EPA/600/3-85/034
April 1985
AERIAL PHOTOGRAPHY AND GROUND
VERIFICATION AT POWER PLANT SITES
Wisconsin Power Plant Impact Study
by
Sarah L. Wynn
and Ralph W. Kiefer
Civil and Environmental Engineering
University of Wisconsin-Madison
Madison, Wisconsin 53706
Grant No. R803971
Project Officer
Gary E, Glass
Environmental Research Laboratory-Duluth
Duluth, Minnesota
This study was conducted in cooperation with
Wisconsin Power and Light Company,
Madison Gas and Electric Company,
Wisconsin Public Service Corporation,
Wisconsin Public Service Commission,
and Wisconsin Department of Natural Resources
ENVIRONMENTAL RESEARCH LABORATORY-DULUTH
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
DULUTH, MINNESOTA 55804
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing/
i. REPORTNO.
EPA/600/3-85/034
3. RECIPIENT'S ACCESSION NO.
. PB85 19/35
4. TITLE AND SUBTITLE
Aerial Photography and Ground Verification at Power
Plant Sites: Wisconsin Power Plant Impact Study
5. REPORT DATE
April 1985
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
S. L. Wynn and R. W. Kiefer
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Civil and Environmental Engineering
University of Wisconsin-Madison
Madison, Wisconsin 53706
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
803971
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Duluth, Minnesota 55804
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/03
600/3
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This study demonstrated and evaluated nine methods for monitoring the deterioration of
a large wetland on the site of a newly-constructed coal-fired power plant in Columbia
County, Wisconsin. Four of the nine methods used data from ground sampling; two were
remote sensing methods without ground verification; and three were remote sensing
methods which either used ground verification or relied on the analyst's "on-the-
ground" knowledge of the area.
These methods were evaluated on the basis of whether they monitor change at a species
or a community level, whether they monitor community change in terms of area or
location or both, and whether they provide information about trends in plant
communities. They were also evaluated in terms of time, cost, sensitivity, and
reliability. Changes in the wetland over a 3-year period are presented, as determined
by each of the methods. Eight appendices provide information and raw data for
several of the methods, color/texture keys for interpreting airphotos, and an
annotated bibliography on remote sensing methods.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
IS. DISTRIBUTION STATEMENT
Release to public
19. SECURITY CLASS (This Report/
Unclassified
21. NO. OF PAGES
292
20. SECURITY CLASS (Thispage)
Unclassifiee
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
-------
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse-
ment or recommendation for use.
11
-------
TABLES
Table
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Example of Encoded Ground Sampling Data and the Scheme
with which They Were Coded
Columbia Generating Station Airphotos
Diversity Index
Summary of Subjective Classification
Reclassif ication of Ground Sampling Data Using Subjective
Results of Chi Square Test on Classification Results
Visually Dominant Species Used in the Association Analysis.....
Reclassification of Ground Sampling Data Using Association
Summary of Vegetation Structure Analysis Results
Reclassification of Ground Sampling Data Using Vegetation
Airphotos Used to Generate Disturbance Maps
Vegetation Classes Identified Using Airphoto Grid Analysis
Classes Originally Used to Sum Grid Analysis Data
Chi Square Tests on Grid Analysis Data
Pagj
18
20
23
26
27
28
33
34
35
38
39
41
58
68
70
72
73
74
X
-------
Figure Page
43 Output from program DLOGE 99
44 Density to log exposure correction using program CORRECT 100
45 D-log E curves generated by program CORRECT 101
46 Overprinted 10-level density slice 104
47 A 36-level density slice 105
48 Listing provided by program TRAIN 106
49 Bar diagrams generated by program CLASSBAR 108
50 Histograms created by program HSGRAM 109
51 Scatter diagram created by program SCATTER 110
52 Output from program CLEANTR Ill
53 Percentage matrix from program CLSTRN 113
54 Output from program BOX4 114
55 Portion of printed out parallelepiped classification created
from program BOX4 115
56 How a decision region can be broken into rectangles whose
borders closely describe the training set distribution 116
57 Portion of a maximum likelihood classification of the site 117
58 Probability values plotted on a three-dimensional graph of a
scatter diagram 119
59 Portion of a classification of the site using program
TABCLASS 120
60 Computer assisted map made from CIR airphoto on
September 15, 1975 121
61 Computer assisted map made from CIR airphoto on July 14,
1976 121
62 Computer assisted map made from CIR airphoto on June 25,
1977 122
63 Computer assisted map made from color airphoto on June 25,
1977 122
ix
-------
Figure Page
19 Color airphoto on June 25, 1977 53
20 CIR airphoto on October 3, 1977 54
21 Color airphoto on March 1, 1975 55
22 Color airphoto on March 1, 1976 56
23 Color airphoto on March 1, 1977 57
24 Disturbance map on September 25, 1975 60
25 Disturbance map on September 24, 1976 61
26 Disturbance map on June 25, 1977 62
27 Disturbance map on October 3, 1977 63
28 Disturbance map on March 1, 1975 64
29 Disturbance map on March 1, 1976 65
30 Disturbance map on March 1, 1977 66
31 Mylar overlay locating transects and sampling stations 69
32 Photo interpreted vegetation map on June 4, 1972 81
33 Photo interpreted vegetation map on July 31, 1974 .• 82
34 Photo interpreted vegetation map on September 25, 1975 83
35 Photo interpreted vegetation map on July 24, 1976 85
36 Photo interpreted vegetation map on September 24, 1976 86
37 Photo interpreted vegetation map on June 25, 1977 87
38 Photo interpreted vegetation map on June 15, 1977 88
39 Photo interpreted vegetation map on October 3, 1977 89
40 Spot densitometer 94
41 Two types of scanning microdensitometers 96
42 D-log E curves 97
viii
-------
FIGURES
Figure Page
1 Columbia Generating Station site 2
2 Vegetation map of site 3
3 Groundwater contours 4
4 Groundwater movement through the site before and after
construction of cooling lake 6
5 Time line of construction events at the Columbia Generating
Station site 7
6 Groundwater isotherms at various depths and distances from
the cooling lake dike 8
7 Wetland water level fluctuations 9
8 Colorinfrared airphoto of study site on July 24, 1976 17
9 Subjective classification of the ground sampling data 30
10 Dendrogram of ground sampling data by association analysis 32
11 Association analysis of the ground sampling data 36
12 CIR airphoto used to determine airphoto grid size on
October 3, 1977 44
13 CIR airphoto on June 4, 1972 46
14 CIR airphoto on July 31, 1974 47
15 CIR airphoto on September 25, 1975 48
16 CIR airphoto on July 14, 1976 50
17 CIR airphoto on September 24, 1976 51
18 CIR airphoto on June 25, 1977 52
vii
-------
CONTENTS
Foreword ill
Abstract iv
Figures vii
Tables x
1. Introduction 1
Impacts of plant construction and operation on the study site 5
2. Conclusions 12
3. Recommendations • 14
4. Materials and Methods 16
Ground sampling data collection 16
Airphoto data collection 19
Monitoring vegetation using ground sampling data 19
Use of airphotos to monitor change 42
Monitoring vegetation change with airphoto and ground
sampling data 67
Generating computer assisted vegetation maps 93
5. Results 129
Method efficiency 142
Method sensitivity and reliability 143
The ground sampling data methods 145
Airphoto data only methods 147
Airphoto grid analysis, airphoto interpreted vegetation mapping
and computer assisted mapping 149
6. Discussion 153
Using methods together 154
References 156
Appendices
A. Species Code List 160
B. Station and Transect Numbers 162
C. Diversity Index 163
D. Subjective Classification 165
E. Association Analysis 166
F. Airphoto Interpretation Keys for Airphoto Grid Assessment 167
G. Airphoto Interpretation Keys for Vegetation Mapping 170
H. Annotated Bibliography 174
VI
-------
classes—disturbed and undisturbed vegetation and open water. However, in
defining only three classes, an exceptionally understandable visual record
of change taking place is created.
Airphoto grid analysis, the second most costly method, records
percentage cover of vegetation classes on a cell by cell basis. This method
documents percent changes in community area but does not directly document
changes in community location.
Airphoto interpreted vegetation mapping, the most costly method, was
used to map the greatest number of vegetation classes of all the methods. A
series of photo-interpreted vegetation maps documents changes in community
location and area. Relative percent cover changes can be quantified using
either a planimeter or overlaying maps with a grid and counting cells of
each vegetation type.
Computer-assisted mapping, the third most expensive method, generates
computer quantified vegetation maps. If vegetation classes can be correctly
defined and identified during training set selection, this is the most
reliable (consistent) of the quantitative cover estimation methods. This
method also demonstrates changes in community location. The color film
products which can be made using this method provide the best visual
documentation of change of any of the nine methods. If the vegetation
classes being mapping are reliably identified, this method definitely
warrants its high cost.
In this particular study where the user of the information collected
and analyzed is the Environmental Protection Agency, it is recommended that
the following combination of methods are the most efficient, sensitive and
reliable Co study wetland vegetation change.
Association analysis (using presence-absence data) should be used to
analyze data collected over time to demonstrate the manner in which
vegetation communities are changing, year to year. If the equipment and
package of programs necessary to use computer assisted mapping are
available, this method should be used to quantify change in community area
and to demonstrate change in community location. If the computer assisted
mapping method can not be used, airphoto interpreted vegetation mapping to
supply community locations and area information is recommended.
-------
ABSTRACT
This study was part of a large Environmental Protection Agency study
designed to monitor the impacts of the construction and operation of a coal-
fired power plant located near Portage, Wisconsin. This study demonstrated
and evaluated nine methods used to document the deterioration over 3-yr
period of a large wetland located within the power plant site. Four of the
nine methods (diversity index, subjective classification, association
analysis, and structure analysis) used ground sampling data. Two methods,
airphoto monitoring and disturbance mapping, used airphoto data only, while
airphoto grid analysis and airphoto interpreted vegetation mapping used
airphoto and ground sampling data. The ninth method, computer assisted
mapping, used only airphoto data but relied heavily on the analyst's "on-
the-ground" knowledge of the area.
These methods were evaluated on the basis of whether they monitor
vegetation change at a species or community level, whether or not they
monitor community change in terms of area and/or location, and whether or
not they provide information about community trends. These methods also
were evaluated for time, cost, sensitivity and reliability.
Of the nine methods, only the diversity index, a ground sampling data
method, documented species change with time. The other three ground
sampling data classification methods dealt with community change, and
documented that change by analysis of species appearance and disappearance.
All methods, other than the diversity index, documented changes in
vegetation community area and/or location. The ground sampling methods
showed point locations of communities which were mapped and quantified
changes in relative percentages of sampling points classified as a
particular class. Subjective classification is the most sensitive and the
most expensive of the three ground sampling methods; vegetation structure
analysis is least sensitive and least expensive while the association
analysis handles large quantities of data better than either of the other
two.
Airphoto methods can better document community area and location
changes than can ground sampling data methods. Airphoto monitoring, a
purely descriptive method, is the least effective of the airphoto methods,
offering only a series of airphotos as a record of change. It does offer
the advantage of not requiring any ground verification nor does disturbance
mapping. Disturbance mapping, a low cost method, uses a series of maps to
record change. Its greatest disadvantage is that it delineates only three
iv
-------
FOREWORD
The U.S. Environmental Protection Agency was established as a focus of
scientific, governmental, and public efforts to improve the quality of the
environment. These efforts require expansion of our understanding of the
mechanisms that govern environmental changes and in particular those changes
that result from our own manimpulations of the environment. One specific
thrust of these efforts must be the continous development of more effective
and more efficient methods for analyzing the environment and the changes
occurring in it.
One such project, which the Environmental Protection Agency is
supporting through its Environmental Research Laboratory in Duluth,
Minnesota, is the study "The Impacts of Coal-Fired Power Plants on the
Environment." The Columbia Generating Station, a coal-fired power plant
near Portage, Wis., has been the focus of all field observations. This
interdisciplinary study, involving investigators and experiments from many
academic departments at the University of Wisconsin, is being carried out by
the Water Resources Center and Institute for Environmental Studies at the
University of Wisconsin-Madison. Several utilities and state agencies are
cooperating in the study: Wisconsin Power and Light Company, Madison Gas
and Electric Company, Wisconsin Public Service Corporation, Wisconsin Public
Service Commission, and Wisconsin Department of Natural Resources.
This investigation demonstrated and evaluated nine methods of
documenting the deterioration of the wetland within the power plant site
over a period of 3 yr. Four methods—diversity index, subjective
classification, association analysis and structure analysis—used ground
sampling methods. Two methods—airphoto monitoring and disturbance mapping
—used airphoto data only, while airphoto grid analysis and airphoto
interpreted vegetation mapping used airphoto and ground sampling data.
Computer assisted mapping used only airphoto data but relied heavily on the
analyst's "on-the-ground" knowledge.
Norbert A. Jaworski
Director
Environmental Research Laboratory-Duluth
Duluth, Minnesota
iii
-------
Table Page
19 List of Airphotos Used to Generate Airphoto Interpreted
Vegetation Maps 75
20 Vegetation Classes Discernible on Photo Interpreted
Vegetation Maps 76
21 Ease with which Vegetation Classes Could Be Identified
on Airphotos 77
22 Relative Percentage Cover of Each Vegetation Class Defined
Using Airphoto Interpreted Vegetation Mapping 78
23 Vegetation Classes Used to Summarize Airphoto Interpreted
Vegetation Maps 80
24 Comparison of Percent Cover Results Using Airphoto Grid
Analysis and Airphoto Interpreted Vegetation Mapping 91
25 Vegetation Classes Derived Using Computer Assisted Mapping 92
26 Vegetation Classes Used to Summarize Computer Assisted
Mapping 125
27 Color and Exposure Values Assigned to Vegetation Classes
Identified Using Computer Assisted Mapping 125
28 Vegetation Classes Identified Using the Eight Classification
and Mapping Methods 130
29 Expertise Needed to Use Each Method 131
30 Capital Equipment Costs for Photographic Data Collection 131
31 Data Collection Materials Cost for One Data Set 132
32 Capital Equipment Used with the Nine Methods 132
33 Capital Equipment Costs for Data Analysis 132
34 Data Processing Materials Costs for One Data Set 133
35 Time For Data Collection and Analysis for a 33.5 Ha Site 134
36 Time for Collection and Analysis for One Data Set Using
Each of the Nine Methods 135
37 A Breakdown of Data Collection and Processing Costs and Labor
Costs for One Set and Four Sets of Data for a 33.5 Hectare
Study Site 136
xi
-------
Page
Data Collection and Processing and Labor Costs for One Data
Set and Four Sets for a 33.5 Ha Site 139
39 Combined Time-Cost (Efficiency) Rating for Each Method 139
40 Method Sensitivity Rating Based on Vegetation Classes and
Data Type 140
41 Method Reliability Based on Data Collection Repeatability,
Data Analyst Interaction, and Quantitative or Qualitative
Results 141
A-l Species Code List 160
B-l Station and Transect Numbers 162
C-l Diversity Index 163
D-l Subjective Classification 165
E-l Association Analysis 166
F-l Airphoto Interpretation Keys For Airphoto Grid Assessment 167
G-l Airphoto Interpretation Keys For Vegetation Mapping 170
xii
-------
SECTION 1
INTRODUCTION
The Columbia Generating Station study began in 1971 when three
Wisconsin utilities (Wisconsin Power and Light Company, Madison Gas and
Electric Company, and Wisconsin Public Service Corporation) asked the
University of Wisconsin-Madison to study the impact of construction and
operation of a new coal-fired generating station at Portage, Wisconsin. It
was hoped that the study would result in newer, less expensive methods for
measuring environmental change.
The three utilities funded the study at $45,000/yr from January 1971 to
25 June 1975. In July 1975, the Environmental Protection Agency became the
funding agency and increased funding approximately ten-fold for the period
July 1975 through July 1978. The study was expanded and reorganized. In
addition to monitoring and documenting the effects of the generating station
on the environment, the investigators now attempted to understand the
interactions among these changes so that power plant impacts could be
predicted and perhaps manipulated in the future. The entire study was
divided into two major components—one to study terrestrial systems and one
to study aquatic systems. The remote sensing group was made part of the
aquatic systems division.
