baird Land 3.02 16
LeConte Salt Water (Marine Bay) 9.20 43
Patterson Fresh Water (Lake) 7.50 60
Shakes Fresh Water (Lake) 3.80 101

The delay between a significant change in climate and glacier terminus response to that
change is referred to as the lag time (T
s
) (Pelto & Hedlund, 2001). Lag time can be thought of as
the time it takes for a glacier to realize that a significant change in climate has occurred and
begin to mostly respond to that change (60% conformity), either by retreating or advancing. Lag
time is specific to each glacier and depends upon the glacier’s mass balance history and other
physical characteristics (Pelto et al., 2001). The determination of a glacier’s reaction time
requires a study over a lengthy period of time, during which the size and shape of the glacier are
measured consistently. Once a glacier recognizes that a significant climate change had occurred,
the time it takes for the glacier to approach a new steady state is referred to as response time (T
m
)
(Pelto et al., 2001).
A thorough literature review revealed very little information regarding lag and response
times for the glaciers in this study as Baird, Patterson, LeConte, and Shakes glaciers are simply
too remote and inaccessible to warrant comprehensive studies. However, lag and response times
have been established for other glaciers and some similarities between these glaciers and those in
this study can be expected. Harper (1993) suggests that glaciers on Mt. Baker, Washington, have
a response time of approximately 20 years. Also, in a study of alpine glaciers in the North
76

Cascade Mountains, it was observed that the glaciers followed three general movement trends
which are summarized in Table 19 (Pelto et al., 2001).
Table 19. Summary of glaciers in the North Cascades glacier study project. General characteristics,
movement trends, lag time and response times are provided. Data summarized from Pelto et al., 2001.
Type General
Characteristics
1890 to 1950 1950 to 1976 1976 to Present Lag
Time (T
s
)
Response
Time (T
m
)
I Steeper Slopes,
Higher Terminus
Velocities
Continuously
Retreated
Advanced Retreated 20-30
Years
4-16
Years
II Intermediate
Slopes,
Intermediate
Terminus Velocities
Rapidly
Retreated
Slowly
Retreated/Unchanged
Rapidly
Retreated
40-60
Years
4-16
Years
III Low Slopes, Low
Terminus Velocities
Retreated Retreated Retreated 60-100
Years
4-16
Years

This project differs from the North Cascades glacier study in three ways: (1) the study
time period is comparatively short (1975-2010); (2) neither Baird, Patterson, LeConte, nor
Shakes Glaciers are alpine glaciers; (3) the climate in the study area is markedly different than
the climate in the northern Cascades. However, while many differences exist between the
glaciers studied in this project and the glaciers studied in the Northern Cascades projects, all
glaciers, regardless of location, have a specific lag time and response time to significant climate
changes (Pelto et al., 2001). Additionally, Meier and Post (1986) concluded in a study of the
Columbia Glacier, a tidewater glacier located in south central Alaska, that grounded tidewater
glaciers rates of flow are independent of climate variables. Rather, flow rates are determined by
glacier and glacial valley physical characteristics. Meier et al., 1986 believe that assigning
climatic significance to grounded tidewater glacier movement is suspect.
77

Meier further refines his conclusions in a later study where Bahr, Pfeffer, Sassolas, and
Meier (1998) summarize the traditional definition of glacier response time as a ratio of ice
thickness to mass balance rate. Glacial mass balance rate is calculated by comparing
measurements from glacial accumulation areas with measurement from glacial ablation areas
(Krenke & Menshutin, 1987). When mass balance rate is taken into account with ice thickness,
ablation, mass balance gradients, hypsometry, and ice surface slope, glacial sensitivity to climate
change can be approximated (Davies, 2014).
Davies (2014) ascertains that maritime glaciers located in temperate climates, like
LeConte Glacier in Central Panhandle climate region (Figure 7), have a response time of 15-60
years. While no definitive response time is available for Baird, Patterson, or Shakes Glaciers, it
is available for Mendenhall Glacier, which is approximately 180km north of the project study
area. Mendenhall Glacier is similar to Patterson and Shakes Glaciers in that its terminus is
located in a freshwater lake. The response time of Mendenhall Glacier is currently estimated at
45 years (Motyka, O’Neel, Connor, & Echelmeyer, 2003).
It remains to be seen if the glaciers in the study area will respond to the climatic change
that has occurred over the length of the study period. Examination of the mean yearly
temperature, as shown in Figure 36, reveals the climate had warmed approximately 0.85°C from
1973 to 2009. This warming trend may be insufficient to force glacial responses. Additional data
is needed to determine glacial mass balance rate of the glaciers in the project study area as well
as relevant glacial physical characteristics. When these values are determined, it may be possible
to determine response times for Baird, Patterson, and Shakes Glaciers. This data is impossible to
derive from the Landsat data used in this study and would require onsite study of these glaciers.
78

In addition to accessing glacial ice mass balance rates, other physical forces should be analyzed
to determine their effect on glacial movement rates. Referring to Table 18, LeConte, Patterson,
and Shakes Glaciers, with terminuses located in water, experienced the greatest average yearly
movement rates. Two hypothetical reasons for this occurrence are a heat sink effect and tidal
influences. Simply stated, ice melts faster in water than in air (Helmenstine, 2014). Helmenstine
(2014) further explains that molecules of water are much closer together than molecules of air
which allows for greater contact between water and ice. For any given volume, it must contain
more molecules of water than ice. Each molecule of water has an inherent amount of thermal
energy as exhibited by the molecules’ atomic motion. Thermal energy, or heat, always moves
from an area of higher concentration to an area of lower concentration in an attempt to achieve
equilibrium. Because there are more molecules of water than ice in any given volume, the
volume of water has more energy to “share” with the ice and as a result, the molecules of ice
begin to move faster and faster as they warm up and change from solid ice to liquid water
(Helmenstine, 2014). Because of this natural phenomenon, LeConte, Patterson, and Shakes
Glaciers, with terminuses in water, should melt faster (ablate) and exhibit greater amounts of
movement (retreat) than Baird Glacier, which has its terminus located on Land.
LeConte Glacier, with its terminus located in a marine bay, is subjected to tidal
influences. Tidal fluctuations in vicinity of LeConte Glacier range from 2m below mean sea
level to 7m above mean sea level (Bruland, 2014). Hypothetically, this large variability in tidal
levels should exert up and down forces on LeConte Glacier’s terminus. In addition, as the tide
moves in and out of LeConte Bay, tidal forces pull and push LeConte Glacier’s terminus. In a
study of a marine terminating glacier in Greenland, Walter, Box, Slawek, Brodsky, Howat, Ahn,
79

and Brown (2012) discovered that tidal forces do exhibit a great deal of force in both the
horizontal and vertical planes, which facilitates glacial ablation. Due to the extreme tidal range,
present at LeConte Glacier, calving at weak points along the terminus should contribute to
glacial movement.
5.2 Comparison of glaciers with similar terminal terrain conditions
During the period of time encompassed by this project, approximately 1975 to 2010,
Baird Glacier’s terminus is a moraine field on land, LeConte Glacier’s terminus is a marine bay,
and Patterson and Shakes Glaciers’ terminuses were freshwater lakes. Among these four glaciers,
three different terminal terrain conditions existed: dry land, marine, and fresh water. With a
diversity of terminus terrain conditions, only Shakes and Patterson glaciers were similar enough
for a valid comparison.
Shakes Glacier versus Patterson Glacier
Analysis of Table 15 values revealed significant differences in the movement rates for
Shakes and Patterson Glaciers for two of four GLS dataset periods. From 1975 to 1990, Shakes
Glacier’s average movement substantially exceeded Patterson Glacier. From 1990 to 2005,
Shakes Glacier’s average movement is less than Patterson Glacier. Then from 2005 to 2010,
Shakes Glacier’s average movement is significantly more than Patterson Glacier. Table 20 and
Figure 38 summarize these findings. For the entire period covered by all GLS datasets, GLS1975
to GLS2010, Shakes Glacier’s average movement rate is 1.68 times greater than Patterson
Glacier (101m / 60m = 1.6833333m ~ 1.68m). While Shakes Glacier’s average movement rate
exceeded Patterson Glacier in two of the four study periods, it is similar enough in the other two
study periods to raise the question of what other forces affect glacier movement rates. Referring
80