The Columbia Generating Station is in west central Columbia County,
along the Wisconsin River floodplain, 4 miles (6.4 km) south of Portage,
Wisconsin. The site covers 1,100 ha (Figure 1) and includes the generating
units and coal storage area (110 ha), the cooling lake (200 ha) and the
ashpit (30 ha).
Before plant construction the site consisted of extensive marsh/sedge
meadow, floodplain forest, and a few low semi-wooded knolls (Figure 2). The
marsh/sedge meadow community floods to varying depths during the spring and
occasionally during the fall. The study site is wettest in the spring and
becomes progressively drier through summer. Although the area was comprised
primarily of sedges and grasses, it included small areas of open water with
emergent vegetation as well as pockets of shrub carr and alder thicket.
The marsh/sedge meadow soil is a peat mat 3 to 4 feet thick overlying
sand. The meadow is bounded on the north by Duck Creek and on the south by
Rocky Run Creek. Before plant construction there was no apparent surface
flow between these streams across the sedge meadow. However, ground water
moved through the meadow toward the river from a large area to the southeast
(Figure 3).
-------
f~)
I South (4? ^..x
/ Knoll \
Figure 1. Columbia Generating Station site showing location of ashpit,
generation station, coal pile, cooling lake, and marsh/sedge
meadow.
-------
u>
£.--Uj EMERGENT AQUATICS
5?'-j
isJ SUBHEKCENT AQUATICS
SivJ SAND DUNE
^77}
;.:, E.V. -J OLD FIELD
'LAND HARDWOODS
.JJ LOWLAND
^a
S ^^J MARS"/SEDGE MEADOW
;•;• r^'s^i
b »rj TAMARACK
I 1 RESIDE,
T
CROPLAND
Figure 2. Dames and Moore
vegetation
map. of site drawn in 1972.
-------
-COAL'PILE
OGLING LAKE
. •Milwaukee
MADISON }
Figure 3. Groundwater contours of the area (from Andrews, 1977),
-------
IMPACTS OF PLANT CONSTRUCTION AND OPERATION ON THE STUDY SITE
Major impact commenced in 1971 when much of the site was cleared and
construction on the cooling lake was started.
The power plant generating units and coal storage area were built on
high ground previously occupied by a mix of dry oak forest, prairie, and old
field vegetation. However, the cooling pond and a portion of the ashpit
were built directly over the sedge meadow. This required construction of an
earthfill dike (April 1971) to contain the pond and a drainage ditch on the
southeast side of the marsh (September 1973) to, divert natural ground and
surface flow around the cooling lake. Figure 4 shows ground-water movement
through the study site before and after construction of the cooling lake and
drainage ditch. Construction on the ashpit also was begun in 1973 (Figure
5).
An attempt was made to fill the cooling lake in June 1974 but it leaked
badly. Therefore it was drained and a clay sealer, bentonite, was sprayed
on the bottom and sides. The pond was refilled in January 1975 and two
months later Unit I went on line.
Most severe ground-water temperature dislocations occur within 100 m of
the dike (Figure 6). On 28 October 1976 ground-water temperatures at a
depth of 1.5 m beneath the surface ranged from greater than 23°C to less
than 10°C. Six months later, on 26 April 1977 (before Unit II went on line)
temperatures at a depth of 1.5 m ranged from 13°C to 5°C, showing that the
ground water near the dike was becoming cooler rather than warmer at this
time of the year. By 25 July 1977 ground-water temperatures again ranged
from 10°C to more than 23°C. With two units in operation, the area of
ground water elevated to 20°C is larger than with only one unit in
operation.
After construction of the cooling lake, the marsh/sedge meadow
vegetation responded dramatically to three impacts—elimination of seasonal
water level fluctuations (Figure 7), increase of ground-water flow, and
year-round thermal loading of the ground water. The increased ground-water
flow is rapidly eroding the peat mat. In addition, thermal loading is
causing some vegetation to die so that its anchoring roots no longer produce
new peat mat. This augments the erosiveness of the ground water. Large
channels of open water are now visible through the marsh/sedge meadow where
originally there were none.
Plant species have reacted in various ways. According to Bedford
(1977), the major responses are: 1) Species dying out due to increased
water level; 2) species increasing due to increased water level; 3) species
dying out due to out-of-synch water temperatures; 4) species increasing due
to increased water temperatures; 5) species dying out due to mechanical
water action which removes protective litter, particularly important for the
-------
River
ai
River ^-Study Area
Cooling Lake
500
1000
METERS
1500
2000
Figure 4. Groundwater movement through the site before and after
construction of cooling lake (from Andrews, 1977).
-------
M
JL
J.
Q.
.N
Jl
J.
-E.
M
A
-N,
J2
A.
M.
J2.
Ji
D
M.
A
.
.0.
Ji
Jl
1971
WesLDike Construction Begins
lflZ2_
Main Building Frame Topped Off
1973
Stack Construction Beains
Finish Stank Construction /Settlin Basin Started
Drainage Ditch Dredged
1974
Intake Channel Started
Filling of Cooling Lake Started
Transmission Line Right-nf-Way Started
M
A
M
J.
±
A
.S.
i'
975
Qolinn I akp Refilled / Boiler Testing Started
Jnit 1 On Line
Unit II Building Started
it II Stack Construction Started
Cooling Tower Construction Beains
1977
Stack Construction Finished
1978
ooling. Tower Completed
Uniti]
Main Building Topped Off
Figure 5. Time line of construction events at the Columbia Generating Station.
-------
October 28, 1976
I 15'
W A
April 26, 1977
r
.
to' -
1 . 1
July 25. 1977
20'-
15'-
100
0 meters
Figure 6. Groundwater isotherms at various depths and distances from the
cooling lake dike in October 1976 and January, April and July
1977. Note the depth and distance of the isotherms showing
warmest temperatures (from Andrews, 1977).
-------
M
ec
UJ
h-
Ul
S 238
UJ
ui
flC
UJ
237.5
cooling lake filled
JAN
1972
JAN
1973
JAN
1974
JAN
1975
JAN
1976
JAN
1977
Figure 7. Wetland water level fluctuations (from Andrews, 1977).
-------
over wintering of the next year's growth; and 6) weedy annual species
increasing in areas where peat and mud are exposed.
Vegetation responses to the cooling lake impacts were analyzed from
several different perspectives: Changes in community growth form and
associated bird life (Jaeger 1979); physiological causes of species'
declines or increases (Bedford 1977); and changes in species and community
occurrences with time.
The latter approach is the subject of this report. The study attempts
to 1) document community and species change from the ground and from the air
and 2) compare the efficiency, sensitivity and reliability of various
monitoring methods for detecting vegetation change.
Much of the literature describing environmental monitoring and impact
assessment techniques is directed at the needs of those preparing environ-
mental impact statements. Many theoretical papers discuss what the environ-
mental impact statement process should include (Bisset 1978, Sondheim
1978). Some address components that might be considered in a statement
(Leopold et al. 1971, Fischer and Davies 1973) and various ways of treating
them (Warner and Bromley 1974, McHarg 1969). A few state specifically what
data might be collected for terrestrial or aquatic studies (Johnson 1974).
Fewer still either suggest or evaluate data collection methods (Eberhardt
1976).
Traditionally, vegetation monitoring studies have been conducted on the
ground (Hough 1965, Bunce and Shaw 1973, Muller-Dombois and Ellenberg 1974,
and Smith et al. 1975). Now, however, remote sensing methods are used to
map vegetation (Cowardin and Myers 1971, Johnson 1974, Brown 1978, and
Gammon and Carter 1979), locate manmade linear features (roads, pipelines),
and evaluate terrain sensitivity (Dirschl and Dabbs 1972). While many
remote sensing papers emphasize the need for adequate ground verification,
few describe how verification should be accomplished (Enslin and Sullivan
1974). The techniques discussed here are a step in this direction.
Incorporating several classification methods to analyze ground sampling
and remote sensing data, this study demonstrates vegetation changes with
time. The methods chosen represent a variety of traditional and new
methods; together they provide many levels of information. Selection of the
various methods was based on their appropriateness for analyzing and
documenting vegetation changes specific to the Portage setting, where a
coal-fired power plant was constructed and operated in.a marsh/sedge meadow.
This study attempts 1) to document species and community change from
the ground and from the air and 2) to compare the efficiency, sensitivity
and reliability of the selected monitoring methods in detecting vegetation
change. Recommendations are made for effectively combining ground sampling
and airphoto data collection and analysis methods. These ground sampling
and airphoto methods were specifically selected to monitor change for an
Environmental Protection Agency study; they might also find use in
environmental impact statements, resource surveys, land-information systems,
10
-------
and litigation (Anderson and Wobber 1973, Fornes and Reimold 1973, Frazier
and Lee 1975, and Lillesand and Kiefer 1979).
Extensive ground sampling data were collected in the Portage wetland
from 1974-77. These data were analyzed to create: 1) A diversity index, 2)
a subjective classification, 3) an association analysis classification, and
4) a vegetation structure classification.
Numerous color and color infrared airphotos of the wetland were taken
over the same period of time. These photos were used as a data base to
record community change. Generalized disturbance maps were drafted from
them. Airphoto and ground sampIng data were used together to generate: 1)
relative percent cover changes, 2) airphoto interpreted vegetation maps, and
3) computer-assisted vegetation maps.
Specifically, the questions under examination in this report are:
1) Using these techniques, can species changes be detected with time?
2) Which techniques detect species changes most efficiently?
3) Using these techniques, can community area and location changes be
detected with time?
4) Which techniques detect community changes most efficiently?
5) Can trends in vegetation changes be documented with these
techniques?
6) Which techniques do so most efficiently?
11
-------
SECTION 2
CONCLUSIONS
This study was part of a large U.S. Environmental Protection Agency
study designed to monitor the impacts of the construction and operation of a
coal-fired power plant located near Portage, Wisconsin. The study
demonstrated and evaluated nine methods used to document the deterioration
of a large wetland located at the power plant site over a 3-yr period. Four
of the nine methods (diversity index, subjective classification, association
analysis, and structure analysis) used ground sampling data. Two methods,
airphoto monitoring and disturbance mapping, used airphoto data only, while
airphoto grid analysis and airphoto interpreted vegetation mapping used
airphoto and ground sampling data. The ninth method, computer-assisted
mapping, used only airphoto data but relied heavily on the analyst's on-the-
ground knowledge of the area.
These methods were evaluated on a basis of whether they monitor vegeta-
tional change at a species or community level, whether or not they monitor
community change in terras of area and/or location, and whether or not they
provide information about community trends. These methods were also
evaluated on the basis of time, cost, sensitivity and reliability.
Of the nine methods, only the diversity index, a ground sampling data
method, documented species change with time. The other three ground
sampling data classification methods dealt with community change and
documented that change by analysis of species appearance and disappearance.
All methods—other than the diversity index—documented changes in
vegetation community area, vegetation community location, or both. The
ground sampling methods showed point locations of communities which were
mapped. They also quantified changes in relative percent Of sampling points
classified as a particular class. Subjective classification is the most
sensitive and the most expensive of the three ground sampling methods;
vegetation structure analysis is least sensitive and least expensive while
the association analysis handles large quantities of data better than either
of the other two classification methods.
Airphoto methods can better document community area and location
changes than can ground sampling data methods. Airphoto monitoring, a
purely descriptive method, is the least effective of the airphoto methods,
offering only a series of airphotos as a record of change. It does offer
the advantage of not requiring any ground verification, a feature which it
shares with the disturbance mapping method. Disturbance mapping, a low cost
method, uses a series of maps to record change. Its greatest disadvantage
12
-------
is that it delineates only three classes—disturbed and undisturbed vegeta-
tion and open water. In defining only three classes, however, this method
creates an exceptionally understandable visual record of change taking
place.
Airphoto grid analysis, the second most costly method, records percent
cover of vegetaton classes on a cell basis. This method documents percent
changes in community area but does not directly document changes in
community location.
Airphoto interpreted vegetation mapping, the most costly method, was
used to map the greatest number of vegetation classes of all the methods. A
series of photo-interpreted vegetation maps documents changes in community
location and area. Relative percent cover changes can be quantified using
either a planimeter or overlaying maps with a grid and counting cells of
each vegetation type.
Computer-assisted mapping, the third most expensive method, generates
computer-quantified vegetation maps. If vegetation classes can be correctly
defined and identified during training set selection, this is the most
reliable (consistent) of the quantitative cover estimation methods. This
method also demonstrates changes in community location. The color film
products which can be made using this method provide the best visual
documentation of change of any of the nine methods. If the vegetation
classes being mapped are reliably identified, the film products and
quantified area cover information this method generates, definitely warrant
its high cost.
In this particular study where the user of the information collected
and analyzed is the U.S. Environmental Protection Agency, the following
combination of methods are probably the most efficient, sensitive and
reliable to study wetland vegetation change.
Association analysis (using presence-absence data) should be used to
analyze data collected over time to demonstrate the manner in which
vegetation communities are changing, year by year. If the equipment and
package of programs necessary to use computer-assisted mapping are
available, the analyst recommends using this method to quantify change in
community area and to demonstrate change in community location. If
computer-assisted mapping cannot be used, the analyst recommends using
airphoto-interpreted vegetation mapping to supply community location and
area information.
13
-------
SECTION 3
RECOMMENDATIONS
This study had several shortcomings. Perhaps the greatest is that
there is no "truth" against which to compare the results of the methods
being tested. No detailed vegetation map exists of the site prior to the
start of construction.
Another shortcoming was the attempt to do ground sampling over too
large an area which led to inconsistent ground sampling, from one year to
the next. Sampling should not have been done over such an extensive area
because the greatest amount of sampling time and effort went into hiking
between sampling stations instead of into the actual sampling.
Another shortcoming was the failure to place permanent surveyed air-
photo targets at selected sampling stations so there would be two or three
targets identifying each vegetation community. This was not done due to a
reluctance to install permanent markers which would kill the vegetation. In
view of the destruction which has taken place at the study site, target
damage to the site would have been inconsequential. Perhaps a more valid
reason for not placing targets was that the difficult terrain made placing
any kind of markers a monumental task. Targets are needed on airphotos to
use in vegetation class identification, particularly when using computer-
assisted mapping. Selecting training sets from target areas of known
vegetation is one way to assure the method's accuracy.
In approaching a similar situation in the future, documenting impact at
a study site without knowing where it would happen or the form it would
take, the procedure described in the next two paragraphs would most likely
provide the best results.
Permanent markers (targets) would be located at a selection of the
sampling stakes so that each community was targeted at least twice and
preferably three times. The location of these targets would be based on
surveying. Target size would be such that it could be identified on the
smallest scale airphotos used.
Data analysis could depend on use. The ground data would be classified
using either subjective classification or association analysis in order to
record and understand the changes exhibited by the vegetation. And the
various vegetation communities could be mapped to visually demonstrate
changes in area and location with time. The first choice of method to use
14
-------
for this would be computer-assisted mapping if sufficient funds were
available. With targeted communities, accurate classification would be
assured. Otherwise, airphoto interpreted vegetation mapping would be
recommended.
15
-------
SECTION 4
MATERIALS AND METHODS
Extensive ground sampling and airphoto data were collected between 1974
and 1977 in the study area (the 33.5 ha of marsh/sedge meadow west of the
cooling lake's west dike (Figure 8)). The data collection techniques that
were chosen represent a range of scales, costs and detail.
GROUND SAMPLING DATA COLLECTION
Ground sampling data were collected from 1974 to 1977 by members of the
Remote Sensing Group and the Wetlands Ecology Group. The 33.5 ha study site
was marked off into east-west transects spaced at 50 m intervals. Sampling
stations were established at 50 m intervals along these transects, forming a
grid. At each sampling station two circular 0.25 m quadrats were laid
out. Species and/or numbers of stems/species were recorded for each
quadrat. The size of the circular quadrats (0.25 m ) was selected because
the grasslike growth form of sedge and grass species made use of a larger
quadrat impractical. Wetland vegetation of this type is susceptible to
trampling. To minimize the effects of trampling, two quadrats were laid out
randomly within the four quadrants at each station.
Water temperature (°C) was measured at the surface and in the
vegetation rooting zone; water depth (cm) was measured at the level of solid
substrate below the rooting zone. Air temperature (°C) also was measured at
each sampling station. Water quality was assessed subjectively, by visual
estimate. The presence or absence of a floating mat was noted at each
sampling station as well as direction of water flow. Volume of water flow
was subjectively classified as absent, gentle, moderate or swift. Soil
substrate was classified according to six types; muck, open water,
consolidated peat, unconsolidated peat, sand or upland.