to Table 16, Patterson Glacier’s slope is 7.50% and its average yearly movement rate is 60m;
however, Shakes Glacier’s slope is 3.80% and its average yearly movement rate is 101m. During
the study period, Shakes Glacier had the greatest average movement rate and the least slope;
Patterson Glacier had the greatest slope and a significantly lower average movement rate. Based
upon slope, and glacier movement rates, a definitive pattern does not emerge and it cannot be
conclusively determined from this data if glacier terminuses that are located in freshwater lakes
within the project study area produce similar glacier movement rates.
Table 20. Movement summary for Shakes Glacier versus Patterson Glacier. Movement rate multipliers
greater than 1.00 indicate that Shakes Glacier’s movement rate is greater than Patterson Glacier. In
addition to analyzing each GLS dataset period, an overall movement rate multiplier was determined for
the entire period of time encompassed by all GLS datasets. That multiplier indicated that Shakes Glacier,
on average, moved 1.68 times farther than Patterson Glacier.
Shakes Glacier versus Patterson Glacier
GLS Dataset Temporal Range Average Movement Rate Multiplier
GLS1975 to GLS1990 3.69
GLS1990 to GLS2000 0.90
GLS2000 to GLS2005 0.57
GLS2005 to GLS2010 6.00
GLS1975 to GLS2010 1.68

81

Figure 38. This graph compares the movement of Shakes Glacier to the movement of Patterson Glacier.
Movement rate multipliers greater than one indicate that Shakes Glacier’s movement rate is greater than
Patterson Glacier. The greatest amount of movement is during the interval between GLS2005 and
GLS2010 datasets where Shakes Glacier’s movement is six times greater than Patterson Glacier’s
movement rate.
5.3 Comparison of glaciers with dissimilar terminal terrain conditions
The terminuses of Patterson and Shakes Glaciers were in freshwater lakes, LeConte
Glacier is in a marine bay, and Baird Glacier is on land. Of the four glaciers, Baird Glacier had
the most dissimilar terminal terrain conditions. Of the two glaciers that end in freshwater,
Patterson Glacier’s average movement rate is more similar to LeConte Glacier’s movement rate.
For this analysis, Baird Glacier is compared to Patterson and LeConte Glaciers; also, LeConte
Glacier is compared to Patterson Glacier.

0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
GLS1975 to
GLS1990
GLS1990 to
GLS2000
GLS2000 to
GLS2005
GLS2005 to
GLS2010
Average MovementRate Multiplier
GLS Dataset Temporal Range
Average Movement Rate Multiplier for Glaciers with Similar
Terminal Terrain Conditions
Shakes Glacier versus
Patterson Glacier
82

Baird Glacier versus Patterson Glacier
The comparison of Baird Glacier versus Patterson Glacier considered land and fresh
water terminal terrain conditions, respectively. Analysis of Table 15 values revealed that in three
of the four dataset periods, Baird Glacier’s average movement rate is less than Patterson Glacier;
from 1975 to 2005. Only from 2005 to 2010 did Baird Glacier’s average movement rate exceed
Patterson Glacier. For the entire period covered by all GLS datasets, GLS1975 to GLS2010,
Baird Glacier had an average movement rate of only 0.27 that of Patterson Glacier (16m / 60m =
0.266667m ~ 0.27m). Table 20 and Figure 38 summarize these findings. However, when slope is
considered, a different pattern emerged. Referring to Table 16, Baird Glacier’s slope is 3.02%
and its average yearly movement rate is 16m; conversely, Patterson Glacier’s slope is 7.50% and
its average yearly movement rate is 60m. Based only upon average yearly movement rates, it
might be concluded that glacier terminuses which end in fresh water are more conducive to
glacier movement (retreat or advance) than terminuses which were located on land. However,
both Baird and Patterson Glaciers typified the relationship between slope and movement rates.
When slope is considered, Patterson Glacier’s greater movement rate may have less to do with
glacier terminus terrain types and more to do with increased slope (7.50% versus 3.02%). More
variables would need to be considered before a definitive conclusion can be reached.
Baird Glacier versus LeConte Glacier
The comparison of Baird Glacier versus LeConte Glacier considered land and marine
terminal terrain conditions, respectively. Analysis of the Table 15 values revealed that in three of
the four dataset periods, Baird Glacier’s average movement rate is less than LeConte Glacier;
from 1975 to 2005. Baird Glacier’s average movement rate is greater than LeConte Glacier from
2005 to 2010. For the entire period covered by all GLS datasets, GLS1975 to GLS2010, Baird
83

Glacier had an average movement rate of only 0.37 that of LeConte Glacier (16m / 43m =
0.372093m ~ 0.37m). Table 20 and Figure 38 summarize these findings. However, when slope
was considered, a different pattern emerged. Referring to Table 16, Baird Glacier’s slope is
3.02% and its average yearly movement rate is 16m; conversely, LeConte Glacier’s slope is
9.20% and its average yearly movement rate is 43m. Based only upon average yearly movement
rates, it may be concluded that glacier terminuses that end in marine were more conducive to
glacier movement (retreat or advance) than a terminus that were located on land. However, both
Baird and LeConte Glaciers typified the relationship between slope and movement rates. When
slope was considered, LeConte Glacier’s greater movement rate might have less to do with
glacier terminus terrain types and more to do with increased slope (7.50% versus 3.02%). More
variables would need to be considered before a definitive conclusion can be reached.
LeConte Glacier versus Patterson Glacier
The comparison of LeConte Glacier versus Patterson Glacier considered marine and
freshwater terminal terrain conditions, respectively. Analysis of the Table 15 values revealed that
in two of the four dataset periods, LeConte Glacier’s average movement rate is less than
Patterson Glacier; from 1975 to 1990 and from 2000 to 2005. LeConte Glacier’s average
movement rate is greater than Patterson Glacier from 1990 to 2000 and 2005 to 2010. For the
entire period covered by all GLS datasets, GLS1975 to GLS2010, LeConte Glacier had an
average movement rate of only 0.72 that of Patterson Glacier (43m / 60m = 0.716667m ~
0.72m). Table 21 and Figure 39 summarize these findings. Based solely upon average yearly
movement rates, it might be concluded that glacier terminuses that end in fresh water are more
conducive to glacier movement (retreat or advance) than a terminus which is located in marine.
84

However, when slope is considered, a contradiction appeared. Referring to Table 16, LeConte
Glacier’s slope is 9.20% and its average yearly movement rate is 43m; conversely, Patterson
Glacier’s slope is 7.50% and its average yearly movement rate is 60m. In this comparison,
LeConte Glacier’s slope is greater than Patterson Glacier’s (9.20% versus 7.50%), but its average
movement rate is less (43m versus 60m). Based upon average movement rates and slope, a
pattern did not emerge. Move variables should be considered before a definitive conclusion can
be reached regarding whether fresh water or marine is more conducive to glacier movement.
Table 21. Movement summary for Baird Glacier versus Patterson Glacier, Baird Glacier versus LeConte
Glacier, and LeConte Glacier versus Patterson Glacier. Movement rate multipliers greater than 1.00
indicate that the movement rate of the first glacier in each comparison is greater than the movement rate
of the second glacier. In addition to analyzing each GLS dataset period, an overall movement rate
multiplier was determined for the entire length of the study area.
Average Movement Rate Multiplier
GLS Dataset Temporal Range
Baird Glacier
versus Patterson
Glacier
Baird Glacier
versus LeConte
Glacier
LeConte Glacier versus
Patterson Glacier
GLS1975 to GLS1990 0.20 0.64 0.31
GLS1990 to GLS2000 0.04 0.03 1.23
GLS2000 to GLS2005 0.14 0.63 0.22
GLS2005 to GLS2010 4.15 1.36 3.05
GLS1975 to GLS2010 0.27 0.37 0.72