Table 1 provides an example of encoded data collected during fall
1976. The first four columns identify the year and month in which the data
were recorded. Columns five to nine identify transect, stake, and quadrat
numbers. Columns 10 to 26 record environmental parameter data (data used by
the Wetlands Ecology Group). Column 31 begins a series of six column
units: the first three columns identify the species, while the last three
record the number of .stems counted for that particular species. The species
codes are listed in Appendix A.
16
-------
Figure 8. Color infrared airphoto of study site on July 24, 1976.
17
-------
TABLE 1. EXAMPLE OF ENCODED GROUND SAMPLING DATA
AND SCHEME WITH WHICH THEY WERE CODED
77
77
77
77
77
77
77
77
77
430061
430064
430074
430072
430082
430084
430091
430094
15
14
12
12
12
12
10
10
430104110
Col limn number
1-2
3-4
5-6
7-8
9
10
11-12
13-14
15-16
17-18
19-20
21
22
23
24-25
26
31+
141634 4 016075028002
142266 4 030007008008
1209371 4 016075030013
1213581 4 016075030013
12 17 174 008017016050
121826 174 008001016050
1010521 4 016050131004030002
101452 4 016050030014
101649 4 016025030005
Information
Year
Month
Transect
Stake
Quadrat
Water quality
Water temperature at surface
Water temperature at base of plant
Water temperature in rooting zone
Water depth to solid substrate
Water depth to rooting zone
Floating mat; l=present, 2=absent
Direction of water flow; 0=absent, 1=N, 2=NE, 3=E, 4=SE,
5=S, 6=SW, 74J, 8=NW
Volume of water flow; 0=absent, l=gentle, 2=moderate,
3=swift
Air temperature
Soil substrate code; l=muck, 2=open water, 3=consolidated
peat, 4=unconsolidated peat, 5=sand, 6=upland
Three letter species code followed by three letter
species count
18
-------
Sampling took place each summer and fall from 1974 to 1977. However,
the number of stations sampled varied throughout the study. During summer
and fall 1974 species stem counts data were collected at every other station
at 100 m intervals. In 1975, species presence-absence data and stem counts
data were collected at every other station at 100 m intervals. In 1975,
species presence-absence data and stem count data were collected at
alternating stations. In the fall seasons of 1976 and 1977, judging stem
counts data more valuable than presence-absence data, the Remote Sensing and
Wetland Ecology Groups collected stem counts data at every station. In the
summers of 1976 and 1977 only a select number of stations were sampled by
the Wetlands Ecology Group. These variations in the sampling scheme reduced
the data actually used for analysis in this study. Analysis was limited to
records obtained from stations that were sampled in each of the fall dates
1974 to 1977. These 62 stations are located between transects 18 and 40 and
are listed in Appendix B.
AIRPHOTO DATA COLLECTION
Airphoto data were collected at more frequent intervals than the ground
sampling data. Airphotos were taken once a month during the growing season
and several times over the winter months. In the spring and fall of 1976,
airphotos were taken every few days in order to document the rate at which
the vegetation sprouted and later died back, as recorded on the airphotos.
In 1973 and 1974, Nikon cameras were used to take 35 mm color and color
infrared photos of the study site at several scales. Beginning in March
1973, two Hasselblad cameras were used to take 70 mm color and color
infrared airphotos using Kodak 2443 and 2448 film. Photos were usually
taken at scales of 1:38,200 (5,000 ft above mean terrain) and 1:19,100
(2,500 ft above mean terrain). Occasionally, photos were taken at scales of
1:76,400 (10,000 ft above mean terrain) and 1:19,100 (2,500 ft above mean
terrain). Occasionally, photos were taken at scales of 1:76,400 (10,000 ft
above mean terrain) and 1:11,500 (1,500 ft above mean terrain). Several
scales were tested to determine which was best suited to computer-assisted
mapping. Table 2 lists the imagery obtained for this study. .,
MONITORING VEGETATION USING GROUND SAMPLING DATA
Ground sampling data were analyzed to generate 1) a diversity index, 2)
a subjective classification, 3) association analyses, and 4) a vegetation
structure classification. These techniques represent a range of detail and
costs; they are described separately below.
19
-------
TABLE 2. COLUMBIA GENERATING STATION AIRPHOTOS, 1949 to 1977
Date
27 May 1949
7 Jan. 1967
13 May 1968
4 April 1970
24 May 1971
through
17 Aug. 1972
Autumn, 1971
4 April 1971
14 April 1971
13 May 1971
4 Aug. 1971
10 April 1972
11 April 1972
13 May 1972
4 June 1972
27 Sept. 1972
13 Nov. 1972
18 March 1973
14 June 1973
12 Oct. 1973
8 Nov. 1973
15 April 1974
16 April 1974
24 May 1974
4 June 1974
17 June 1974
31 July 1974
7 Aug. 1974
8 Aug. 1974
8 Aug. 1974
10 Oct. 1974
10 Oct. 1974
22 Nov. 1974
19 March 1975
19 March 1975
Film type
Black & White
Black & White
Black & White
Black & White
Black & White
Obliques
Color Infrared
Black & White
Black & White
Black & White
Black & White
Color
Infrared
Color
Infrared
Black & White
Color & Color IR
Black & White IR
Color IR
Color
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color
Color
Color IR
Color & Color IR
Color IR
Color IR
Color
Color IR
Color
Black & White
Color & Color IR
Color
Scale
1:20,000
1:15,480
1:20,000
1:6,000
1:7,900
1:6,600
1:7,900
1:1,200
1:120,000
1:24,000
1:12,000
1:24,000
1:7,900
1:120,000
1:7,900
1:32,300
1:30,500
1:29,200
1:60,900
1:60,900
1:12,200
1:24,400
1:17,000
1:17,000
1:12,200
1:12,200
1:120,000
1:24,400
1:54,800
1:24,400
1:54,800
1:18,300
1:22,200
1:24,400
1:24,000
1:19,100
1:38,200
1:76,400
Source
ASCSa
DOTb
ASCS
ALSTERC
Airpixd •
LLe
DNRe
DNR
DNR
DOT
UW8
UW
DNR
NASA"
DNR
UW
DNR
UW
UW
UW
UW
UW
UW
UW
UW
NASA
UW
UW
UW
UW
UW
DOT
DNR
DNR
Transparency
format
9"x9"
9"x9"
9"x9"
9"xl5"
9"x9"
35 mm
35 mm
9"x9"
9"x9"
9"x9"
35 mm
9"x9"
35 mm
35 mm
35 mm
35 mm
35 mm
35 mm
35 mm
35 mm
35 mm
35 mm
35 mm
35 mm
35 mm
9"x9"
70 mm
9"x9"
20
(continued)
-------
TABLE 2 (continued)
Date
19 March 1975
13 June 1975
10 July 1975
16 July 1975
17 Aug. 1975
26 Aug. 1975
25 Sept. 1975
6 Oct. 1975
5 Nov. 1975
7 Jan. 1976
1 March 1976
11 March 1976
20 March 1976
1 April 1976
8 April 1976
19 April 1976
4 May 1976
11 May 1976
21 May 1976
5 June 1976
17 June 1976
Film type
Color
Color
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color
Color
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Scale
1:76,400
1:38,200
1:76,400
1:76,400
1:76,400
1:76,400
1:38,200
1:76,400
1:38,200
1:76,400
1:19,100
1:38,200
1:11,500
1:38,200
1:19,100
1:11,500
Obliques
Obliques
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
Source
DNR
DNR
DNR
UW
UW
UW
UW
UW
UW
UW
UW
UW
UW
UW
UW
UW
UW
UW
UW
UW
UW
Transparency
format
9"x9"
9"vQ "
s AJ
/ 0 IOD1
70 min
70 mm
/ 0 niin
70 mm
/ fl TTITH
/ \J Uim
70 mm
70 mm
70 mm
35 mm
35 mm
70 mm
70 mm
70 mm
70 mm
70 mm
70 mm
70 mm
70 mm
(continued)
21
-------
TABLE 2 (continued)
24
3
24
26
7
1
26
25
11
15
Date
July 1976
Sept. 1976
Sept. 1976
Oct. 1976
Feb. 1977
March 1977
April 1977
June 1977
July 1977
Aug. 1977
3 Oct. 1977
Film type
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Color & Color IR
Scale
1:38,200
1:19,100
1:11,500
1:5,700
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:5,700
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:38,200
1:19,100
1:11,500
1:11,500
Source
UW
uw
UW
uw
uw
uw
uw
uw
uw
uw
uw
Transparency
format
70 mm
70 mm
70 mm
70 mm
70 mm
70 mm
70 mm
70 mm
70 mm
70 mm
70 mm
aAgricultural Stabilization and Conservation Service (USDA).
Department of Transportation, State of Wisconsin.
GAlster & Associates, Madison, Wisconsin.
dAIRPIX, 2610 North Laramie, Chicago, Illinois.
^Landscapes Limited, Madison, Wisconsin (No longer in operation).
Department of Natural Resources, State of Wisconsin.
^University of Wisconsin, Institute for Environmental Studies,
Environmental Monitoring and Data Acquisition Group, Madison, Wisconsin.
National Aeronautics and Space Administration.
22
-------
Diversity Index
The diversity index (Table 3) gives information about yearly changes in
the numbers of species based upon field data collected over several years.
The total number of species found at all 62 sampling stations was summed for
each year of the study (see Appendix C). In 1975, for example, the total
number of species found at all 62 stations was 379. By 1977, the total had
dropped to 266, a loss of 30%.
TABLE 3. DIVERSITY INDEX, 1974-77
Year
1974
1975
1976
1977
Numbers of
species
379
357
296
266
Species loss/yr
(%)
5.8
17
10
Species losses reflect the impacts taking place each year. In 1974 and
1975 the cooling lake was filled and peat mat erosion had begun due to the
increased ground-water flow. Species loss reached 22 species or 5.8% by the
time Unit I went on line in March 1975. At that time water moving from the
cooling lake into the marsh/sedge meadow began carrying a thermal load which
speeded up wetland destruction. This is reflected in the 61 species (17%)
lost that year. By 1976 and 1977 many of the species most sensitive to
disturbance had already disappeared. Species loss for that year dropped
back to 30 species or 10% of the original total.
The diversity index also provided a means to record the total number of
occurrences of weedy species appearing each year. Species defined as weedy
were the annuals Bidena aermua, Pilea pumila, and Lernna minor*. If a
sampling quadrat was empty of all vegetation, indicating it was very badly
disturbed, a value of three was recorded for it. Each weedy species was
assigned a value of one. When totals were obtained by this method, weedy
species were shown to increase from 17 in 1974 to 69 in 1977, a 405%
increase.
Subjective Classification
In order to demonstrate community change station by station, the data
collected at each of the 62 sampling stations were assigned a vegetation
23
-------
classification according to species present, 1974-77. Most of the
vegetation classes used were defined subjectively by Bedford (1977).
1) Carex striata prominent; mixed with other sedges, grasses, forbs
and ferns. Carex lacustris, Calamagrostis aanadensis Spartina
peatinata, and Sagi.tta.ria latifolia locally abundant.
2) Carex laoustris prominent, with Carex striota, Calamagrostis
oanadensis, and Sagittaria latifolia locally abundant.
3) Mixed Transition zone between sedge meadow and emergent aquatics;
Calamagrostis oanadensis, Sagittaria latifolia, Carex rostrata,
Carex laaustris locally abundant.
4) Locally prominent emergent aquatics: Typha latifolia, Carex
laoustris, Sparganiwn euryoarpum, Scirpus fluviatilis, Aaorus
calamus, Sagittaria latifolia, Carex rostrata.
5) Spiraea alba prominent in the shrub overstory; understory similar
to Carex striata areas.
6) Shrubs (Salix spp. and Cornus stolonifera) and lowland trees
(Populus tremuloides, Ulmus amerioana, Salix nigra, Acer
saoaharinum).
In order for a sampling station to be classified as one of these six
classes, those species listed as prominent had to make up 60% of the stems
in the quadrat. .
Since the study area changed in composition quite rapidly, several
classes had to be added. The first stage of change at the study site
required the addition of four Degraded classes:
7) Degraded Carex striota
8) Degraded Carex lacustris
9) Degraded Transition
10) Degraded Emergents
Stations placed in any of these Degraded classes had enough of the
prominent species to identify them as a particular community but lacked
accompanying species and adequate numbers of all species. All stations
classified as degraded had high Lemna minor (duckweed) counts indicating
open, still water and little shading vegetation.
Further damage, resulting in complete erosion of the peat mat, required
the addition of two more classes:
24
-------
11) Weedy annuals prominent; areas of floating peat exposed to the air
and often covered with Bidens aernus and/or Pilea pumila seedings
12) Open water areas where both the vegetation and peat mat had b.een so
thoroughly destroyed that only open water remained.
The subjective classification demonstrated changes with time in terms
of the changing numbers of stations assigned to each vegetation class (see
Appendix D). This method also made it possible to correlate vegetational
changes with changes in water temperature, volume of water flow, and peat
mat erosion.
Table 4 lists the number and percent of the total number of sampling
stations subjectively assigned to each vegetation class over a 3-yr
period. In 1974, 47% of the sampling stations were classified as sedges
(Carex striata and Carex laaustris classes). By 1976 this number had
decreased to 26% and dropped to 16% in 1977. The Carex laaustris community
almost disappeared entirely between the summers of 1974 and 1976. Carex
laaustris is a clone forming species. When the cooling lake was filled
(1974 and 1975), upwelling and greatly increased surface flow in the
marsh/sedge meadow resulted in exposure and mechanical damage to the next
year's shoots of Carex laaustris. Flowing water carried away detritus,
which normally acts as a protective mulch for young sedge shoots, causing
peat mat erosion. The Carex striata community survived better initially
because its tussock growth form kept the next year's shoots above water
during the winter. However, Carex striata is being destroyed more slowly by
erosion eating away the peat mat where the individual tussocks are anchored.
In 1974, 19% of the sampling stations were classified as Transition.
By 1976 and 1977, only 6% of these stations could be classified
Transition. The Emergents community decreased only slightly in numbers from
17% in 1974 to 13% in 1977. The Emergents class, which grows in deeper
water, has been able to spread into new habitat with the deeper water
condition and has therefore shown smaller losses than the Transition class.
Stations classified Degraded, Weedy Annual or Open Water increased from
1.5% in 1974 to 11% in 1975 and to 37% in 1977. This again indicates that
the greatest damage took place in 1975 and 1976; the first year of thermal
loading. Sampling stations classified Open Water increased from none in
1974 to 14% in 1977. Stations classified as Spiraea or Shrubs remained
constant between 1974 and 1977, reflecting their slower response to changing
conditions.
Table 5, which summarizes the reclassification of the sampling stations
using subjective classification, demonstrates the fragility of each
community. Over half of the stations originally classified as Carex
laaustris, Transition, and Emergents evolved in 3 yrs to Degraded, Weedy
Annuals or Open Water classifications. The Carex laaustris community
appears the most fragile since only one station out of an initial 15 (6,7%)
retained that classification. In the Transition community, one station out
25
-------
TABLE 4. SUMMARY OF SUBJECTIVE CLASSIFICATION
1974
Vegetation class
Carex striota
Degraded Carex stricta
Carex lacustris
Degraded Carex laeustris
Transition
Degraded Transition
Emergent s
Degraded Emergents
Spi raea
Shrub
Weedy Annual
Open Water
No. of
stations
15
14
12
11
3
6
1
Relative
%
24
23
19
17
4.8
9.7
1.6
1975
No. of
stations
16
1
9
1
9
2
12
1
3
6
1
1
Relative
%
26
1.6
14
1.6
14
3.2
19
1.6
4.8
9.7
1.6
1.6
1976
No. of
stations
14
4
2
6
4
4
11
3
3
6
2
3
Relative
%
23
6.5
3.2
9.7
6.5
6.5
18
4.8
4.8
9.7
3.2
4.8
1977
No. of
stations
9
10
1
3
2
4
8
6
3
5
2
9
Relative
y
c'°
14.5
16.1
1.6
4.8
3.2
6.5
12.9
9.6
4.8
8.1
3.2
14.5
-------
TABLE 5. RECLASSIFICATION OF GROUND SAMPLING DATA
BETWEEN 1974 AND 1977 USING SUBJECTIVE CLASSIFICATION
No. of
stations
Community in 1974 1974
Carex laaustris 15
Carex striata 15
Transition 12
Emergents 11
Shrubs 6
Spiraea 3
No. of
stations
1977
1
4
3
2
3
1
1
5
9
1
1
5
4
2
2
6
3
5
1
3
Relative
%
6.7
26.7
20.0
13.3
20.0
6.7
6.7
33.3
60.0
6.7
8.3
42
33
17
18
55
27
83
17
100
Community in 1977
Carex laaustris
Open Water
Degraded C. lacustris
Weedy Annuals
Carex strieta
Degraded C. strieta
Emergents
Carex striata
Degraded C. striata
Transition
Transition
Emergents
Degraded Transition
Open Water
Emergents
Degraded Emergents
Open Water
Shrubs
Carex strieta
Spiraea
27
-------
of 12 (8.3%) retained the Transition classification whereas five (42%)
changed to an Emergents classification and four (33%) to a Degraded
Transition classification. The Emergents community showed the same trend
toward reclassification as either Degraded or Open Water.