85

Figure 39. Movement comparison for Baird Glacier versus Patterson Glacier, Baird Glacier versus
LeConte Glacier, and LeConte Glacier versus Patterson Glacier. Movement rate multipliers greater than
1.00 indicate that the movement rate of the first glacier in each comparison was greater than the
movement rate of the second glacier.
Analysis of the average rates of change over the full range of GLS datasets, as
summarized in Table 15, inferred that glacier terminuses that end in freshwater (Shakes and
Patterson) were more conducive to movement than those that were located in a marine
environment (LeConte) or on land (Baird). Likewise, terminuses located in marine environments
(LeConte) were more conducive to movement than those located on land (Baird). The average
movement per year for Shakes Glacier is 101m; Patterson Glacier is 60m. LeConte Glacier is
43m, and Baird Glacier is 16m. Although Shakes Glacier averaged 101m, it probably should be
considered atypical for this study due to its rapid movement. Considering Baird, Patterson, and
LeConte Glaciers, the difference in average movement rates between the most active glacier,
0
1
2
3
4
5
GLS1975 to
GLS1990
GLS1990 to
GLS2000
GLS2000 to
GLS2005
GLS2005 to
GLS2010
Average Movement Rate Multiplier
GLS Dataset Temporal Range
Average Movement Rate Multiplier for Glaciers with Dissimilar
Terminal Terrain Conditions
Baird Glacier
versus Patterson
Glacier
Baird Glacier
versus Le Conte
Glacier
Le Conte Glacier
versus Patterson
Glacier
86

Patterson, and the least active glacier, Baird, is 44m. Analysis of Table 20 also revealed wide
fluctuation in glacier movement values during the various GLS dataset periods.
When glacier slope is considered, relative glacier movement appeared to be more
dependent upon glacier slope than terminal terrain conditions. Baird Glacier had the least slope
and lowest average movement rates (3.02% and 16m); LeConte Glacier had the greatest slope
and second lowest average movement rate (9.20% and 43m); Patterson Glacier had the second
greatest slope and third greatest average movement rate (7.50% and 60m); and Shakes Glacier
had the second lowest slope and the greatest average movement rate (3.80% and 101m). The
most apt conclusion that may be inferred from the data presented in this project is that each
glacier acted independently of other glaciers and each glacier’s movement is the result of the
input from many variables.
Simply considering rates of change and glacier slopes for Baird, Patterson, LeConte, and
Shakes Glaciers is insufficient to determine if any particular terminal terrain condition is more
conducive to glacier position change. Rather, other variables, such as glacier cross-section,
glacier valley profile, and direction of flow should be considered. For example, Shakes Glacier’s
glacial valley is physically straighter than Baird, Patterson, or LeConte Glaciers’ and contained
no bottlenecks to restrict the flow of ice. While Shakes Glacier had the greatest average
movement rate of 101m, it only had 3.80% slope; this relationship is contrary to what is expected
(Waddington, 2009).

87

CHAPTER SIX: CONCLUSIONS AND SUGGESTED COMPLIMENTARY
STUDIES
6.1 Conclusions
This project initially assumed that there would be a correlation between glacier
movement rate and glacier slope. However, the results did not conclusively support this
assumption. Shakes Glacier had the greatest average yearly movement rate of 101m, but had a
very slight slope of 3.80%. Conversely, LeConte Glacier had much a smaller average yearly
movement rate of 43m but had the greatest slope of 9.20%. Glacier movement rates and slope
values are summarized in Table 18. Based upon these findings, it must be assumed that factors
other than slope affect glacier movement.
While no definitive relationship between glacier slope and average movement rates
could be established, there did appear to be a relationship between increased average movement
rates and glacier terminuses that were located in water. Referring to Table 18, Baird Glacier,
whose terminus is located on land, had the lowest average movement rate of 16m. LeConte
Glacier, with a terminus located in salt water, had a higher yearly average movement rate of
43m. Patterson and Shakes Glaciers, which both end in fresh water, had the greatest average
yearly movement rates of 60m and 101m respectively. Based upon these findings, the presence
of water, especially fresh water, at the terminal end of glaciers had a greater effect on glacier
movement than slope. Possible explanations for this effect might include a heat sink effect or
tidal motions that hasten glacier disintegration in the ablation zone. In a heat sink scenario, it is
hypothesized that the water bodies that LeConte, Patterson, and Shakes Glaciers terminus are
located in act as a thermal energy transfer medium that increases glacier melting and subsequent
retreat. Helmenstine (2014) proposes that the increased molecular density of water versus ice and
88

the nature of thermal energy to move from areas of higher concentration to areas of lower
concentration are largely responsible for water induced glacial ice melting. Tidal motions
hypothetically act as horizontal and vertical push/pull forces, which increase the fracturing rate,
calving, and subsequent retreat of LeConte Glacier. Walter et al. (2012) discovered that tidal
influences on a marine-terminating Greenland glacier significantly affected glacier ablation.
Further studies are necessary to test these hypotheses to determine if a heat sink effect and tidal
motions significantly affect the movement rates for the glaciers in this study. Over the length of
the study period, there was 0.85°C increase in annual air temperatures. While this value may
seem low, Davies (2014) was able to force a change in a glacier prediction model using only
0.5°C change in temperature. This temperature increase of 0.85°C may prove important when
determining glacial mass balance rates.
A significant result of this study is the creation of shapefiles delineating the positions of
Shakes Glaciers in the Global Land Survey (GLS) dataset time periods. These shapefiles can be
submitted to the Global Land Ice Measurements from Space (GLIMS) program for inclusion in
their master worldwide glacier database, which currently does not include Shakes Glacier.
Although the submission process for the shapefile is not included in the thesis, it is ongoing, and
general information is deferred to the GLIMS site and possibly available within six months from
the time of this writing. This information should be useful to other researchers and its inclusion
in GLIMS would insure that they could readily access information about Shakes Glacier.
6.2 Lessons learned
This project was analytically challenging. The self-imposed constraint of using the GLS
datasets was both beneficial and problematic. From a historical perspective, the GLS datasets
89

provided an un-parallel source of multispectral imagery. The disadvantage to this dataset is the
30m spatial resolution (60m for GLS°975). Although the false-color shortwave infrared
composite (RGB 4, 5, 7) was invaluable for determining the extent of clean glacial ice, it had
difficulty in determining the extent of dirty ice. Dirty ice, with its large concentration of rock and
soil, exhibited spectral properties similar to the bare ground and soil that surrounds the periphery
of the glacier terminus. Differentiating the extent of glacial ice with this visual analysis alone
was often insufficient.
Image band combinations, like the false-color shortwave infrared composite (RGB: 4, 5,
7), and the true color composite (RGB: 3, 2, 1); unsupervised classifications; and individual
image bands, like the near infrared (Band 4), were often difficult to interpret and could not be
used exclusively to delineate glacier terminuses. Often, the line between glacial ice and
surrounding terrain was indistinct. Determining each terminus required constantly switching
between the various composite images and image bands to verify whether each pixel at a
glacier’s terminus represents a point on a glacier’s terminus. This method was time consuming
and each glacier terminus was revised several times before the results were acceptable.
The Global Land Ice Measurements from Space (GLIMS) data did not match the GLS
dataset very well, as discussed in Section 5.3. For example, the GLIMS data for LeConte Glacier
was determined from the GLS2005 dataset. When the GLIMS data was overlaid on the GLS2005
imagery, the difference between the terminuses of LeConte Glacier was off in several places by
hundreds of meters. Areas that GLIMS indicates as ice was in fact bare ground and vice-versa. It
is important to note, that an authoritative data source, like GLIMS, can also potentially have
errors and should be used cautiously.
90