The subjective classification changes reflect a successional trend
toward deeper water species, the result of deepening water levels in the
marsh and continuing peat mat erosion. The Carex lacustris community which
grew close to the cooling lake dike (in the area of greatest upwelling)
suffered the greatest damage, so that very little Carex lacustris remains.
As erosion has continued and extended diagonally through the center of the
study area, the deeper water Transition and Emergents communities also have
suffered severe losses.
The Carex stricta community has survived better than the Carex
lacustris community because of its tussock building character. Nonetheless,
five of the 15 stations classified Carex stricta in 1974 were reclassified
as Degraded Carex stricta by 1977. Least fragile were the shrubby
communities since they do not respond as quickly to change as the grasses
and forbs•
A chi square test for two independent samples (Siegal 1956) was used to
compare classification results year by year (Table 6). The chi square test
TABLE 6. RESULTS OF CHI SQUARE TEST ON CLASSIFICATION RESULTS
Type and year of
Classification
Not
significant
0.10 0.05 0.02 0.001
Subjective classification
1974 to 1975
1975 to 1976
1976 to 1977
1974 and 1977
1974, 1975, 1976, 1977
Association analysis
1974 to 1975
1975 to 1976
1976 to!977
1974 and 1977
1974, 1975, 1976, 1977
Vegetation structure analysis
1974 to 1975
1975 to 1976
1976 to 1977
1974 and 1977
1974, 1975, 1976, 1977
x
X
X
X
X
X
X
X
28
-------
for two independent samples can be performed if two criteria are met:
1) expected values are > 5.0; and 2) expected values of < 5.0 but > 1.0 are
permitted in only 20% of the cases being considered in the test.
To meet these criteria, vegetation classes had to be combined. The
vegetation classes used to compute chi square were Sedges (Carex stricta and
C. lacustris), Degraded Sedges, Transition, Emergents, Shrubs, Other
Disturbance (which takes in all Degraded categories other than Degraded
Sedges), Open Water and Weedy Annuals.
The chi square tests indicate that the change that took place between
1975 and 1976, as recorded by the subjective classification, is significant
at the 0.05 level. A chi square comparison of the 1974 and 1977
classifications is significant at the 0.001 level; so is the chi square test
performed on the classification results for the 4 yr.
Figure 9 shows the spatial locations of the 62 study site sampling
stations and their subjective classifications for 1974, 1975, 1976, and
1977. Over these years, the northern section of the study site became badly
disturbed and changed from a Sedges classification to a Degraded Sedges or
Open Water classification. The central section of the study area, which was
classified Transition in 1974, became an Emergents area. In the southern
section of the study site, stations initially classified as Sedges were
reclassified as Degraded Sedges, Weedy Annuals, or Open Water. Only the
southernmost stations escaped destruction.
Association Analysis
Association analysis (Williams and Lambert 1959, 1960) was used to
generate an objective classification of field data covering a 3-yr period.
This association analysis method differs from a subjective
classification in that the classification of the data from each station is
done using a computer program with set classification criteria. The user
selects the method of classification to be used and various options within
that method but the method is carried out by a computer program.
Consequently the method is more consistent, although not necessarily as
exact as subjective classification.
The association analysis was done using program DIVIDE found in the
Clustan Manual (Wishart 1970), a computer package of several association
analysis methods available at the Madison Area Computing Center.
To use association analysis, one cluster, containing data from all 62
sampling stations, is divided into two clusters of maximum species
dissimilarity. Dissimilarity is based on whether each sample contains a
particular species or not. The species used to divide the cluster is one
that will create the greatest species dissimilarity in the resulting two
clusters, using as a dissimilarity coefficient, sum chi square.
29
-------
1974
1975
1976
& Weedy Annuals
O Shrubs
a Transition
O Emergents
• Open Water
o Carex Stricta
x Carex Lacustns
o Degraded Species
1977
Figure 9. Subjective classification of ground sampling data
in 1974, 1975, 1976 and 1977.
30
-------
Once the first cluster is divided, one of the two new groups is termed
positive. The positive group includes all sampling stations that contain
the species used to create the original division. The other group is termed
negative since the dividing species is absent from it. Division then
continues along the positive branch until a statistically optimum number of
clusters is reached. In this case, the optimum was 20 clusters. Division
then takes place along the negative branch until a complete hierarchic
division is obtained. A dendrogram (Figure 10) is a convenient way to show
the species used to make the division in the analysis.
Program DIVIDE (an association analysis method) uses only binary data
(presence-absence data expressed as 1 or 0 respectively). ("M" binary
attributes are measured for population of "N" objects.) The ground sampling
data for each year, 1974 to 1977, consisted of 62 objects (in this case
sampling stations) while 34 attributes (selected species) were considered.
The species selected for this analysis were those believed to be visually
dominant in terms of height, cover and/or number, and responsible for the
patterns seen on airphotos of the study site. The cost of this method
depends on the size of the matrix created by multiplying attributes by
objects. Since these 34 species are fairly common, enough information about
them existed to have them be useful. Table 7 lists the 34 species used in
the association analysis.
In order to compare the results of the association analysis with the
subjective classification, the same vegetation classes (Carex stricta,
Transition, etc.) were used for both analyses. For the purposes of this
comparison, an additional class, Emergents-Open, was added to the
association analysis since these two communities often were not separated
using this method. Table 8 gives the results of the association analysis.
As with the subjective classification, the association analysis is repeated
for each year's data. The changes in clusters reflect the vegetation
changes in the wetland (see Appendix E).
Using association analysis, stations classified as Carex etricta
decreased from 27% of the total in 1974 to 9.7% in 1977 while stations
classified as Carex lacustris decreased from 32% in 1974 to none in 1977
(Table 8). Stations classified as Sedges decreased from 60% of the total in
1974 to 9.7% in 1977. Stations classified as Transition decreased from 20%
of the total to 4.8% in 1977. The number of stations classified as
Emergents increased slightly from 9.7% to 13% for the same period of time.
Stations classified as Degraded increased from zero in 1974 to 64% in
1977. The number of stations classified as shrubby remained stable while
stations classified Open Water and Degraded increased from zero in 1974 to
64% in 1977.
Table 9 shows the 1974 classification of the 62 sampling stations and
how those stations were reclassifled in 1977. Forty percent of the stations
originally classified Carex lacuetris in 1974 were reclassified as Degraded
Carex etriata in 1977. Although four (24%) of the stations classified Carex
stricta in 1974 retained this classification in 1977, 58% were reclassified
as Degraded Carex striata. Two-thirds of the stations classified Emergents
31
-------
K>
-23
Acorus calamus
-22 I i
Eleocharis acicularis
-21
Carex stricta
=J
Calamagrostis canadensis I
_2 "[ ,-2 Carex aquatilis
-13
Eupatorium maculatum
-1 I O
Carex lacustrls
-4
«4
Dryoptens thelypteris
-3 I «
Lemna minor
• 39
Iris shrevii
-38
-35
Spiraea alba
Carex stricta
-28
_ Carex rostrata
Sagittaria latifolia
Calamagnostis canadensis
-1 I »
Carex rostrata
-5 I .3 -
Typhalatifolia
-15 I
Rumex orbiculatus
-12
Carex lacustris
-2
Figure 10. Dendrogram of ground sampling data by association analysis
in the fall of 1976.
-------
TABLE 7. THIRTY-FOUR VISUALLY DOMINANT SPECIES
(ATTRIBUTES) USED IN THE ASSOCIATION ANALYSIS
1. Aaorus calamus
2. Calamagrostis aanadensis
3. Carex aquatilis
4. Carex etriata group
5. Carex hay den-Li
6. Carex lacustrie
7. Carex lasioaarpa
8. Carex rostrata
9. Dryopteris thelypteris
10. Eupatorium maculatum
11. Eupatoriwn perfoliatum
12. Helianthus grossesserratus
13. Iris shrevei
14. Leersia oryzoides
15. Lemna minor
16. Onoclea sensibilis
17. Polygonum coccineum and Polygonum natans
18. Rumex orbiculatus
19. Sagittaria latifolia
20. Scirpus cyperinue
21. Scirpus fluviatilis
22. Sci rpwe ua Z t
-------
TABLE 8. SUMMARY OF ASSOCIATION ANALYSIS RESULTS
OJ
1974
Vegetation class
Carex stria to.
Degraded Carex striota
Carex lacustris
Degraded Carex lacustris
Transition
Degraded Transition
Emergents
Degraded Emergents
Spiraea
Shrubs
Weedy Annuals
Open Water
No. of
stations
17
20
12
6
3
4
Relative
%
27
32
19
9.7
4.8
6.5
1975
No. of
stations
19
9
15
13
2
4
Relative
%
31
14
24
21
3.2
6.5
1976
No. of
stations
14
10
2
11
14
2
4
1
4
Relative
%
23
16
3.7
18
23
3.2
6.5
1.6
6.5
1977
No. of
stations
6
18
3
4
8
13
1
4
5
Relative
%
9.7
29
4.8
6.5
13
21
1.6
6.5
8.0
alncludes Degraded Emergents and Open Water.
these two classes.
Association analysis did not successfully differentiate
-------
TABLE 9. RECLASSIFICATION OF GROUND SAMPLING DATA
USING ASSOCIATION ANALYSIS
No. of
Community In 1974 stations
Carex lacustris 20
Carex stricta
Transition 12
Emergents 6
Shrub 4
Spiraea 3
No. of
stations
0
8
3
3
2
2
1
1
4
10
1
1
1
1
1
1
4
2
4
2
2
2
1
Relative
percent
0.0
40
15
15
10
10
5.0
5.0
23
58
5.9
5.9
5.9
8.3
8.3
50
33
33
67
50
50
67
33
Community in 1977
Carex lacustris
Degraded Carex stricta
Emergent s-open
Degraded transition
Transition
Weedy annuals
Emergents
Degraded transition
Carex stricta
Degraded Carex stricta
Spiraea
Weedy annuals
Shrub
Transition
Degraded transition
Emergent s-open
Emergents
Emergents
Emergents-open
Shrubs
Emergents-open
Carex etriata
Spiraea
in 1974 were reclassified as Emergents-Open in 1977 and 50% of the stations
classified as shrubby in 1974 were reclassified as Emergents-Open Water in
1977.
The association analysis method of classification shows (like the
subjective classification) that the Carex lacustris, Transition, and
Emergents communities are more fragile than the Carex stricta and shrubby
communities.
Figure 11 shows spatially the classification changes, based on the
association analysis, of the 62 sampling stations between 1974 and 1977.
35
-------
1974
A Weedy Annuals
O Shrubs
a Transition
O Emergents
• Open Water
a CarexStricta
x Carex Lacustris
Degraded Species
Figure 11. Association analyses of the ground sampling data
in 1974, 1975, 1976 and 1977.
36
-------
Results show the same trends as those obtained with the subjective
classification. Areas classified as sedges in the northern and central
areas of the site evolved to either a Degraded Sedges or Open Water
classifcation by 1977. The large central area, originally classified
Transition, converted to emergents and open water. The area along the west
edge of the study site containing emergents expanded over 3 yrs while the
sedge area in the southern portion of the site evolved to a Degraded Sedges
and Degraded Transition classification.
Chi square tests for two independent samples were done to compare
classification results for association analysis year by year. The
vegetation changes were significant except for the year spanning 1974 and
1975 (Table 6).
Vegetation Structure Analysis
Another way to demonstrate change in the study area was to document
changes in vegetative structure with time. Ground sampling data were
assigned to five categories of vegetative structure:
1) Grasslike: includes fine textured sedges, grasses and forbs, 2 to 4
ft tall, viewed from the air, this type of vegetation growth creates
a continuous cover with no apparent gaps.
2) Tall^-Coarse: includes species 4 to 8 ft tall with 50:50 or greater
ratios of water to vegetation. This type of vegetation includes
cattails and bulrushes and creates a coarse texture on airphotos.
3) Grasslike-Tall: includes sedge and grass species together with the
emergent species described under Tall-Coarse. Vegetation classified
as Grasslike-Tall displays more interspersion than vegetation
classified as Grasslike.
4) Shrubby: includes predominantly shrubby species mixed with small,
lowland trees.
5) Open: describes areas containing at least 75% water. When it does
occur, rooted vegetation is very sparse. Lerrtna minor is included in
Open.
Table 10 shows that stations classified Grasslike decreased over 3 yrs
from 47 to 14.5% of the total. Stations classified Grasslike-Tall increased
from 19% of the total to 24%, while stations classified Tall-Coarse
increased from 19% to 29%. Stations classified shrubby remained stable in
number. Stations classified Open increased from 1.6% of the total in 1974
to 19.4% in 1977.
Table 11 shows that more than two-thirds of the 29 stations classified
Grasslike in 1974 were reclassified Grasslike-Tall or Open in 1977. By 1977
92% of those stations classified Grasslike-Tall in 1974 were reclassified
37
-------
TABLE 10. SUMMARY OF VEGETATION STRUCTURE ANALYSIS RESULTS
OJ
00
1974
Structure class
Grasslike
Grasslike-Tall
Tall-Coarse
Shrubby
Open Water
No. of
stations
29
12
11
9
1
Relative
%
47
19
18
14
1.6
1975
No. of
stations
25
11
14
9
3
Relative
%
40
18
23
14
4.8
1976
No. of
stations
16
14
15
9
8
Relative
%
26
23
24
14
13
1977
No. of
stations
9
15
18
8
12
Relative
%
14.5
24.2
29
12.9
19.4
-------
Tall-Coarse or Open. Of the 12 stations originally classified Tall-Coarse
in 1974 73% retained that classification 4 yrs later, while 89% of the
stations classified Shrubby in 1974 retained that classification. Table 11
demonstrates that the Grasslike and Grasslike-Tall classes were sensitive to
impact and that the study area changed from a predominantly Grasslike and
Grasslike-Tall vegetation structure to a Tall-Coarse and Open Water
structure.
TABLE 11. RECLASSIFICATION OF GROUND SAMPLING DATA
USING VEGETATION STRUCTURE ANALYSIS
Community in 1974
Grasslike
Grasslike-Tall
Tall-Coarse
Shrubby
Open Water
No. of
stations
1974
29
12
11
9
•
1
No. of
stations
1977
8
14
1
6
1
9
2
8
3
8
1
1
Relative
28
48
3.4
20.3
8.3
75
17
73
27
89
11
100
Community in 1977
Grasslike
Grasslike-Tall
Tall-Coarse
Open Water
Grasslike-Tall
Tall-Coarse
Open Water
Tall-Coarse
Open
Shrubby
Grasslike
Open
A chi square test indicated that the changes recorded for 1974 to 1977
were significant at the 0.02 level. A chi square test performed on the data
from all 3 yrs was significant at the 0.001 level (Table 6).
Summary of Ground Sampling Data Methods
The diversity index demonstrated a 29% loss of species between 1974 and
1977. This technique detected changes in numbers of species present, year
by year. It should be emphasized that the diversity index was useful in
documenting changes at the Columbia study site where there was marked
overall loss of species present. The diversity index would not work in
disturbance situations where some species decreased but other species
39
-------
Invaded or increased, resulting overall in no change or an increase in the
number of species. The diversity index is the Only method of the nine
methods discussed which deals exclusively with species change.
Classification Methods—
Tables 4, 8, and 10 show changes in the numbers of sampling stations
assigned to each vegetation class over 4 yr. All herbaceous classes showed
losses in numbers, while the number of stations classified as Open Water
increased dramatically. Only the shrubby classes tended to remain
constant. Using the methods of subjective classification, association
analysis and vegetation structure classification, 28%, 15%, and 42%
respectively of the sampling stations retained their original classification
over 3 yr.
Tables 5, 9, and 11 show how the vegetation at specific stations
changed with time. Almost all stations classified as Carex lacustris in
1974 changed to a Degraded Carex stricta or Open-Emergents classification by
1977. (The Carex lacustris community has almost entirely disappeared from
the study site.)
The Carex stricta community survived somewhat better than the Carex
lacustris community. Twenty-three to 33% (depending on the classification
method) of those stations classified Carex stricta in 1974, still retained
that classification in 1977. Carex stricta stations were most frequently
reclassified as Degraded Carex stricta.
The Transition community is as fragile as the Carex lacustris
community. Only 8% of the stations classified Transition in 1974 retained
this classification in 1977. Most Transition stations were reclassified as
Emergents or Open Water by 1977.
The Emergents community appeared to only be slightly less fragile than
the Transition community. Using subjective classification, 82% of those
stations classified as Emergents in 1974 had gone to a Degraded Emergents or
Open Water classification by 1977. Using association analysis 67% of the
stations originally classified as Emergents had gone to an Emergents-Open
classification.
Only the Shrubs and Spiraea communities retained their species
integrity. All other communities changed, at the very least, to a degraded
classification. Substantial numbers of stations from all communities were
reclassified as Emergents, Degraded Emergents, or Open Water after 3 yr.
This reflects the widespread erosion of the peat mat and the consequent
widespread destruction of most of the marsh/sedge meadow species.
Change of classification of each of the 62 sampling stations was
compared for 1974 to 1975, 1974 to 1976, and 1974 to 1977 using each of the
three ground sampling classification methods. Table 12 shows the percentage
of sampling stations which changed their classification using the methods of
40
-------
subjective classification, association analysis, and vegetation structure
analysis.
TABLE 12. SAMPLING STATIONS CHANGING THEIR
CLASSIFICATION 1974-75, 1974-76, AND 1974-77
1974-75 1974-76 1974-77
Classification method % % %
Subjective classification
Association analysis
Vegetation structure analysis
19
47
19
52
64
48
60
79
55
The amount of change shown by each of the three classification methods
is partly an artifact of each method. The association analysis classi-
fication method shows the greatest amount of change in sampling station
classification. Association analysis defines vegetation classes using
presence-absence data based on one species at a time. According to this
method, if a sampling station lacks one species which is being used to form
a particular cluster even though it contains all the other prominent species
which define that community, it will not be identified as belonging to that
community. All sampling stations classified as a cluster using this
association method must be given the same classification, even though they
might be labeled differently using subjective classification. But
association analysis has several advantages. Once the type of analysis and
options are selected, the analysis is consistent and objective. And
association analysis can handle large quantities of data in far less time
and at less cost than subjective classification.
Subjective classification offers the most finely tuned classification
because 1) it is based on species stem counts data which offer more
information than Is available with binary data, and 2) it offers greater
data-analyst interaction. Its greatest shortcoming is lack of objectivity.
In using this method the analyst must be careful to set up and adhere
strictly to specific criteria for each vegetation class.
The vegetation structure classification method is crude in comparison
to the other two methods because it is based on changes in the types of
vegetation structure rather than changes in either species stem counts or
presence-absence data. For this classification technique, only five
vegetation structures were defined in constrast to 12 vegetation classes
which were defined using other methods. Furthermore, the character of the
vegetation at a sampling station can change without its vegetation structure
changing. For example, data classified as Carex laaustris could change to a
Degraded Carex lacustrie or a Degraded Carex striota classification. Using
41
-------
vegetation structure analysis, these data initially would be classified
Grasslike and would retain that classification.
The subjective classification and vegetation .structure analysis methods
indicate changes in sampling station classification as shown in Table 12,
because the vegetation structure analysis used in this study was based on
ground sampling stem counts data instead of the more appropriate percentage
cover data.
The mapped results from the three types of classification provide a
spot representation of changes in community area and location. The mapped
classification results document the transition at the study site from more
shallow water species to deeper water species and weedy mudflat species.
These results also show that few of the original species other than shrubs
can withstand the constantly expanding erosion and destruction of the peat
mat in the central part of the study site.
In summary, whereas the diversity index analyzes ground sampling data
at only the species level, the subjective classification and association
analysis classify data by community type, based on the species found at each
sampling station. It is the analyst's opinion that subjective classifica-
tion is the more sensitive method since it uses species stem counts data,
unlike association analysis which uses prresence-absence data. Vegetation
structure analysis is a relatively crude classification tool. Its classi-
fication is based exclusively on vegetation structure data; consequently it
cannot offer the degree of differentiation that species-based classification
offers.
USE OF AIRPHOTOS TO MONITOR CHANGE
A series of airphotos was analyzed using the methods of airphoto
monitoring and disturbance mapping. Sequential airphotos permit the
observation of changes in pattern, texture, color and other features that
occur over time at a site.
Factors Affecting Airphoto Interpretation of Wetlands
Wetland vegetation classes are easier to depict using color infrared
film rather than color film (Brown 1978). With color film the many shades
of green seen on summer imagery are too subtle for consistent and easy photo
interpretation. Color infrared film better defines living vegetation in
distinguishable shades of pink and red (Olson 1964, Gammon and Carter
1979). Water definition on wetland airphotos is particularly important.
Color infrared film defines water, which appears a deep blue or black tone,
more clearly than color film (Shima 1973, Gammon and Carter 1979).
Late spring and early fall are the best times to obtain airphotos of
southern Wisconsin wetlands (Meyer 1977). At these times the spectral
42
-------
response patterns of various vegetation classes are still consistent but
more easily distinguishable, one from another, than they are at midsummer.
In the middle of summer all spectral response patterns are so saturated in
the infrared-sensitive film layer than they are difficult to differentiate
(Scher and Tueller 1973).
Color and texture were both critical elements in identifying vegetation
classes. Since vegetation class color on airphotos changes with film type,
processing, sun angle and vegetaton vigor, color could not be used over time
to identify vegetation classes (Gallagher, Thompson and Reimold 1972).
Color and texture were used simultaneously to identify vegetation classes
(Whitman and Marcellus 1973).
Airphoto scales used were 1:120,000, 1:38,200, 1:19,100 and 1:11,500.
The 1:120,000 scale is so small that texture could not be determined,
depriving the analyst of important photo interpretation information. The
1:38,200 scale is adequate for vegetation mapping, however, a scale of
1:19,100 offers more textural and color detail. Figure 12 is an enlargement
of a 5 October 1977 color infrared airphoto, original scale 1:11,500. At
this scale of 1:11,500 so much color and texture detail is available that it
becomes difficult to generalize it sufficiently to create a legible map. As
the scale becomes larger, the problem of the same vegetation class appearing
differently on an airphoto depending on the amount of vegetation-water
interspersion becomes more acute (Olson 1964; Scher and Tueller 1973; Shima,
Anderson and Carter 1976; Brown 1978).
Airphoto Monitoring
Airphotos can be used without ground verification to monitor vegetation
change with time at a site which is inaccessible from the ground (Hubbard
and Grimes 1972). For best results, the airphotos should be taken at the
same time of day. For easy comparison they should be taken at the same
scale and at the same season or seasons each year.
A series of airphotos of the marsh/sedge meadow were taken from June
1972 to October 1977. All the photos are color infrared. All were taken in
summer or early fall with the exception of the June 1977 color airphoto and
three color photos taken on 1 March 1975, 1976, and 1977.
In general the pinkish red magenta tones of the color infrared air-
photos taken from June to mid-September indicate actively synthesizing
vegetation. (Reflectance in healthy vegetation increases dramatically from
0.7-1.3 m in the spectrum (Swain and Davis 1978).)
Light tan patches indicate areas on the color infrared film where the
vegetation is dying back. On color infrared film, dead or dying vegetation
does not have the intense pink-red tones characteristic of living vegeta-
tion. Tan patches are also indicators of disturbance on these color
infrared airphotos. On color photos, dying vegetation shows a yellowish-tan
tone.
43
-------
Figure 12. CIR airphoto used to determine airphoto grid size on October 3,
1977 (original scale 1:11,500).
-------
Besides these general characteristics the following features can be
accurately interpreted from airphotos with little or no ground verification.
1) Trees appear as the coarsest texture on the 11 airphotos in a deep
magenta or intense red tone.
2) Marsh/sedge meadow appears as a red-pink, relatively smooth
textured area between the trees to the west and the dike and
cooling lake to the east.
3) Water appears as a deep blue or black. Initially it is visible in
the river, and then in the cooling lake. Beginning in September
1975, the area of blue-black color increases in the study area as
the peat mat erodes. As such, it is an indicator of disturbance.
4) Constructed surfaces are generally characterized by straight lines
and a very high reflectance. The dike and the intake channel are
easily identified as man-made features by their bright, whitish
color and linear form. Another prominent man-made feature in the
study area is the keyhole well, a keyhole shaped weedy, shrubby
sand-dump area directly north of the overflow channel.
5) Sand has a very high reflectance; it appears tannish white on both
color infrared and color airphotos.
6) Lernna minor* (Duckweed) is a tiny surface floating plant with a very
high reflectance which appears as whitish-pink patches within the
marsh/sedge meadow on color infrared photos. Duckweed grows on
shallow water where it can be exposed to nearly full or full
sunlight. It is incompatible with sedges and grasses whose dense
canopies effectively shade it out. In this study, duckweed is an
indicator of disturbance. As more and more duckweed appears with
time, it means vegetation which would shade it out is dying off,
making increasing amounts of habitat available to it.
The 4 June 1972 airphoto (Figure 13), the earliest in the series, shows
the sedge meadow before extensive degradation occurred. The airphoto from
31 July 1974 (Figure 14) shows the area 2 yrs later. Both photos are color
infrared high altitude airphotos, original scale 1:120,000, and both show
the marsh/sedge meadow forming a dense, continuous surface. Dark patches of
open water are only visible near the lowland forest and in one area in the
northern end of the study site.
The vegetation mat is still largely intact on a September 1975 color
infrared airphoto (Figure 15), original scale 1:38,200. However, far more
detail can be seen on this photo, made at a larger scale than the previous
two. Keeping in mind that filling the cooling lake was started in January
1975, the increased surface water visible in this photo reflects the
increased groundwater flowing into the area. Several light toned, almost
whitish areas of duckweed (Leima minor) can also be seen in Figure 15
indicating that duckweed has shown up in the marsh/sedge meadow where the
peat mat has broken up. As such it is an indicator of disturbance.
45
-------
Figure 13. CIR airphoto on June 4, 1972 (original scale 1:120,.000).
46
-------
Figure 14.
CIR airphoto on July 31, 1974 (original scale 1:120,000).
47
-------
Figure 15. CIR airphoto on September 25, 1975 (original scale 1:38,200),
48
-------
Figure 16, a 24 July 1976 color infrared airphoto (original scale
1:38,200) shows that areas of open water or water covered with duckweed now
abound. Much of the northern section of the study site now appears to be
highly disturbed. Some of the areas of very high reflectance (whitish pink
tones) consist of floating and exposed peat, often covered with seedlings of
annuals and a layer of duckweed. Dark green-blue black tones in the center
of the study site indicate areas of standing water. In the lower center of
the photo, a channel is beginning to form. An actual pond, characterized by
considerable flow, is evident in the upper right-hand corner. The previous
fall, this area was covered with duckweed. A large dark area in the lower
left-hand corner of the photo consists of hydrophytic species widely
interspersed with deeper water.
Comparing Figure 16 with Figures 14 and 15 reveals that the open water
area in the upper right-hand corner (indicated by the arrow) has extended
northward. Definite channels now flow from this open area through the tree
island (extending from the upland out into the marsh) to join the large
central channel. In addition, new areas of definite disturbance are
appearing both west and east of that channel.
Figure 17, a photo made from a 24 September 1976 CIR airphoto, original
scale 1:19,100, shows that the channel running along the north knoll and the
channels flowing through the tree island are more clearly defined than 2
months earlier. Far more duckweed is in evidence south of the tree island
and west and south of the keyhole well. The quantity of water in the west-
central portion of the study site can be more easily seen than was possible
on the July photo because this September photo is at a larger scale and
also, the vegetation is dying back here. More open water is visible on this
airphoto than could be seen in Figure 15, taken a year earlier.
Figure 18 is a photo made from a 25 June 1977 color infrared airphoto,
original scale 1:38,200. This picture shows that a large channel has cut
through the center of the study area while the channels along the north
knoll have become localized. The pond where this channel ends has become
enlarged.
Another airphoto from 25 June 1977 (Figure 19) is the same scale as the
previous one but it is a color photo rather than color infrared. Although
areas of dead vegetation show up strikingly in a tannish-white, areas of
water are more difficult to detect on this photo. Nuances of color among
the living vegetation are also far less discernible here than on the
infrared photo.
Figure 20 is made from an October 1977 color infrared airphoto,
original scale 1:11,500. This photo shows areas of open water continuing to
enlarge. The central channel and the pond below it have widened. Most
light areas are areas of disturbance and destruction. At this point in
time, vegetation damage is fairly ubiquitous.
Figures 21, 22 and 23 were made from color airphotos taken on 1 March
1975, 1976, and 1977. The 1 March 1975 airphoto (original scale 1:19,100)
was taken 2 months after the cooling lake was filled and before Unit I went
49
-------
Figure 16. CIR airphoto on July 24, 1976 (original scale 1:38,200).
50
-------
- ' .'
».- ••*• «
Figure 17. CIR airphoto on September 24, 1976 (original scale 1:19 100).
51
-------
Figure 18. CIR airphoto on June 25, 1977 (original scale 1:38,200).
52
-------
Figure 19. Color airphoto on June 25, 1977
(original scale 1:38,200).
53
-------
Figure
CIR airphoto on October 3,
54
1977 (original scale 1:11,500).
-------
Figure 21. Color airphoto on March 1, 1975 (original scale 1:19,100).
-------
Figure 22. Color airphoto on March 1, 1976 (original scale 1:38,200).
56
-------
Figure 23. Color airphoto on March 1, 1977 (original scale 1:38,200).
-------
on line. Once the cooling lake filled, increased ground water flowed into
the already frozen marsh/sedge meadow, breaking through the frozen surface,
spreading out and freezing into large sheets of ice.
The 1 March 1976 airphoto (Figure 22, original scale 1:38,200) shows
much open surface water. The water was present, in part, because of a thaw
that lasted from mid February through early March. In addition, however,
the open ponds visible in the northern end of the study site and a substan-
tial channel flowing from this area down into the rest of the study site
provide evidence that substantial peat mat erosion continued throughout the
winter months.
The 1 March 1977 photo (Figure 23, original scale 1:38,200) shows two
open areas in the northern end of the study site, a definite channel flowing
along the north knoll, and channels flowing down into the center of the
study site.
On all three photos, ice and open water at the study site correspond
with areas of significant disturbance where often the vegetation has died of
exposure (Bedford 1977) and the peat mat has floated up. These three March
airphotos provide important documentation of the channel cutting which
continues throughout the winter at the study site. Open water is as clearly
defined on these winter color photos as it is on winter color infrared
photos but ice is more visible on the color photos. If the analyst were
only concerned with monitoring open water, color infrared photos should be
used exclusively (Niemann et al. 1975).
Airphoto Interpreted Disturbance Maps
Airphoto interpreted maps were made outlining areas of disturbed and
undisturbed vegetation and open water. This was done by enlarging selected
color infrared photos 6.67 times and drawing the images that were identified
on frosted mylar. Table 13 lists the dates and scales of the original
imagery used to generate these maps •
TABLE 13. AIRPHOTOS USED TO TO GENERATE DISTURBANCE MAPS
Date Scale
25 September 1975
24 September 1976
25 June 1977
3 October 1977
1 March 1975
1 March 1976
1 March 1977
1:38,200
1:19,100
1:19,100
1:11,500
1:19,100
1:38,200
1:38,200
58
-------
A series of four vegetation disturbance maps were drawn which show
Disturbed and Undisturbed areas and Open Water over a 3-yr period.