Iterative Self-Organizing (ISO) Data Cluster Unsupervised Classification methods were
useful but were often difficult to interpret. Often, the process resulted in misclassified image
pixels which required traditional image interpretation skills to assess the validity. It was often the
case that a particular pixel was examined in multiple images to determine if it was glacial ice.
This was tedious and required a considerable investment in time to accomplish. More often than
not, resulting classes were not pure and had to be labeled based upon the most prevalent feature
in that class. Despite the many challenges that were presented during the course of this project, it
should be noted that all imagery and other geospatially enabled data was free.
6.3 Suggested complimentary studies
There are large differences in glacier movement values that resulted from this project, as
summarized in Table 14. Although the goal of the project was not seeking to specifically
determine why there was such high rate of variability, considerations were made based on terrain
conditions and slope. It is clear that considering only terminal terrain conditions and slope is
insufficient to determine whether one terminal terrain condition is more conducive to glacier
movement than another. Additional studies should be attempted to determine other variables that
could affect glacier movement in this area, for which suggestions are provided below.
For example, it is well established that glaciers respond to changes in climate conditions.
Glacier lag and response times are unique to each glacier (Pelto et al., 2001). Establishing a
particular glacier’s lag and response time requires a thorough understanding of its mass balance
change rate (Davies, 2014). Research suggests that mass balance change rates have yet to be
established definitively for Baird, Patterson, LeConte or Shakes Glaciers. If lag and response
91

times were determined for these Glaciers, it might offer some insight into the movement patterns
for these glaciers.
This study considered air temperature in the context that warming trends could explain
the movement patterns of the glaciers in the study (Waddington, 2009). This historical data was
collected from a weather station in Petersburg, Alaska, which is approximately 30km to the west
of the study area. Establishing on-site meteorological collection equipment would provide more
useful data. In addition to weather data, collecting water temperature data for Patterson,
LeConte, and Shakes Glaciers would be useful to test for any heat-sink effect.
Another factor that contributes to glacier movement rates is subsurface bed composition
and profile (Michna, 2012). Technologies like ground penetrating radar can be used to examine
these characteristics. Understanding the physical characteristics of the channel that each glacier
flows through may help to explain movement patterns.
Glaciers change shape and size simultaneously in three dimensions: length, width, and
thickness. This study only considered length; additional studies should consider change in all
dimensions. Newer technologies, like Light Detection and Ranging (LiDAR), would be very
useful for determining the three-dimensional shape of each glacier. A limiting factor of LiDAR
surveys is that the sensors must be able to “see” the ground to sense it. The study area, located in
the Tongass, was often obscured from satellite observation due to overcast, misty, or rainy
weather conditions (Sections 2.1 and 3.1). These same weather conditions precluded high flying
aircraft for LiDAR surveys. The steep terrain (Section 2.1) also precluded low flying survey
aircraft due to the operational risk for aircraft operating in steep, often ascending terrain. Because
92

of these difficulties, it may be necessary to identify a new type of aerial platform that is capable
of cost-effectively surveying glaciers in remote settings and with minimal risk to human life.

93

REFERENCES
Aher, Sainath P. and Sanjaykumar Dalvi. 2012. "Remote Sensing Techniques for Monitoring the
Glacier Retreating Process and Climatic Changes Study." Indian Streams Research
Journal 2 (8). Accessed October 1, 2012. http://www.isrj.net/PublishArticles/1200.pdf.
Alaska Climate Research Center. 2010. “Alaska Climatology.” Fairbanks: Alaska Climate
Research Center. Accessed January 31, 2013.
http://climate.gi.alaska.edu/Climate/index.html.
Alaska Department of Fish and Game. 2012. “Commercial Fisheries.” Juneau: Alaska
Department of Fish and Game. Accessed November 22, 2012.
http://www.adfg.alaska.gov/index.cfm?adfg=fishingcommercial.main.
Alaska Department of Fish and Game. 2012. “Southcentral.” Juneau: Alaska Department of Fish
and Game. Accessed November 22, 2012.
http://www.adfg.alaska.gov/index.cfm?adfg=fishingCommercialByArea.southcentral.
Alaska Humanities Forum. 2013. “The Geography of Alaska: Places and Regions.” Anchorage:
Alaska Humanities. Accessed February 15, 2013.
http://www.akhistorycourse.org/articles/article.php?artID=122.
Alesheikh, A. A., A. Ghorbanali, and N. Nouri. 2007. "Coastline Change Detection using
Remote Sensing." International Journal of Environmental Science and Technology:
(IJEST) 4 (1): 61-66. Accessed October 1, 2012.
http://search.proquest.com/docview/199331137?accountid=14749.
ArcGIS (Version 10.2) [Software]. (2013). Redlands: Environmental Systems Research Institute,
Inc. Retrieved from http://www.esri.com/.
Arendt, A., T. Bolch, J.G. Cogley, A. Gardner, J.-O. Hagen, R. Hock, G. Kaser, W.T. Pfeffer, G.
Moholdt, F. Paul, V. Radić, L. Andreassen, S. Bajracharya, M. Beedle, E. Berthier, R.
Bhambri, A. Bliss, I. Brown, E. Burgess, D. Burgess, F. Cawkwell, T. Chinn, L. Copland,
B. Davies, H. De Angelis, E. Dolgova, K. Filbert, R. Forester, A. Fountain, H. Frey, B.
Giffen, N. Glasser, S. Gurney, W. Hagg, D. Hall, U.K. Haritashya, G. Hartmann, C.
Helm, S. Herreid, I. Howat, G. Kapustin, T. Khromova, C. Kienholz, M. Koenig, J.
Kohler, D. Kriegel, S. Kutuzov, I. Lavrentiev, R. LeBris, J. Lund, W. Manley, C. Mayer,
E. Miles, X. Li, B. Menounos, A. Mercer, N. Moelg, P. Mool, G. Nosenko, A. Negrete,
C. Nuth, R. Pettersson, A. Racoviteanu, R. Ranzi, P. Rastner, F. Rau, B.H. Raup, J. Rich,
H. Rott, C. Schneider, Y. Seliverstov, M. Sharp, O. Sigurðsson, C. Stokes, R. Wheate, S.
Winsvold, G. Wolken, F. Wyatt, and N. Zheltyhina. 2012. “Randolph Glacier Inventory
[v2.0]: A Dataset of Global Glacier Outlines.” Boulder: Global Land Ice Measurements
94

from Space. Accessed February 26, 2011.
http://www.glims.org/RGI/RGI_Tech_Report_V2.0.pdf.
Bahr, David B., W. Tad Pfeffer, Christophe Sassolas, and Mark F. Meier. 1998. “Response Time
of Glaciers as a Function of Size and Mass Balance.” Journal of Geophysical Research
103 (B5): 9777-9782.
Baolin, Li, Zhang Yichi, and Zhou Chenghu. 2004. "Remote Sensing Detection of Glacier
Changes in Tianshan Mountains for the Past 40 Years." Journal of Geographical
Sciences 14 (3): 296-302.
Beuchle, Rene, Hugh D. Eva, Hans-Jürgen Stibig, Catherine Bodar, Andereas Brink, Philippe
Mayaux, Desiree Johansson, Frederic Achard, and Alan Belward. 2011. “A Satellite
Dataset for Tropical Forest Area Change Assessment.” International Journal of Remote
Sensing 32 (22): 7009-7031. doi:10.1080/01431161.2011.611186.
Bieniek, Peter A., Uma S. Bhatt, Richard L. Thoman, Heather Angeloff, James Partain, John
Papineau, Frederick Fritsch, Eric Holloway, John E. Walsh, Christopher Daly, Martha
Shulski, Gary Hufford, David F. Hill, Stavros Carlos, and Rudiger Gens. 2012. “Climate
Zones of Alaska.” Journal of Applied Meteorology and Climatology 51: 1276-1289.
Billington, James H. 2013. “America’s Story.” Washington: Library of Congress. Accessed
March 6, 2013. http://www.americaslibrary.gov/jb/recon/jb_recon_alaska_1.html.
Bishop, Michael P., Jeffrey A. Olsenholler, John F. Shroder, Roger G. Barry, Bruce H. Raup,
Andrew B. G. Bush, Luke Copland, John L. Dwyer, Andrew G. Fountain, Wilfried
Haeberli, Andreas Kääb, Frank Paul, Dorothy K. Hall, Jeffrey S. Kargel, Bruce F.
Molnia, Dennis C. Trabant, and Rick Wessels. 2004. “Global Land Ice Measurements
from Space (GLIMS): Remote Sensing and GIS Investigations of the Earth’s
Cryosphere.” Geocarto International 19 (2): 57-74. Accessed February 26, 2013.
http://www.geocarto.com.hk/cgi-bin/pages1/june04/bishop.pdf.
Boyce, Eleanor S., Roman J. Motyka, and Martin Truffer. 2007. “Flotation and Retreat of a
Lake-Calving Terminus, Mendenhall Glacier, Southeast Alaska, USA.” Journal of
Glaciology 53 (181): 211-224. doi: 10.3189/172756507782202928.
Bruland, Kris. 2014. “Petersburg, Alaska Tides.” Coupeville: Admiralty Software, LLC.
Accessed March 14, 2014. http://www.americantides.com/tide-predictions/petersburg-
alaska.
95