Undisturbed areas are those areas which retained their original vegetation
cover as recorded in September 1975. Disturbed areas are characterized by
vegetation that has degraded due to peat mat erosion. Duckweed was
generally found scattered among the vegetation in these areas and there are
fewer of each species than in the Undisturbed areas. Open Water areas exist
where ground water upwelling is so great it has worn a hole in the peat mat
which continues to erode or where surface water flow is substantial enough
to show open water on airphotos.
The 25 September 1975 map (Figure 24) shows only a few small areas of
disturbed vegetation in the northern end of .the study site. The only areas
where open water is visible are the drainage ditch running across the study
area and the channel along the far left edge of the study area.
Figure 25, a map drawn from a 24 September 1976 airphoto, shows how
areas of disturbed vegetation and open water have both expanded. Disturbed
vegetation has increased in the northern part of the study area and to the
south and west. Definite pools and channels of open water are now in
evidence, indicating places where the peat has disintegrated and the flow is
so strong that Lemna minor* (duckweed) cannot remain on the surface but is
carried off downstream.
On the 25 June 1977 map (Figure 26) areas of open water and disturbed
vegetation have expanded greatly, indicating that the peat mat has continued
to erode and disintegrate. This map shows a definite channel forming from
the northern end of the study area down to the southwest.
The map drawn from a 3 October 1977 airphoto (Figure 27) shows that the
central channel has widened and become better defined. In this map, new
areas of open water are appearing in areas that were classified disturbed in
June 1977. And this map provides evidence that the disturbed area has
increased considerably since June. By October 1977 much of the disturbed
area was characterized by floating mat-mud flat; vegetation was largely
absent.
Figures 28, 29, and 30 are maps that indicate open water, sheet ice,
and snow cover. These maps were drawn from imagery taken 1 March 1975,
1 March 1976, and 1 March 1977 respectively. These maps are particularly
interesting because they show the surface drainage patterns and the
continuing erosion of the peat mat that are evident in winter.
The cooling lake was filled during January 1975 but the power plant did
not go on-line until late March 1975. Consequently, during the winter of
1974-75, the study area had time to freeze before extensive leakage from the
cooling lake could begin. This resulted in much open water and sheet ice on
the surface at the northern and southern ends of the study site although
leakage is heaviest at the northern end (Figure 28).
The map in Figure 29 (1 March 1976) reveals extensive open water at the
northern end of the study site which later formed into the channel system
59
-------
September 25, 1975
D Undisturbed
D Disturbed
• Open Water
Figure 24. Disturbance map on September 25, 1975.
60
-------
September 24, 1976
D Undisturbed
D Disturbed
• Open Water
Figure 25. Disturbance map on September 24, 1976.
61
-------
June 25, 1977
Q Undisturbed
E3 Disturbed
• Open Water
Figure 26. Disturbance map in June 25, 1976.
62
-------
October 3, 1977
D Undisturbed
Q Disturbed
• Open Water
/-.
Figure 27. Disturbance map in October 3, 1977
63
-------
March 1,1975
D Undisturbed
El Disturbed
• Open Water
Figure 28. Disturbance map on March 1, 1975.
64
-------
March 1, 1976
D Undisturbed
EH Disturbed
• Open Water
Figure 29. Disturbance map on March 1, 1976.
65
-------
March 1, 1977
D Snow Cover
ED Sheet Ice
• Open Water
Figure 30. Disturbance map on March 1, 1977.
66
-------
seen on both the September 1976 and June 1977 maps. (Water moving out from
the cooling lake was heated at this point in time.) In the southern end of
the study site where leakage from the cooling lake is reduced, extensive
surface ice has formed.
The 1 March 1977 map (Figure 30) shows how the open water channel has
continued to develop over a year's time, draining off most of the open water
in the northern end of the study site. Surface ice, extensive over much of
the study site, correlates with the areas of disturbance delineated on the
June and October 1977 maps (Figures 26 and 27).
Summary of Airphoto Data Methods
Airphoto monitoring consists of obtaining a series of airphotos at the
same site over time in order to record change. It is a purely descriptive
method. The airphotos may provide a high degree of detail but no provision
is made for mapping or quantifying that information.
Airphoto disturbance mapping offers several advantages which airphoto
monitoring does not include: 1) in examining the airphotos under
magnification, far more detail can be seen and recorded than is available
using the original scale of the photography; 2) in mapping areas of
disturbed and undisturbed vegetation and open water, a permanent record of
this information is created; and 3) in generalizing all disturbance into one
category, a series of maps, more easily readable than airphotos, is created
allowing the extent of disturbance or change to be seen at a glance.
MONITORING VEGETATION CHANGE WITH AIRPHOTO AND GROUND SAMPLING DATA
In order to demonstrate change in vegetation with time, two techniques
were used: airphoto grid assessment and airphoto interpreted vegetation
mapping. These methods combine airphoto and ground sampling data.
Airphoto Grid Assessment
Airphotos selected to cover the study site from 1972 to 77 (Table 14)
were overlaid with a grid to scale. The percent of cell for each of several
vegetation classes was determined. This information was summed for each
vegetation class over each airphoto such that percent changes in the area of
each class could be compared with time.
The selected imagery was viewed on a Richards light table. A grid size
representing approximately 50 m on the ground was selected because it was
best suited to the size of the vegetation patterns seen on a 3 October 1977
airphoto (originally at a scale of 1:11,500). This grid size was reduced to
fit airphotos at scales of 1:19,100, 1:38,200, and 1:120,000 and a mylar
67
-------
overlay, showing the location of the transects and stations in the study
area, also was made to fit each scale (Figure 31). Use of this transect-
station overlay meant that ground sampling data observations could be used
to help identify the vegetation patterns observed on each airphoto.
TABLE 14. LIST OF AIRPHOTOS USED IN AIRPHOTO GRID ASSESSMENT
Date Scale CIR/Color
_ 1 _. j. . . _ . -
4 June 1972 1:120,000 CIR
31 July 1974 1:120,000 ' CIR
25 September 1975 1:38,200 CIR
24 July 1975 1:38,200 CIR
24 September 1976 1:19,100 CIR
25 June 1977 1:38,200 CIR
25 June 1977 1:38,200 Color
3 October 1977 1:11,500 CIR
For each airphoto listed in Table 14, the percent of cell was recorded
for each vegetation class present. Table 15 lists the vegetation classes
which could be discerned on each photo.
A key was assembled for each airphoto describing the appearance of the
different vegetation classes (Appendix F). These descriptions were obtained
by identifying, on the airphoto, the locations where the vegetation classes
appeared in the ground sampling data. The appearance of each vegetation
class was defined in terms of color (tone) and texture. Many of the
vegetation classes used in the airphoto grid assessment are the same as
those used in the subjective classification of the ground sampling data.
Additional classes which could be identified by this method were Sedges
and Grasses, Degraded Sedges and Grasses, Typha latifolia, Scirpus
fluviatilie, Floating Mat, and Lemna .minor. A Sedges.and Grasses
classification was used when Carex laauetris and Carex striata could not be
differentiated. In September 1975, the two Carex communities supported a
dense Calamagrostis canadensis component. Once this species was greatly
reduced in numbers, the Carex etriata and Carex laauetris classes were
easier to separate. Dense clones of Typha latifolia and Scirpue fluviatilis
were identifiable on airphotos at a scale of 1:28,200 or larger. Floating
Mat refers to large areas where layers of the peat mat have separated from
the wetland bottom and floated up to the surface. These areas of peat-mat
are interspersed with the seedlings of densely clustered annuals. Together
these two categories have a high reflectance. Lemna minor areas are easily
68
-------
lO
-.-i-.-.'-'1
s a s i s~~5 g s
Figure 31. Mylar overlay locating transects and sampling stations.
-------
TABLE 15. VEGETATION CLASSES IDENTIFIED USING AIRPHOTO GRID ANALYSIS
Vegetation classes
Carex laoustris
Degraded Carex laoustris
Carex striata
Degraded Carex striota
Sedges and grasses
Degraded sedges and grasses
Transition
Degraded transition
Emergents
Transit ion-emergents
Disturbed emergents
Typha lati folia
Scirpue fluviatilis
Spiraea alba
Shrubs
Weedy annuals
Floating mat
Lemna minor
Open water
Shrub carr
Trees
4 June
1972
X
X
X
X
X
X
X
X
31 July
1974
X
X
X
X
X
*
X
X
X
25 Sept.
1975
X
X
X
X
X
X
X
X
X
X
X
X
24 July
1976
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
24 Sept.
1976
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
25 June
1977a
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
25 June
1977a
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
3 Oct.
1977
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
aColor film. Other photos are CIR.
-------
identified by their high reflectance, distinctive color and flat, smooth
texture.
Once data was recorded for all the cells considered on an airphoto, the
percent of each vegetation class was determined for the entire scene by
totaling the percent of each type of vegetation and dividing by the total
number of cells times 100. This allowed changes in percent cover of the
various vegetation classes to be compared with time.
Table 16 gives the percentages of each vegetation class for each of the
airphotos analyzed. Not all classes could be identified on all dates of
imagery due to differences of scale and data. In order to compare results
from each airphoto, the 14 original vegetation classes were combined as
follows. Shrubs, Spiraea alba and Open Water classes remained the same as
previously. Transition and Emergents and Typha latifolia were combined into
a Transition-Emergents category; Carex strieta and Carex laeustris were
combined into a Sedges category; Degraded Carex laaustris and Degraded Carex
stricta were combined into a Degraded Sedges class. The Other Disturbance
class combined Lenrna minor, Floating Mat, and Weedy Annuals- Degraded or
disturbance classes did not appear on imagery taken prior to 1975. Table 17
gives the results of the grid analysis using these categories.
The percentage cover estimates of the different vegetation classes were
more accurate using some airphotos than other methods. Vegetation features
on the two high altitude photos (scale 1:120,000) were so small that neither
vegetation classes nor total area covered could be identified accurately.
The June 1977 color photo masked vegetation class differences and percent
cover data are unreliable. The tones of the October 1977 airphoto
highlighted disturbance (it appears greenish-purple) but made it difficult
to distinguish different disturbance types. Consequently the analyst
believes the vegetation class area estimates made from the color photos of
September 1975, July 1976, September 1976 and June 1977 are the most
representative of what appeared on the ground at that time. If
approximately the same percentage cover estimate was made for a vegetation
class from two photos dated consecutively, these two estimates are assumed
to be correct.
Using airphoto grid assessment estimates, Shrubs made up approximately
7.0% of the area in the study site in 1972 and 1974 and continued to do so
in 1977. Open Water areas consisted of only 2.0% of the area in 1972 and
1974 but increased to 10% in 1977. Open Water area is much greater (33%) if
Floating Mat and Lernna minor areas also are included. From Table 17 it
would appear that the amount of Open Water was overestimated in the July
1976 color infrared photo and underestimated in the June 1977 color photo.
On the July 1976 airphoto, areas with much interspersion are a very dark
green and may have been recorded in some instances as Open Water. On the
June 1977 color image, the color of the water in the study site is brownish-
green and difficult to distinguish from some of the vegetation.
The Transition and Emergents classes were difficult to distinguish, so
they were combined for purposes of this analysis. The Transition class
often was difficult to distinguish from the Floating Mat class. As a
71
-------
TABLE 16. CLASSES ORIGINALLY USED TO SUM GRID ANALYSIS DATA
Vegetation classes
Shrub
Open
Typha lati folia
Transition
Emergent
Sedges
Degraded sedges
Carex etriata
Degraded Carex stricta
Carex laauetris
Degraded Carex laoustrie
Spiraea alba
Lernna minor
Floating mat
Weedy Annuals
4 June
1972
6.8
2.1
-
25
9.3
50
-
-
-
..
-
6.5
-
-
—
31 July 25 Sept.
1974 1975
4.6 7.4
2.2 5.2
_
15 16
3.3 9.6
75
- -
31
-
19
-
a 9.5
2.2
-
— —
24 July
1976
6.8
9.5
-
13
18
-
-
20
3.3
6.4
6.5
8.2
7.1
-
1.3
24 Sept.
1976
5.9
7.6
1.4
10
9.8
-
-
23
10
5.9
7.8
9.4
8.1
-
0.2
25 June
1977a
6.0
7.4
5.8
12
15
-
0.5
17
13
1.9
11
5.7
4.4
-
7.0
25 June
1977a
5.3
5.5
0.5
-
24
23
5.2
-
6.8
-
8.9
8.5
3.6
8.7
—
3 Oct
1977
7.8
10
1.9
-
24
5.0
3.6
6.1
8.5
-
-
9.2
21
1.6
1.1
Not distinguishable on this image.
-------
TABLE 17. GRID ANALYSIS PERCENT COVER DATA
Vegetation classes
Shrub
Spiraea
Open
Transition-Emergents
Sedges
Degraded Sedges
Other Disturbance
4 June
1972
6.8
6.5
2.1
35
50
'
—
31 July
1974
4.6
-
2.2
19
75
-
—
25 Sept.
1975
7.4
9.5
5.2
25
50
-
2.2
24 July
1976
6.8
8.2
9.5
30
27
9.7
8.4
24 Sept.
1976
5.9
9.4
7.6
21
29
18
8.3
25 June
1977a
6.0
5.7
7.5
33
19
24
11
25 June
1977a
5.3
8.5
5.5
24
25
21
12
3 Oct.
1977
7.8
9.2
10
25
11
12
24
-------
result, the percentage cover for this category varies. The Transition-
Etnergents class was particularly difficult to distinguish on high-altitude
imagery where little detail is visible.
As water depths increased by 10 cm in the wetland, the growth of the
Transiton and Emergents species was encouraged. This was countered by the
fact that the areas of Open Water were increasing at the expense of the
Transition class. While Emergents species were increasing, Transition
species were decreasing so that total Transition-Emergents remained
approximately the same over 4 yr.
Sedges cover was estimated at 50% on the September 1972 and September
1975 airphotos. The most rapid destruction at the study site took place
from 1975 to 1976. During this time the Sedges class cover decreased from
50% to 25 to 30%. The June 1977 color infrared photo shows Sedges cover
being further reduced to 19%. Sedges were difficult to identify on the
October 1977 airphoto which explains their very low (11%) cover estimate.
As Sedges decreased by 39% from 1974 to 1977, Degraded Sedges increased
from 0 to 24%. The estimate of Degraded Sedges dropped low on the October
197.7 photo where they could not be distinguished from the Other Disturbance
class. Other Disturbance increased from 2.2% in September 1975 to 11 to 12%
in June 1977. Eight percent of the site was classified Spiraea alba
throughout the span of the study.
Total Open Water, Disturbed Sedges, and Other Disturbance increased
from 2.5% in 1972 to 46% in 1977. This seems to be an accurate
approximation, based on ground knowledge.
Table 18 lists results of the chi square tests done with grid analysis
results from a number of dates. Chi square tests done comparing data
collected in 1975 and later were significant at the 0.01 or 0.001 level. In
order to meet the requirements of the test, some of the vegetation
categories listed in Table 16 had to be combined. Vegetation categories
used for these tests were Sedges, Degraded Sedges, Transition-Emergents,
Spiraea, Shrubs, and Other Disturbance which included the Open Water
category.
TABLE 18. CHI SQUARE TESTS ON GRID ANALYSIS DATA
Grid analysis dates Not significant 0.01 0.001
June 1972 and September 1975 X
June 1972 and July 1976 X
September 1975 and June 1977 X
June 1972 and June 1977 X
74
-------
Airphoto Interpreted Vegetation Maps
Vegetation classification maps were drawn in the same manner as the
disturbance maps. The original photo was enlarged 6.67 times and projected
on a ground glass screen. This enlarged image was overlaid with a mylar
sheet (to scale) showing the location of transects 18 to 42 and the sampling
along each of these transects. Ground sampling data for each station which
had been printed out by computer and subjectively classified for all
stations served as the basis for identifying vegetation classes on the
mylar. The data based on ground sampling could be located specifically at
the sampling points shown on the magnified images of the photos. The
patterns seen on the photo images then could be labeled. The vegetation
classes were first delineated on the mylar with pencil, then redrawn in ink
and labeled. Each map was enlarged or reduced as necessary to bring all
the maps to a common scale.
When the maps were brought to a common scale, each was covered with a
sheet of graph paper. Total study site area and the area of each vegetation
class were determined for each map. The total area of each vegetation class
was divided by the total site area and multiplied by 100 to provide the
relative percentages for each class. Table 19 lists the eight photos used
to generate the airphoto interpreted vegetation maps.