Campbell, James B. and Randolph H. Wynne. 2011. “Introduction to Remote Sensing.” 5
th

Edition. New York & London: The Guilford Press.
Canadian Space Agency. 2013. “RADARSAT-1.” Saint-Hubert: Canadian Space Agency.
Accessed November 30, 2013. http://www.asc-
csa.gc.ca/eng/satellites/radarsat1/components.asp.
Cape Decision Lighthouse Society. 2013. “Tongass Facts.” Sitka: Cape Decision Lighthouse
Society. Accessed February 11, 2013. http://capedecisionlight.org/Tongass-facts.html.
Davies, Bethan. 2014. “Glacier Response Time.” Antarctic Glaciers.org. Accessed February 12,
2013. http:// antarcticglaciers.org/glaciers-and-climate/glacier-response-time/.
Earth Resources Observation and Science Center. 2012. “Global Land Survey (GLS).”
Washington: United States Geological Survey. Accessed February 22, 2013.
http://eros.usgs.gov/#/Find_Data/Products_and_Data_Available/GLS.
Environmental Systems Research Institute, Inc. 2013. “Online base map data.” Redlands:
Environmental Systems Research Institute. Accessed February 11, 2013.
http://www.esri.com/.
Frery, Alejandro C., Susana Ferrero, and Oscar H. Bustos. 2009. "The Influence of Training
Errors, Context and Number of Bands in the Accuracy of Image Classification."
International Journal of Remote Sensing 30 (6): 1425-1440. Accessed October 1, 2012.
doi:10.1080/01431160802448919.
Geophysical Survey Systems, Inc. 2012. “Ground Penetrating Radar for Ice and Snow Profiling”.
Salem: Geophysical Survey Systems, Inc. Accessed December 2, 2013.
http://www.geophysical.com/directionstogssi.htm.
GLIMS: Global Land Ice Measurements from Space. 2007. “GLIMS Impacts.” Boulder:
National Snow and Ice Data Center. Accessed February 26, 2013.
http://www.glims.org/About/impacts.html.
GLIMS: Global Land Ice Measurements from Space. 2013. “Home Page.” Boulder: National
Snow and Ice Data Center. Accessed February 28, 2013. http://www.glims.org/.
Gutman G, R. Byrnes, J. Masek, S. Covington, C. Justice, S. Franks, and R. Headley. 2008.
“Towards monitoring land cover and land use changes at a global scale: the global land
survey 2005.” Photogrammetric Engineering & Remote Sensing 74 (1): 6–10. Accessed
March 10, 2013. http://gls.umd.edu/documents/GLS_2005_PERS_Jan2008_Gutman.pdf.
96

Haq, NM Anul, Kamal Jain, and KPR Menon. 2012. "Change Monitoring of Gangotri Glacier
using Remote Sensing." International Journal of Soft Computing & Engineering 1 (6):
259-261.
Haritashya, Umesh K., Michael P. Bishop, John F. Shroder, Andrew B. Bush, G., Henry N.
Bulley, and N. 2009. "Space-Based Assessment of Glacier Fluctuations in the Wakhan
Pamir, Afghanistan." Climatic Change 94 (1-2): 5-18. doi:10.1007/s10584-009-9555-9.
Harper, Joel T. 1993. “Glacier Terminus Fluctuations on Mount Baker, Washington, U.S.A.,
1940-1990, and Climatic Variations.” Arctic and Alpine Research 25 (4) 332-340.
Accessed February 13, 2014. http://www.jstor.org/stable/1551916.
Hekkers, Michael. 2010. “Mendenhall Glacier Science.” Juneau: University of Alaska Southeast.
Accessed November 19, 2012.
http://www.uas.alaska.edu/arts_sciences/naturalsciences/envs/links.html.
Helmenstine, Anne Marie. 2014. “Does Ice Melt Faster in Water or Air.” About.Com Chemistry.
Accessed March 14, 2014. http://chemistry.about.com/od/howthingswork/f/ice-melt-
faster-water-air.htm.
Hood, Eran., Jason Fellman, Robert G. M. Spencer, Peter J. Hernes, Rick Edwards, David
D’Amore, and Durelle Scott. 2009. “Glaciers as a Source of Ancient and Labile Organic
Matter to the Marine Environment.” Nature 462 (24): 1044-7. Accessed October 1, 2012.
doi:10.1038/nature08580.
Irons, James R. 2013. “The Worldwide Reference System.” Washington: National Aeronautics
and Space Administration. Accessed February 23, 2013.
http://landsat.gsfc.nasa.gov/about/wrs.html.
Jet Propulsion Laboratory. 2013. “ASTER Advanced Spaceborne Thermal Emission and
Reflection Radiometer.” Pasadena: Jet Propulsion Laboratory at California Institute of
Technology. Accessed November 30, 2013. http://asterweb.jpl.nasa.gov/.
Justice, Chris. 2013. “Global Land Survey.” College Park: National Aeronautics and Space
Administration: Land-Cover and Land-Use Change. Accessed February 23, 2013.
http://gls.umd.edu/index.html.
Kim, Yonl-Il, Yang-Dam Eo, and Byoung-Kil Lee. 1999. "Analyzing the Spatial Distribution
Pattern of Image Classification Error." Journal of the Japan Society of Photogrammetry
and Remote Sensing 38 (2): 53-60.
97

Klibanoff, Peter, Alvaro Sandroni, Boaz Moselle, and Brett Saraniti. 2005. “Managerial
Statistics: A Case-Based Approach.” 1
st
Edition. Cincinnati: South-Western College Pub.
Krenke, A. N. and V.M. Menshutin. 1987. “Calculation of Mass Balance of Glaciers by Remote-
Sensing Imagery Using Similarity of Accumulation and Ablation Isoline Patterns.”
Journal of Glaciology 33 (115): 363-368.
Lillesand, Thomas M., Ralph W. Kiefer, and Jonathan W. Chipman. 2008. “Remote Sensing and
Image Interpretation.” 6
th
Edition. New York: John Wiley & Sons, Inc.
Lindquist, E. J., M. C. Hansen, D. P. Roy, and C. O. Justice. 2008. “The Suitability of Decadal
Image Datasets for Mapping Tropical Forest Cover Change in The Democratic Republic
of Congo: Implications for the Global Land Survey.” International Journal of Remote
Sensing.” 29 (24): 7269-7275. doi: 10.1080/01431160802275890.
Masek, Jeff. 2011. “LEDAPS.” Washington: National Aeronautics and Space Administration.
Accessed February 24, 2013. http://ledaps.nascom.nasa.gov/.
Meier, M. F., and Austin Post. 1986. “Fast Tidewater Glaciers.” Journal of Geophysical
Research. 92 (B9): 9051-9058. Accessed February 12, 2014. doi:
10.1029/JB092iB09p09051.
Michna, Paul. 2012. “Glaciers.” Earth Science Australia. Accessed March 21, 2013.
http://earthsci.org/education/teacher/basicgeol/glacier/glacier.html#MovementofGlaciers.
Microsoft Excel (Version 10) [Software]. (2010). Redmond: Microsoft Corporation, Inc.
Retrieved from http://www.microsoft.com/en-us/default.aspx.
Molnia, B.F., 2008. Glaciers of North America -- Glaciers of Alaska, in Williams, R.S., Jr., and
Ferrigno, J.G., eds., Satellite image atlas of glaciers of the world: U.S. Geological Survey
Professional Paper 1386-K, 525 p. Accessed March 1, 2013.
http://pubs.usgs.gov/pp/p1386k/.
Motyka, Roman J., Shad O’Neel, Cathy L. Connot, and Keith A. Echelmeyer. 2003. “Twentieth
Century Thinning of Mendenhall Glacier, Alaska, and its Relationship to Climate, Lake
Calving, and Glacier Run-off.” Global and Planetary Change. 35 (1-2): 93-112. doi:
10.1016/S0921-8181(02)00138-8.
Muralidhar, Ashwini. 2011. "Augmented Image Classification using Image Registration
Techniques." M.S., Arizona State University. Accessed October 1, 2012.
98

National Aeronautics and Space Administration. 2012. “From the Beginning.” Washington:
National Aeronautics and Space Administration. Accessed November 19, 2012.
http://landsat.gsfc.nasa.gov/history.html.
National Ocean Service. 2013. “LIDAR-Light Detection and Ranging-is a Method to Examine
the Surface of the Earth.” Silver Spring: National Oceanic and Atmospheric
Administration’s National Ocean Service. Accessed November 30, 2013.
http://oceanservice.noaa.gov/facts/lidar.html.
National Snow and Ice Data Center. 2013. “Programs and Projects ant NSIDC.” Boulder:
National Snow and Ice Data Center. Accessed February 25, 2013.
http://nsidc.org/programs/.
National Snow and Ice Data Center. 2013. “Submit Data.” Boulder: National Snow and Ice Data
Center. Accessed February 25, 2013. http://nsidc.org/data/submit.html.
Noderer, Michael. 2010. Intermediate Spectral Exploitation and Analysis Course II (ISEA-II).
Springfield: National Geospatial-Intelligence Agency, National Geospatial-Intelligence
College.
Noderer, Michael. 2007. Remotely Sensed Imagery Course (RSI). Fort Belvoir: National
Geospatial-Intelligence Agency, National Geospatial-Intelligence College.
Patra, Swarnajyoti, Susmita Ghosh, and Ashish Ghosh. 2011. "Histogram Thresholding for
Unsupervised Change Detection of Remote Sensing Images." International Journal of
Remote Sensing 32 (21): 6071-6089. Accessed October 1, 2012.
doi:10.1080/01431161.2010.507793.
Pelto, Mauri S. and Cliff Hedlund. 2001. “Terminus Behavior and Response Time of North
Cascade Glaciers, Washington, U.S.A.” Journal of Glaciology 47 (158): 497-506.
Accessed February 12, 2014. doi: 10.3189/172756501781832098.
Pidwirny, Michael and Scott Jones. 2010. “Physical Geography.net: Glossary of Terms.”
Okanagan: University of British Columbia. Accessed March 6, 2013.
http://www.physicalgeography.net/physgeoglos/i.html.
Quincey, D. J. and A. Luckman. 2009. "Progress in Satellite Remote Sensing of Ice Sheets."
Progress in Physical Geography 33 (4): 547-567.
Raper, Sarah C. B. and Roger J. Braithwaite. 2006. "Low Sea Level Rise Projections from
Mountain Glaciers and Icecaps Under Global Warming." Nature 439 (7074): 311-3.
99

doi:http://dx.doi.org/10.1038/nature04448.
http://search.proquest.com/docview/204516467?accountid=14749. Raup, Bruce, Andreas
Kääb, Jeffrey S. Kargel, Michael P. Bishop, Gordon Hamilton, Ella Lee, Frank Paul,
Frank Rau, Deborah Soltesz, Siri Jodha Singh Khalsa, Mathew Beedle, and Christopher
Helm. 2007. "Remote Sensing and GIS Technology in the Global Land Ice
Measurements from Space (GLIMS) Project." Computers & Geosciences 33 (1): 104-
125. Accessed October 1, 2012. doi:10.1016/j.cageo.2006.05.015.
Raup, Bruce, Andreas Kääb, Jeffrey S. Kargel, Michael P. Bishop, Gordon Hamilton, Ella Lee,
Frank Paul, Frank Rau, Deborah Soltesz, Siri Jodha, Singh Khalsa, Mathew Beedle, and
Christopher Helm. 2007. “Remote Sensing and GIS Technology in the Global Land Ice
Measurements from Space (GLIMS) Project.” Computers and Science 33(2007): 104-
125. doi: 10.1016/j.cageo.2006.05.015.
Schweiker, Roy and David Olson. 2012. “U.S. National Forest High Point.” Accessed February
20, 2013. http://www.peakbagger.com/list.aspx?lid=1825.
Sharma, Bharat R., Upali A. Amarasinghe, and Alok Sikka. 2008. “Indo-Gangetic River Basins:
Summary Situation Analysis.” International Water Management Institute. New Delhi
Office, New Delhi, India.
Shukla, A., M. K. Arora, and R. P. Gupta. 2010. "Synergistic Approach for Mapping Debris-
Covered Glaciers using optical–thermal Remote Sensing Data with Inputs from
Geomorphometric Parameters." Remote Sensing of Environment 114 (7): 1378-1387.
Accessed October 1, 2012. doi:10.1016/j.rse.2010.01.015.
Space Sciences Laboratory. 2013. “SETI@Home.” Berkley: University of California. Accessed
February 1, 2013. http://setiathome.ssl.berkley.edu/.
State of Alaska Department of Labor and Workforce Development. 2013. “Research and
Analysis.” Juneau: State of Alaska Department of Labor and Workforce Development.
Accessed February 11, 2013. http://labor.alaska.gov/research/census/maps.htm.
State of Alaska Department of Natural Resources. 2012. “ASGDC: Alaska State Geo-Spatial
Data Clearinghouse.” Anchorage: State of Alaska Department of Natural Resources.
Accessed February 28, 2013. http://www.asgdc.state.ak.us/.
Tappan, Ray and Matthew Cushing. 2004. Use of SLC-Off Landsat Image Data for Monitoring
Land use / Land Cover Trends in West Africa. Sioux Falls, SD: United States Geological
Survey.
100

Théau, Jérôme. 2012. "Change Detection." In Springer Handbook of Geographic Information,
75-94. Berlin, Heidelberg: Springer Berlin Heidelberg.
United States Census Bureau. 2013. “American Fact Finder.” Washington: United States Census
Bureau. Accessed February 11, 2013.
http://factfinder2.census.gov/faces/nav/jsf/pages/index.xhtml.
United States Department of Agriculture, Forest Service. 2013. “About the Forest.” Washington:
United States Department of Agriculture, Forest Service. Accessed January 31, 2012.
http://www.fs.usda.gov/main/tongass/about-forest.
United States Department of Agriculture, Forest Service Geospatial Service and Technology
Center. 2013. “FSGeoData Clearing House.” Salt Lake City, United States Department of
Agriculture Forest Service Geospatial Service and Technology Center. Accessed
February 10, 2013.
United States Geological Survey. 2013. “Earth Explorer Data Portal.” Washington: United States
Geological Survey. Accessed February 23, 2013. http://earthexplorer.usgs.gov/.
United States Geological Survey. 2012. “Glaciers and Icecaps: Storehouses of Freshwater.”
Washington: United States Department of the Interior, United States Geological Survey.
Accessed October 1, 2012. http://ga.water.usgs.gov/edu/earthglacier.html.
United States Geological Survey. 2013. “Global Land Survey.” Washington: United States
Geological Survey. Accessed February 20, 2013 with Earth Explorer Data Portal.
http://earthexplorer.usgs.gov/.
United States Geological Survey. 2012. “Landsat: A Global Land-Imaging Mission.” Sioux
Falls: United States Geological Survey. Accessed February 23, 2013.
http://pubs.usgs.gov/fs/2012/3072/fs2012-3072.pdf.
United States Geological Survey. 2013. “Landsat Missions.” Reston: United States Geological
Survey. Accessed November 30, 2013. http://landsat.usgs.gov/.
United States Geological Survey. 2013. “The USGS Land Cover Institute (LCI).” Washington:
U.S. Department of the Interior, U.S. Geological Service. Accessed February 28, 2013.
http://landcover.usgs.gov/.
University of Alaska Fairbanks. 2012. “Alaska Mapped and the Statewide Digital Mapping
Initiative.” Fairbanks: University of Alaska Fairbanks. Accessed February 15, 2013.
http://www.alaskamapped.org/.
101

Waddington, Ed. 2009. “Principles of Glaciology ESS 431.” Seattle: University of Washington.
Accessed March 20, 2013.
http://courses.washington.edu/ess431/Lectures/lecture_2009/Lect_07_Glacier_flow_2009
.pdf.
Walter, Jacob I., Jason E. Box, Slawek Tulaczyk, Emily E. Brodsky, Ian M. Howat, Yushin Ahn,
and Abel Brown. 2012. “Oceanic Mechanical Forcing of a Marine-Terminating
Greenland Glacier.” Annals of Glaciology 53(60): 1-12. Accessed March 14, 2014. doi:
10.3189/2012AoG60A083.
Western Regional Climate Center. 2013. “Climate Zones of Alaska.” Reno: Western Regional
Climate Center. Accessed January 31, 2013.
http://www.wrcc.dri.edu/narratives/ALASKA.htm.
Wolfe, Robert, Jeffrey Masek, Nazmi Saleous, and Forest Hall. 2004. “LEDAPS: Mapping
North American Disturbance from the Landsat Record.” Greenbelt: National Aeronautics
and Space Administration: Goddard Space Flight Center. Accessed February 24, 2013.
http://ledaps.nascom.nasa.gov/docs/pdf/ledaps_igarss04.pdf.