TABLE 19. LIST OF AIRPHOTOS USED TO GENERATE
AIRPHOTO INTERPRETED VEGETATION MAPS
Date Scale Color/CIR
4 June 1972
31 July 1974
25 September 1975
24 July 1976
24 September 1976
25 June 1977
25 June 1977
3 October 1977
1:120,000
1:120,000
1:38,200
1:38,200
1:19,100
1:38,200
1:38,200
1:11,500
CIR
CIR
CIR
CIR
CIR
CIR
Color
CIR
A color and texture key was assembled for each airphoto, to facilitate
consistent mapping of the various communities (Appendix G). Sixteen
vegetation classes were identified using color infrared airphotos taken
25 September 1975 and 25 June 1977. Although one less class was identified
on the 25 June 1977 color airphoto and the 3 October 1977 airphoto, the
analyst felt less confident about vegetation class identification using
these photos. Table 20 lists the vegetation classes discernible on each of
the airphoto interpreted maps.
75
-------
TABLE 20. VEGETATION CLASSES DISCERNIBLE ON PHOTO INTERPRETED VEGETATION MAPS
A.
B.
C.
D.
E.
F.
G.
H.
I.
J.
K.
L.
M.
N.
0.
P.
Q-
R.
S.
4 June 31 July
Vegetation classes 1972 1974
Carex lacustrie
Disturbed Carex laaustris
Carex stricta
Disturbed Carex stricta
Sedges and Grasses X X
Transition X X
Disturbed Transition
Emergents X X
Disturbed Emergents
Typha lati folia
Soirpus fluviatilis
Spiraea alba X X
Shrubs X X
Weedy Annuals
Floating Mat
Lemna minor
Open Water X X
Shrub carr X X
Trees X X
25 Sept.
1975
X
X
X
X
X
X
X
X
X
X
X
X
X
24 July
1976
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
24 Sept.
1976
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
25 June 25 June
1977 1977
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
.
X
X
X
X
X
X
X
X
X
X
X
3 Oct
1977
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
-------
Table 21 classifies the vegetation classes according to how easy they
were to identify: easy, moderately difficult, difficult, and very
difficult.
TABLE 21. EASE WITH WHICH VEGETATION CLASSES COULD
BE IDENTIFIED ON AIRPHOTOS
Easy
Sedges and Grasses
Degraded Sedges
Shrubs
Lernna minor
Open Water
Shrub carr
Trees
^•^•^^^^W^^^^^«MI^^^^B^H^B.V*^«^^^-^>~i^B^^^*W^^B^V*H^»^^^H
Moderately
Easy
Carex lacuetris
Carex stricta
Typha lati folia
Scirpus fluviatilis
Spiraea alba
^•M^^P^b^MM^HM^— ^^^P^M^hW^^KMHA^^B.M^^^^H
Difficult
Transition
Emergents
Floating Mat
Weedy Annuals
Degraded Carex
lacustris
Degraded Carex
stricta
Very
Difficult
Degraded
Emergents
Sedges and Grasses and Degraded Sedges were easy to identify. It was
slightly more difficult to distinguish Carex lacustris from Degraded Carex
stricta. The Shrubs, Trees and Shrub carr classes were all easy to identify
due to their coarse textures and generally rounded crown shapes. Open Water
was easily identifiable on all color infrared imagery. Lemna minor, with
its bright reflectance and flat texture, could be identified easily.
Typha latifolia and Soirpus fluviatilis, occurring in dense patches,
were moderately easy to identify due to their distinctive clone growing
patterns and coarse textures. Spiraea alba was easy to recognize when it
occurred in dense patches, but difficult to identify when interspersed with
dense Carex stricta and Calamagrostis canadensis.
The Transition and Emergents classes were always difficult to
distinguish due to their similarity of color and texture. The same was true
of the Floating Mat and Weedy Annuals classes.
Table 22 lists all the vegetation classes identified on the eight
airphoto interpreted maps and gives the relative percentage of each class
per airphoto. Differences of scale and date prevented identification of all
classes on all dates of imagery. Consequently classes were combined as
follows: The Shrubs, Spiraea, and Open Water classes remained unchanged.
The Transition, Emergents, Typha and Scirpus classes were combined into a
Transition-Emergents category. Carex laaustris and Carex stricta were
combined into a Sedges class. Degraded Carex lacustris and Degraded Carex
stricta were combined into a Degraded Sedges class. The Other Disturbance
77
-------
00
TABLE 22. RELATIVE PERCENTAGE COVER OF EACH VEGETATION CLASS DEFINED
USING AIRPHOTO INTERPRETED VEGETATION MAPPING
A.
B.
C.
D.
E.
F.
G.
H.
I.
J.
K.
L.
M.
0.
P.
Q-
R.
S.
T.
4 June 31 July
Vegetation classes 1972 1974
Carex lacustris
Degraded Carex lacustris
Carex etriata
Degraded Carex striota
Sedges and Grasses 24 30
Transition 17 10
Degraded Transition
Emergents 10 9.9
Degraded Emergents
Typha lati folia
Scirpus fluviatilia
Spiraea alba 10 11
Shrubs 2.7 4.3
Weedy Annuals
Floating Mat
Lerrma minor
Open Water 0.2
Shrub carra
Trees 0.8
25 Sept.
1975
23
22
1.1
12
1.5
7.6
1.8
11
9.2
3.1
1.8
1.5
•
1.6
24 July
1976
5.3
7.6
30
10
1.8
5.3
16
0.5
0.5
11
2.4
2.1
1.1
2.3
.1.6
1.3
24 Sept .
1976
5.9
9
22
5.9
12
1.4
16
1.6
0.5
4.2
4.0
3.4
2.0
4.7
2.2
1.9
25 June
1977
7.2
11
18
11
4.2
7.2
9.7
0.6
0.7
7.6
2.7
3.6
5.1
2.2
5.8
1.7
25 June
1977
9.4
10
5.5
13
12
12
0.3
0.3
11
2.1
1.1
11
2.5
4.8
1.7
3 Oct
1977
7.0
8.9
15
11
4.9
9.5
2.0
0.5
0.5
7.5
4.6
3.5
9.2
2.9
10
1.1
Appears at far southern end of marsh sedge meadow outside the study site
-------
combined all the Degraded classes not already classified, as well as Lernna
minor, Floating Mat and Weedy Annuals (Table 23).
Results of the airphoto interpreted vegetation mapping show that Shrubs
probably made up 5.0 to 6.0% of the area within the study site while Spiraea
accounted for 10% of the area. These figures remained fairly constant over
3 yr. This method shows Open Water increasing from 0.2% in 1974 to 10% in
1977. The analyst believes Open Water was generally underestimated since it
often appeared as rivulets among the vegetation, making it difficult to
accurately estimate its relative cover. The Transition-Emergents class
decreased from 20 to 25% of the total area in 1974 to 1975 to approximately
15% by 1977 while Sedges decreased from 45% to 22% of the total area over
the same period. Percentage area of Sedges was underestimated on the
1:120,000 scale airphotos (4 June 1972 and 31 July 1974). At this scale,
features often were so small that lines could not be drawn around them and
consequently much of the area remained unclassified.
The Degraded Sedges class increased in area from 0 to 22% between 1974
to 1975 and 1977. The Other Disturbance class increased from 0 in 1974 to
16 to 18% of the total area in 1977. Total disturbed area (combining Open
Water, Degraded Sedges, and Other Disturbance) increased from 0.2% in 1972
to 37% in 1977. Percent unclassified area, discounting the 1:120,000
airphotos, varied from 1.3 to 3%.
Changes Observed on the Photo Interpreted Vegetation Maps—
On the 4 June 1972 map (Figure 32) interpreted from a high altitude
color infrared airphoto, scale 1:120,000, two discrete ponds of open water
are visible. Otherwise the vegetation is a continuous mat of sedges and
grasses, transition and emergent vegetation, shrubs and Spiraea (Table 20).
The 31 July 1974 map (Figure 33) was interpreted from another high
altitude color infrared 1:120,000 scale airphoto. Far less detail of color
and texture were available on this airphoto. Consequently no open water
could be seen and vegetation classes could not be identified with the same
level of confidence as the 4 June 1972 airphoto.
The map drawn from the 25 September 1975 color infrared airphoto, scale
1:38,200, shows far more detail (Figure 34). Thirteen vegetation classes
could be identified using this airphoto in contrast with only five to eight
classes identified on the high altitude imagery. The two open ponds visible
on the 4 June 1972 airphoto can be seen on this airphoto. In addition a
channel can be seen running along the far west side of the study area.
Disturbance is apparent in two woody areas; one located near the keyhole
well, the other near the northern end of the study site where a road was
relocated. Another indicator of disturbance, duckweed, which floats on
still, open water can be seen along the north knoll and in the western
portion of the study site along the lowland forest. Figure 35 was made from
an airphoto taken approximately 7 months after the cooling lake was filled
and 4 months after Unit I went on-line.
79
-------
00
o
TABLE 23. VEGETATION CLASSES USED TO SUMMARIZE
AIRPHOTO INTERPRETED VEGETATION MAPS
Relative %
Vegetation classes
Shrub
Spiraea
Open Water
Transition-Emergents
Sedges
Degraded Sedges
Other Disturbance
Unclassified
4 June
1972
8.7
5.2
0.2
27
24
-
-
35
31 July
1974
4.3
11
-
20
30
1.1
-
35
25 Sept,
1975
11
11
1.5
22
45
-
7.5
1.9
of each class
24 July 24 Sept.
1976
3.7
11
1.6
19
35
18
11
1.3
1977a
5.9
4.2
2.2
31
27
15
11
3.0
25 June
1977a
4.6
7.6
5.8
14
25
22
18
1.8
25 June
1977a
3.8
11
4.8
25
15
23
15
2.1
3 Oct
1977
5.7
7.5
10
15
22
20
17
1.3
-------
OO
LOWLAND FOREST
LOWLAND S
FOREST
Figure 32. Photo interpreted vegetation map on June 4, 1972.
-------
OO
NJ
Figure 33. Photo interpreted vegetation map in July 31, 1974.
-------
00
u>
Figure 34. Photo interpreted vegetation map on September 25, 1975.
-------
The 24 July 1976 map (Figure 35), drawn from a color infrared airphoto,
scale 1:38,200, shows that nearly 1 yr later, open water is appearing in the
north end of the marsh/sedge meadow along with weedy annuals and Lerma
minor'. Lemna minor is also becoming prominent in the center of the
wetland. Large areas of Degraded Carex lacuetris and Carex striata are
appearing along the dike.
Figure 36 was drawn from a 24 September 1976 color infrared airphoto,
scale 1:19,100. The larger scale provides more detail. This map does not
show much change from Figure 23 except that there is more open water
visible where there was Lemna minor' previously, indicating continued peat
mat erosion. And on this airphoto, more area in the center of the study
site could be classified as floating mat. This may be a function of the
vegetation's partial dieback, making mud areas more visible.
Figure 37 is a map drawn from a 25 June 1977 color infrared airphoto,
original scale 1:38,200. Channels are now seen extending from the northern
end of the study site through the tree island and into the central portion
of the study site, finally joining Duck Creek. Figure 38, drawn from a
25 June 1977 color airphoto, original scale 1:38,200, shows far more open
water than was visible during the summer and fall of 1976. Open water now
appears to be peppered throughout the study site. The central portion of
the study area appears to be floating mat.
The 3 October 1977 map (Figure 39) was drawn from a color infrared
airphoto, original scale 1:11,500. This scale offered so much subtle color
and texture information that it was difficult to generalize vegetation
patterns. Disturbance is highlighted in pink tones on this airphoto but it
is difficult to differentiate the different types of disturbance. The area
of floating mat has increased to cover the entire central portion of the
study site.
In order to photo-interpret maps showing this degree of detail it is
necessary to collect extensive ground sampling data, including species stem
counts, since numbers of individual species as well as the identification of
species present is important.
Ideally the photo interpretation of airphotos should be done on an
orthophoto base map. This would remove any effects of tilt and also any
changes of scale on the airphoto due to differences in terrain relief.
Relief at the study site was negligible but airphotos did suffer from a
slight element of tilt.
Comparing Airphoto Grid Analysis and Airphoto Interpreted Mapping Results
Airphoto grid analysis and airphoto interpreted vegetation mapping
detected change in vegetation community area and location over 3 yr. Using
each method, a definite trend from healthy marsh/sedge meadow vegetation to
heavily eroded and disturbed wetland vegetation was observed. Airphoto grid
analysis and airphoto interpreted vegetation mapping demonstrated changes in
84
-------
00
In
Figure 35. Photo interpreted vegetation map on July 24, 1976.
-------
CO
LOWLAND FOREST
^.,"-!%
COOLING LAKE
Figure 36. Photo interpreted vegetation map in September 24, 1976.
-------
CO
Figure 37- Photo interpreted vegetation map on June 25, 1977.
-------
co
00
Figure 38. Photo interpreted vegetation map on June 15, 1977.
-------
OO
Figure 39. Photo interpreted vegetation map on October 3, 1977.
-------
the area of a few, large clone-forming species. General comments about each
method and a comparison of their results follow.
The estimation of percent cover of vegetation classes within a grid
cell was the strength and weakness of the airphoto grid analysis method.
Another weakness was the procedure of viewing one cell at a time with little
reference to what was around it through a 15 power microscope. At the same
time, however, cell by cell viewing allowed a more precise percent estimate
of vegetation classes than was possible using airphoto interpreted
vegetation mapping.
Airphoto interpreted mapping necessitated drawing lines around vegeta-
tion classes. Because the Carex lacustrie and Carex stricta communities and
the Transition and Emergents communities tended to grade into one another,
drawing a line around a community could not always be done with confidence.
The technique of placing a grid over a finished map and counting numbers of
cells of each cover type resulted in area estimates that were less
consistent than those done with grid analysis. Although two partial cells
of a vegetation class were counted as one cell, the percent cover of all
vegetation classes was consistently underestimated.
Table 24 lists summary vegetation classes and percent cover results,
airphoto by airphoto, using airphoto grid analysis and airphoto interpreted
vegetation mapping. Poor agreement was found between the two methods. When
the seven vegetation classes of Table 24 are combined in Table 25 and
expressed as Disturbed and Undisturbed Vegetation and Open Water, agreement
among methods is close.
Airphoto grid analysis and airphoto interpreted vegetation mapping both
define many vegetation classes and quantify the percent cover of each
class. Beyond these similarities are differences: with airphoto grid
analysis, percent cover determination is done simultaneously with vegetation
class definition on a cell-by-cell basis whereas with airphoto interpreted
vegetation mapping, the analyst maps the entire scene and then lays a grid
over it to determine the total area mapped and the percent total area made
up by each vegetation class.
The analyst believes airphoto interpreted vegetation mapping is the
less accurate of the two methods because it uses a technique to determine
percent vegetation class cover which is crude in comparison to the technique
used to determine cover in airphoto grid analysis. (This does not have to
be the case. Mapped areas could be digitized to obtain accurate cover
determinations from airphoto interpreted vegetation maps (Niemann 1979).)
This method offers the advantage of generating photo interpreted vegetation
maps. The slightly greater time and dollar costs associated with airphoto
interpreted vegetation mapping result from the drafting which is necessary
to produce finished maps.
90
-------
TABLE 24. COMPARISON OF PERCENT COVER RESULTS USING AIRPHOTO
GRID ANALYSIS AND AIRPHOTO INTERPRETED VEGETATION MAPPING
Grid
analysis
June 4. 1972
Shrubs
Spiraea.
Open Water
Trans 1 1 1 on -Emer gents
Sedges
Degraded Sedges
Other Disturbance
Unclassified
September 25, 1975
Shrubs
Spiraea.
C0en Water
Tr ans 1 1 1 on-Emer gents
Sedges
Degraded Sedges
Other Disturbance
Unclassified
Dike
September 24, 1976
Shrubs
Spiraea
Open Water
Trans Itlon-Emer gents
Sedges
Degraded Sedges
Other Disturbance
Unclassified
June 25, 1977-Color film
Shrubs
Spiraea
Open Water
Trans Itlon-Emergents
Sedges
Degraded Sedges
Other Disturbance
Unclassified
6.8
6.5
2.1
35
50
-
-
-
7.4
9.5
5.2
25
50
-
2.2
-
-
5.9
9.4
7.6
21
29
18
8.3
-
5.3
8.5
5.5
24
25
2t
12
-
Vegetation
mapping
8.7
5.2
0.2
27
24
-
2.5
35
11
11
1.5
22
45
-
7.5
1.9
-
5.9
4.2
2.2
31
27
15
11
3.0
3.8
1 1
4.8
25
15
23
15
2.1
July 31, 1974
Shrubs
Spiraea
Open Water
Trans 1 1 1 on -Emer gents
Sedges
Degraded Sedges
Other Disturbance
Unclassified
July 24, 1976
Shrubs
Spiraea
Open Water
Trans 1 1 1 on-Emer gents
Sedges
Degraded Sedges
Other Disturbance
Unclassified
Dike
June 25, 1977
Shrubs
Spiraea
Open Water
Trans 1 1 1 on-Emer gents
Sedges
Degraded Sedges
Other Disturbance
Unclassified
October 3, 1977
Shrubs
Spiraea
Open Water
Trans 1 1 1 on -Emer gents
Sedges
Degraded Sedges
Other Disturbance
Unclassified
Grid
analysis
4.6
-
2.2
19
75
-
-
-
6.8
8.2
9.5
30
27
9.7
8.4
-
-
6.0
5.7
7.5
33
19
24
11
-
7.8
9.2
10
25
11
12
24
-
Vegetat 1 on
mapping
4.3
It
-
20
30
-
-
35
3,7
1 1
1.6
19
35
18
11
1.3
-
4.6
7.6
5.8
14
25
22
18
1.8
5.8
7.5
10
15
22
20
17
1.3
91
-------
TABLE 25. VEGETATION CLASSES DERIVED USING COMPUTER ASSISTED MAPPING
Vegetation class
Carex lacuatris
Carex striota
Degraded Sedges
Spiraea /Sedges
Spiraea! Shrubs
Shrubs
Transition
Emergence
Transition-Emergents
Lemma minor
Weedy Annuals
Dike
Open Water
Unclassified
Total Area
September
m2 x 10
54
44
44
18
5.7
41
12
4.5
15
1.1
6.0
246
25, 1975
3 %
22
18
18
7.4
2.3
17
4.9
1.8
6.2
0.5
2.4
July 24,
m2 x 103
43
37
33
11
73
26
8.7
14
1.0
246
1976
18
15
13
4.4
30
10
3.5
5.7
0.4
June 25,
of- x 10J
33
53
24
17
42
29
14
4.7
20
9.7
246
1977-CIR
13
21
9.6
6.7
17
12
5.9
1.9
8.1
3.9
June 25
nf4 x 10J
59
43
19
48
23
25
12
15
3.0
246
1977-Color
24
17
7.6
20
9.2
10
4.8
6.1
1,2
-------
GENERATING COMPUTER ASSISTED VEGETATION MAPS
Computer assisted vegetation maps based on film dye density data offer
consistent classification of a large area and quantification of the area
covered by each vegetation class.
Generating computer vegetation maps entails the following steps:
1) Selecting airphotos to be mapped; 2) scanning the imagery to obtain film
dye density data; 3) correcting and transforming the data; 4) choosing and
"cleaning" the training sets selected, to represent the different vegetation
classes observed on the photo; 5) using training set values to assign, a
classification to each pixel (unit of measurement) that makes up a scene;
and 6) classifying the entire scene.
A discussion follows of the two processes, namely, film scanning to
obtain dye density data and dye density exposure classification. Lillesand
and Kiefer (1979) discuss black and white film characteristics, color film
characteristics and color infrared film characteristics.
Film Scanning
Film density is measured with an instrument called a densitometer.
Spot densitometers take density readings on an image by manually translating
the image with respect to the instrument optics. When an entire image is to
be scanned, a scanning densitometer, which covers the entire image scan line
by scan line, is used. Although a scanning densitometer was used to extract
airphoto density data for a portion of this study, it is easier to first
explain how a spot densitometer works. The scanning densitometer functions
in a smilar manner except that it picks up continuous bands of data instead
of one spot at a time.
Spot Densitometer—
The spot densitometer operates by means of a light source which
illuminates the image under study with a reference beam of incident light
(Figure 40). The incident light passes through an aperture assembly which
allows selection of various image spot sizes. As the film is scanned, it
passes between the aperture assembly and a spectral filter assembly so that
data from one film layer at a time is collected. Incident light from the
light source passes to a receiver—a photo-multiplier tube—which responds
electronically to that part of the light beam which has passed through the
image. A readout unit displays the receiver response in terms of image
density. In this study, image density was expressed as values ranging from
0 to 225. Objects that look very dark (such as water) on an airphoto were
93
-------
(4) Receiver
(3) Color filter wheel
Film
Spot being measured
(2) Aperture
(1) Light source
(5) Electronics
(6) Digital display
and/or recording
Figure 40. Spot densitometer (from Lillesand and Kiefer, 1979).
-------
assigned very low values while very bright objects were assigned very high
values•
Scanning Densitometer—
Two types of scanning densitometers exist: flatbed and rotating drum
systems. A flatbed system moves the image in a flat plane with respect to
the source/receiver optics (Figure 41). Readings are taken at discrete
intervals along scan lines. At the end of a line, the instrument steps over
and begins the next scan line parallel and contiguous to the previous
line. This process is repeated until the entire image has been scanned.
In a rotating drum scanner (Figure 41) the film is mounted over a
square hole in a rotating drum such that it forms a portion of the drum's
circumference. The image on the drum spins round while the source/optics
receiver steps over after each drum rotation.
A revolving drum Optronics P 1700 Photomation Mark II Scanning
Microdensitometer was used to scan the airphotos and stepwedges. The
scanning was done using three 0.012 um narrow band filters centered at 0.45
ym, 0.55 urn, 0.65 urn in order to extract blue, green, and red film data.
All images scanned were at a scale of 1:38,200. A pixel size of 50 urn
was selected which gave an on-the-ground spot size of 1.9 ra. As an image is
scanned on a scanning microdensitometer, continuous output from the photo-
multiplier tube is converted to describe integer values on a pixel-by-pixel
basis. These density values ranging from 0 to 255 are recorded on computer-
compatible seven track tapes (Buchanan 1977, Lillesand and Kiefer 1979).
Input from the scanner is in the form of a two dimensional array of
numbers proportional to the dye density. These numbers are written in eight
bit words on magnetic tape.
After the scanning process, film density data are available for
processing. Since film density measurements reflect the combined broad band
sensitivities of all film layers and the film base, film density is not
related linearly to film exposure and light reflected from the ground.
Film exposure on a photo is related to the reflectance of the object
imaged at that point. Characteristic curves (often called density log
exposure or D-log E curves) are used to relate the exposure values for the
photo by the densitometer (Figure 42). (The computer assisted classifica-
tion process carried out. at the University of Wisconsin-Madison is unique
since it works with spectral analytical density data (Scarpace 1978). D-log
E curves are generated by exposing a portion of a film to a series of known
energy steps. This creates a strip of film called a filmwedge, made up of a
series of sections or steps, ranging from a very light density to a very
dark density. Such a filmwedge provides a quantified incremental scale of
densities, each with a known exposure. The D-log E curves make it possible
to convert the film densities provided by a densitometer into exposure
95
-------
Receiver
VO
Receiver
Source
(a) Flatbed version
(b) Drum version
Figure 41. Two types of scanning microdensitometers: In the flatbed version the image
scanned moves in relation to source/receiver, in the drum version, the drum and
source/receiver move (from Lillesand and Kiefer, 1979).
-------
3.0-•
2.0--
O)
Q
1.0--
(a) Black and white negative film
1 2
Relative log exposure
3.0--
2.0--
-------
values. Exposure data can be correlated directly with ground phenomena
(Scarpace 1978).
Flim Data Classification
A number of computer programs have been developed in the Environmental
Remote Sensing Center at the University of Wisconsin-Madison to input,
output, correct, analyze, transform and classify film density data
(Scarpace, Fisher, and Quirk 1978). Program DLOGE uses scanned density
values and established log exposure values to create D-log E curves for the
yellow, magenta, and cyan film layers (Figure A3).
Correcting and Transforming Data—
Program CORRECT corrects the scanned data by transforming the density
values recorded with the scanning microdensitometer to nonlinear exposure
values using the D-log E curves generated by program DLOGE and user supplied
information about the exposure ranges to be used for output values. In
essence, the analyst defines the linear portion of each D-log E curve by
entering exposure values A and B (Figure 44). The interval between A and B
is divided into 256 equal width increments of log exposure, and each is
assigned a value between 0 and 255. CORRECT converts the original density
data values to log exposure values. At the same time CORRECT generates D-
log E curves for all three film layers of data (Figure 45).
Lens falloff, which causes vignetting toward the edges of the film
image also must be corrected as it prevents vegetation response signatures
from being consistent across the image. Program FALLOFF corrects the log
exposure values for lens falloff in one of two ways:
if the cosine option is chosen, the effect of lens falloff is assumed
to be expressed as
E(r) = E(o)cosn6 log E(r) = log E(o) + log [cosn(9)] (1)
where:
E(r) is the exposure at a distance r from the center,
E(o) is the exposure at the center and
E. is the angular distance from the lens axis.
(The program computes a look-up table of the value log (cos 9) if the table
option is used and the distance of any point in the data-file from the
center row, column must be < 1650 pixels.) For each pixel in the data-file,
the log exposure and distance from center are calculated. The log exposure
98
-------
VO
VO
Ii-t.OG E CURVE FOR FILTER 3
1 H
1S»
30+
75t
901
I
N
r 105+
u
T
120+
E
N 135+
S
I
T 150+
t
STO- UllUOE HlflTISTICS FUR FILTER 3--
SIEC
11
V
10 A
I J }<
i:.' i:
13 li
i 4 i:
i •:. r
47 M
JU I
IV J
2O K
21 L
DENSITY LUG EXPOSURE
AVERAGE STD. DEW,
::»'.!. 4BS
MI.OAO
211,970
!:M.:.VO
! I 7. OI'O
I J . I /O
176.24'.,
Kill. 730
137.103
10U.100
UO.J70
SO.793
as.; i:;
2.0A5
.000
.000
.000
,000
1.276
1 .403
1.U78
i .:'j:i
1 .258
I .490
i. v.ri
2.249
2.473
2.V71
3.572
4.U47
O.AHA
3. I 3'J
.000
.000
.000
.000
-.210
- . 360
-,SOO
-.640
-. 790
-1 .0110
-1 .3110
-1
30
1 .6110
1 .1130
1 .990
. 140
.300
.450
.610
.760
.920
3. 080
3.230
V 165+
A
L
U 180+
E
S
240 +
155 +
-3,00
+
-2.50
+ 1 + .,
-2.0O -1.75 -1.50 -1.
OUITUT LOO EXPOSURE UflLUE
+
-1.00
-+—
-.75
.00
Figure 43. Output from program DLOGE.
-------
0 n
a
in
-------
t
N
P
U
T
[i
E
N
S
I
T
r
V
A
L
U
£
S
DATA FILE: JULY 24,1974 sooo FT WIT SCANNED AT so MICRONS
O-LOG E CURUE - 0 70 253 OUTPUT VALUES
0 + 3
.2 3
.12 3
15+1 23
123
123
30+ 123 '
1 23
I 13
45 + 1 23
1 23
1 3
40+ 1 3
1 32
1 3 2
75+ 1 32
\ 3 2
1 3 2
90+ 1 3 2
1 3 2
1 3 2
105+ 1 3 2
1 3 2
1 32
120+ 1 3 2
1 32
I 3 2
135+ 1 32
1 3 2
1 32
150+ 1 3 2
1 3 2
1 3 2
145+ 1 3 2
1 3 2
1 32
ISO* I 3 2
1 32
1 32
195+ I 33
1 23
1 23
210+ 1 23
1 2 3
1 23
223+ '1 2 3
I 2 3
1 2
240 +
(
t
253 +
_
3
21 3
3
3
3
30 43 60 75 90 IDS 120 13S ISO 145 180 19S 310 223 2AO 2S3
OUTPUT LOO EXPOSURE VALUES
Figure 45. D-log E curves generated by program CORRECT.
101
-------
value is corrected by subtracting the look-up table value corresponding to r
from the pixel's log E value.
If the table option is chosen then the effect of lens falloff can be
expressed as:
E(r) = E(o)T(r) log E(r) = log Eo + log (T(r)) (2)
where:
T(r) is the falloff factor at distance r.
The T(r) values were measured at Johnson Spacecraft Center for a variety of
lens, filter and F/stop combinations. Each table represents one element of
the film, and corresponds to one lens-filter-F/stop combination (Kalman and
Scarpace 1979).
The tables contain T(r) values for r going from 0 to 38 mm, in steps of
1 mm. The FALLOFF program interpolates in the table, and generates a look-
up table of log T(r) for r with this option. The distance of any point in
the data-file from the center must be < 500 pixels. If the cosine option is
used, the distance from the center to any pixel must be < 1650 pixels. The
log exposure value and distance r for each pixel in the data file are
computed. The log exposure is corrected by subtracting log T(r). Corrected
log exposure values are stored in a data file.
Transforming and Enhancing the Data—
Several programs are available to transform and enhance data and
separate vegetation response patterns. Two of the programs which seemed
particularly useful are NORMRATIO and SMOOTH.
Program NORMRATIO calculates a normalized ratio of the bands of film or
scanner data. The program calculates
A - 255 ._
A " (A2 + B2 + C2 + D2)L/2 C }
where:
R. is the value representing the calculated value of the normratio
for band 4, and
A, B, C, and D are the values for bands 4, 5, 6, and 7 for each
pixel.
102
-------
The program redefines values below zero or > 255 to zero or 255 and places
the calculated value in the appropriate bands of the new data-file.
Program SMOOTH can be used to generalize density slice or classified
scene data, smoothing out aberrant pixels and generating a more map-like
classification than the original classification.
Each pixel of the classified scene is processed in a sequential
manner. Each pixel and its immediate neighbors are examined and a
determination is made as to whether or not a majority of the pixels belong
to the same class. If the majority of the pixels belong to the same class,
the central pixel of the neighborhood (i.e., the pixel under examination) is
assigned (reclassified) to the majority of class. The one exception to this
generalization scheme occurs when the majority class is the "unclassified"
class. For this case the central pixel is not changed, but retains its
original classification.
Selecting Training Sets—
Once the data are corrected and transformed, training sets of the
various resources are selected. To assist in the average set selection,
program SLICE was used to generate 10 and 36 level density slices for each
band of data (Figures 46 and 47). A 10 level slice is most helpful for
identifying broad patterns. A 36 level slice should be used to delineate
the details within broad patterns for training set selection.
Training set selection is extremely important as a supervised
classification can only be as good as the training sets which are selected
to define the resources and form the basis of classification (Wacker and
Landgrebe 1972). In this study training sets were selected using a 36 level
slice. Any area on a density slice consisting of 20 or more pixels with the
same symbol was a potential training set. In most instances training sets
could be confidently Identified as a particular vegetation class only when
an entire scene was classified, making it possible to see all the locations
designated for that training class.
The actual location of a training set is specified, by entering into
the computer density slice row and column values of the apices of the
area. The computer lists the name of each training set, its corrected and
uncorrected apex points and the number of data points in each training set
(Figure 48). The number of pixels necessary for each training set should be
between 10 n and 100 n where n equals the number of spectral bands of data
being used. Training sets for one vegetation class should come from all
over the scene being classified rather than using one large training set
located in a single area. The spread of training sets over an entire scene
for a single vegetation class increases the change that the training data
will be representative of all variations in the data throughout the scene
(Lillesand and Klefer 1979).
103
-------
JUHC 1977 CI» 5000 FT AMT 00 MM LENS SC4NWEO AT 50
R0*3.« 4 TO 302 COLONS-- 63 TO 77o
EVE»r 4 »0«S A«JO COLUMNS »»E SHO*»I
SLICED ON B«NO 6
90 135 180 225 270 315 360 a 05 050 195 5«0 585 6)0 675 720 765
» . * . .*. * ...*. t .*. ..*.. .*. * . .*. . t ...t. . * *. *
9IMIMMMIIIMI 9
miMIIIMMIMIfl 18
27[|IMMIMMIIMM 27
)6fMMMMMIMMIM 36
OSIIIIMIMMMIMIMII 05
5a!|IIMIMIMIMIMMll 50