102

APPENDIX A: GLOBAL LAND SURVEY (GLS) DATA SOURCING AND
DOWNLOAD
This appendix provides a step-by-step process flow for locating and downloading GLS
datasets.
Locating Global Land Survey (GLS) data
Data source: The GLS data was downloaded using the United States Geological Survey’s
Earth Explorer web portal (United States Geological Survey, 2013). The uniform resource
locator (URL) for Earth Explorer is http://earthexplorer.usgs.gov/.
Cost: There is no cost to download data from this service. However, I could not do it
anonymously; rather I had to create a user profile that identified 1) who I am and 2) why I am
using this data. This type of information usually allows the site administrator to generate site
statistics that can be used to justify future funding or service improvements.
Downloading GLS data
Downloading GLS data is a twelve-step process. Refer to Figures 39-44 for step-by-step
instructions on accomplishing this.
103

Figure 40. United States Geological Survey (USGS) Earth Explorer home page is the starting point for
downloading GLS datasets. 1: select the “Search Criteria” tab; 2: scale the display map into the study area
using the pan and zoom tools.

104

Figure 41. Define the area of interest for GLS image searches. 1: loosely bound the project study area by
left-clicking to form a polygon area of interest (AOI); outlined by the red box labeled. The labels for each
of the vertices correspond to the order in which the AOI is defined.

Figure 42. Switch from AOD definition to dataset(s) selection. 4: select the “Data Sets” tab.

105

Figure 43. Specify the datasets for downloading. 5: in the dataset list, expand the “Global Land Survey”
data list and select the desired datasets; 6: select the “Results” tab.

Figure 44. Earth Explorer search results page. This page displays the available data for the area of
interest. If multiple datasets were specified in Step 5 (Figure 53), then only one may be displayed at a
time. 7: select a dataset to view images for; 8: image metadata summary with WRS path and row address
and date of image collection; 9: image footprint allows for visual confirmation of image location; 10:
image download button. Table 22 summarizes relevant image information for the various GLS datasets
used for this study.

106

Figure 45. GLS data download options. 11: Level 1 Product is selected. For this dataset, the file size is
212.1 MB and is in Geotiff format. 12: “Select Download Option” button will begin the download
process.
The GLS data files are very large, in the case of GLS2010 data, that size is approximately
212 MB (Figure 44). The data file is compressed in a Tape ARchive (.tar) format. WinRAR
software was used to un-compress the data. For the GLS2010 data, the un-compressed size is
481MB. Large datasets like the GLS data can quickly use computer resources and may require
careful planning to ensure that these datasets can be used. As a comparison, a digital video disk
(DVD) can only store approximately eight GLS data scenes. Table 22 summarizes the image
details for all five GLS datasets. After downloading the GLS2010 dataset, GLS2005, GLS2000,
GLS1990, and GLS1975 datasets were downloaded in a similar manner by simply changing the
“Dataset” choice (Figure 43, Step “7”) and following all subsequent steps.

107

Table 22. Image details for Global Land Survey (GLS) datasets summary for Central Southeast Alaska
Glacier Study Project area.

GLS Dataset Entity ID Acquisition Date
WRS
Path
WRS
Row
Download
Option
Un-compressed
Size (MB)
GLS2010
LT505702020091
86PAC01
July 30, 2009 56 20
Level 1
Product
481
GLS2005
LE705602020052
24EDC00
August 12, 2005 56 20
Level 1
Product
636
GLS2000
P056R020_7X19
990812
August 12, 1999 56 20
Level 1
Product
656
GLS1990
P056R020_5X19
890909
September 9, 1989 56 20
Level 1
Product
391
GLS1975
P060R020_1X19
740903
September 3, 1974 60 20
Level 1
Product
70
Total 2234
108

APPENDIX B: GLOBAL LAND ICE MEASUREMENTS FROM SPACE (GLIMS)
DATA SOURCING AND DOWNLOAD
This appendix provides a step-by-step process flow for locating and downloading GLIMS
datasets.
Locating Global Land Ice Measurements from Space (GLIMS) data
Data source: The GLIMS data was downloaded using the National Snow and Ice Data
Center’s (NSIDC) GLIMS Glacier Viewer web portal (GLIMS: Global land Ice Measurements
from Space, 2013). The uniform resource locator (URL) for GLIMS: Global Land Ice
Measurements from Space is http://glims.org/.
Cost: There is no cost to download data from this service. Unlike Earth Explorer, no
login is required.
Downloading GLIMS data
Downloading GLIMS data is an eight-step process. Refer to Figures 45-49 for step-by-
step instructions on accomplishing this.

Figure 46. Global Land Ice Measurements from Space (GLIMS) home page. 1: select “Data Access”
button that will take the user to a screen where GLIMS data can be downloaded.
109

Figure 47. GLIMS glacier database home page. 2: select “Start GLIMS Glacier Viewer” to proceed to the
online data search portal.

Figure 48. GLIMS glacier database graphical summary page for the entire world. 3: change the map size
to “800x600”; 4: use the zoom tool to go to area of interest.

110

Figure 49. GLIMS glacier database graphical summary for display window. Glaciers are shown as
magenta outlines. 5: pan and zoom tools to refine area of interest; 6: download button to download all
glaciers in current extent.

111

Figure 50. GLIMS data download page. 7: select the type of file that the export will be stored as; 8: select
the “Download Data” button to download GLIMS data. For this step, select an ESRI Shapefile for easy
data viewing in ArcGIS 10.2. Note: if the extent is too large, the download process may crash. Also, the
GLIMS portal does not provide a download progress bar, so have patience and assume that the data
request is going to work. Once the data is fully downloaded, the user is notified.

112

APPENDIX C: GLACIER IMAGES USED TO CREATE THE GLACIER
TERMINUSES SHAPEFILES
Collection of the GLS datasets used as either stand-alone images or to create true color
and false color image composites and Iterative Self-Organizing (ISO) Data Unsupervised
Classifications to delineate the glacier terminuses. These images are displayed in Figures 50-69.
Images are grouped by GLS dataset and arranged in order of Baird, Patterson, LeConte, and
Shakes Glaciers. Table 23 is provided as a means to locate a particular glacier/GLS dataset
combination.
Table 23. Glacier images page number summary. The page numbers are given for all images used for
glacier terminus delineation.
Glacier Name
GLS2010
Page #
GLS2005
Page #
GLS2000
Page #
GLS1990
Page #
GLS1975
Page #
Baird 113 117 121 125 129
Patterson 114 118 122 126 129
LeConte 115 119 123 127 130
Shakes 116 120 124 128 130

113

GLS2010 Baird Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 51. Baird Glacier in GLS2010 images used for glacier terminus delineation. The collecting sensor
was Landsat 5 TM. The upper left image is Band 4 (NIR) image. The upper right image is a true-color
composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared composite
(RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification. Legends for
each image are displayed below their respective images.

114

GLS2010 Patterson Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 52. Patterson Glacier in GLS2010 images used for glacier terminus delineation. The collecting
sensor was Landsat 5 TM. The upper left image is Band 4 (NIR) image. The upper right image is a true-
color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared composite
(RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification. Legends for
each image are displayed below their respective images.

115

GLS2010 LeConte Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 53. LeConte Glacier in GLS2010 images used for glacier terminus delineation. The collecting
sensor was Landsat 5 TM. The upper left image is Band 4 (NIR) image. The upper right image is a true-
color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared composite
(RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification. Legends for
each image are displayed below their respective images.

116

GLS2010 Shakes Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 54. Shakes Glacier in GLS2010 images used for glacier terminus delineation. The collecting
sensor was Landsat 5 TM. The upper left image is Band 4 (NIR) image. The upper right image is a true-
color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared composite
(RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification. Legends for
each image are displayed below their respective images.

117

GLS2005 Baird Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 55. Baird Glacier in GLS2005 images used for glacier terminus delineation. The collecting sensor
was Landsat 7 ETM+. The upper left image is Band 4 (NIR) image. The upper right image is a true-color
composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared composite
(RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification. Legends for
each image are displayed below their respective images.

118

GLS2005 Patterson Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 56. Patterson Glacier in GLS2005 images used for glacier terminus delineation. The collecting
sensor was Landsat 7 ETM+. The upper left image is Band 4 (NIR) image. The upper right image is a
true-color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared
composite (RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification.
Legends for each image are displayed below their respective images.

119

GLS2005 LeConte Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 57. LeConte Glacier in GLS2005 images used for glacier terminus delineation. The collecting
sensor was Landsat 7 ETM+. The upper left image is Band 4 (NIR) image. The upper right image is a
true-color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared
composite (RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification.
Legends for each image are displayed below their respective images.

120

GLS2005 Shakes Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 58. Shakes Glacier in GLS2005 images used for glacier terminus delineation. The collecting
sensor was Landsat 7 ETM+. The upper left image is Band 4 (NIR) image. The upper right image is a
true-color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared
composite (RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification.
Legends for each image are displayed below their respective images.

121

GLS2000 Baird Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 59. Baird Glacier in GLS2000 images used for glacier terminus delineation. The collecting sensor
was Landsat 7 ETM+. The upper left image is Band 4 (NIR) image. The upper right image is a true-color
composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared composite
(RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification. Legends for
each image are displayed below their respective images.

122

GLS2000 Patterson Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 60. Patterson Glacier in GLS2000 images used for glacier terminus delineation. The collecting
sensor was Landsat 7 ETM+. The upper left image is Band 4 (NIR) image. The upper right image is a
true-color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared
composite (RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification.
Legends for each image are displayed below their respective images.

123

GLS2000 LeConte Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 61. LeConte Glacier in GLS2000 images used for glacier terminus delineation. The collecting
sensor was Landsat 7 ETM+. The upper left image is Band 4 (NIR) image. The upper right image is a
true-color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared
composite (RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification.
Legends for each image are displayed below their respective images.

124

GLS2000 Shakes Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 62. Shakes Glacier in GLS2000 images used for glacier terminus delineation. The collecting
sensor was Landsat 7 ETM+. The upper left image is Band 4 (NIR) image. The upper right image is a
true-color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared
composite (RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification.
Legends for each image are displayed below their respective images.

125

GLS1990 Baird Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 63. Baird Glacier in GLS1990 images used for glacier terminus delineation. The collecting sensor
was Landsat 5 TM. The upper left image is Band 4 (NIR) image. The upper right image is a true-color
composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared composite
(RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification. Legends for
each image are displayed below their respective images.

126

GLS1990 Patterson Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 64. Patterson Glacier in GLS1990 images used for glacier terminus delineation. The collecting
sensor was Landsat 5 TM. The upper left image is Band 4 (NIR) image. The upper right image is a true-
color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared composite
(RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification. Legends for
each image are displayed below their respective images.

127

GLS1990 LeConte Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 65. LeConte Glacier in GLS1990 images used for glacier terminus delineation. The collecting
sensor was Landsat 5 TM. The upper left image is Band 4 (NIR) image. The upper right image is a true-
color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared composite
(RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification. Legends for
each image are displayed below their respective images.

128

GLS1990 Shakes Glacier
Band 4 (NIR) True-Color Composite (RGB: 3, 2, 1)

False-Color SWIR Composite (RGB: 4, 5, 7) ISO Cluster Unsupervised Classification

Figure 66. Shakes Glacier in GLS1990 images used for glacier terminus delineation. The collecting
sensor was Landsat 5 TM. The upper left image is Band 4 (NIR) image. The upper right image is a true-
color composite image (RGB: 3, 2, 1). The lower left image is a false-color shortwave infrared composite
(RGB: 4, 5, 7). The lower right image is an ISO Data Cluster Unsupervised Classification. Legends for
each image are displayed below their respective images.

129

GLS1975 Baird Glacier
False-Color NIR Composite (RGB: 7, 5, 4) ISO Cluster Unsupervised Classification

Figure 67. Baird Glacier in GLS1975 images used for glacier terminus delineation. The collecting sensor
was Landsat 1 MSS. The lower left image is a false-color near-infrared composite (RGB: 7, 5, 4). The
right image is an ISO Data Cluster Unsupervised Classification. Legends for each image are displayed
below their respective images.

GLS1975 Patterson Glacier
False-Color NIR Composite (RGB: 7, 5, 4) ISO Cluster Unsupervised Classification

Figure 68. Patterson Glacier in GLS1975 images used for glacier terminus delineation. The collecting
sensor was Landsat 1 MSS. The lower left image is a false-color near-infrared composite (RGB: 7, 5, 4).
The right image is an ISO Data Cluster Unsupervised Classification. Legends for each image are
displayed below their respective images.

130

GLS1975 LeConte Glacier
False-Color NIR Composite (RGB: 7, 5, 4) ISO Cluster Unsupervised Classification

Figure 69. LeConte Glacier in GLS1975 images used for glacier terminus delineation. The collecting
sensor was Landsat 1 MSS. The lower left image is a false-color near-infrared composite (RGB: 7, 5, 4).
The right image is an ISO Data Cluster Unsupervised Classification. Legends for each image are
displayed below their respective images.

GLS1975 Shakes Glacier
False-Color NIR Composite (RGB: 7, 5, 4) ISO Cluster Unsupervised Classification

Figure 70. Shakes Glacier in GLS1975 images used for glacier terminus delineation. The collecting
sensor was Landsat 1 MSS. The lower left image is a false-color near-infrared composite (RGB: 7, 5, 4).
The right image is an ISO Data Cluster Unsupervised Classification. Legends for each image are
displayed below their respective images.

Abstract (if available)

Abstract Global Land Survey (GLS) data encompassing Landsat Multispectral Scanner (MSS), Landsat 5’s Thematic Mapper (TM), and Landsat 7’s Enhanced Thematic Mapper Plus (ETM+) were used to determine the terminus locations of Baird, Patterson, LeConte, and Shakes Glaciers in Alaska in the time period 1975-2010. The sequences of the terminuses locations were investigated to determine the movement rates of these glaciers with respect to specific physical and environmental conditions. ❧ GLS data from 1975, 1990, 2000, 2005, and 2010 in false‐color composite images enhancing ice‐snow differentiation and Iterative Self-Organizing (ISO) Data Cluster Unsupervised Classifications were used to 1) quantify the movement rates of Baird, Patterson, LeConte, and Shakes Glaciers

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Comparing Landsat7 ETM+ and NAIP imagery for precision agriculture application in small scale farming: a case study in the south eastern part of Pittsylvania County, VA

PDF

Exploring remote sensing and geographic information systems technologies to understand vegetation changes in response to land management practices at Finke Gorge National Park, Australia Between ...

PDF

Use of remotely sensed imagery to map sudden oak death (Phytophthora ramorum) in the Santa Cruz Mountains

PDF

Using census data, urban-land cover classification, and dasymetric mapping to measure urban growth of the lower Rio Grande Valley, Texas

PDF

Evaluating the utility of a geographic information systems-based mobility model in search and rescue operations

Asset Metadata

Creator Davidson, Robert Howard (author)

Core Title Evaluating glacier movement fluctuations using remote sensing: a case study of the Baird, Patterson, LeConte, and Shakes Glaciers in central southeastern Alaska

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 04/07/2014

Defense Date 02/06/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Alaska,Baird,geographic information systems,GIS,Glaciers,GLIMS,Global Land Ice Measurements from Space,Global Land Survey,GLS,Le Conte,LeConte,OAI-PMH Harvest,Patterson,remote sensing,RSI,Shakes

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Paganelli, Flora (committee chair), Lee, Su Jin (committee member), Stott, Lowell (committee member)

Creator Email rhdavidson37@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-373735

Unique identifier UC11295362

Identifier etd-DavidsonRo-2328.pdf (filename),usctheses-c3-373735 (legacy record id)

Legacy Identifier etd-DavidsonRo-2328.pdf

Dmrecord 373735

Document Type Thesis

Format application/pdf (imt)

Rights Davidson, Robert Howard

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA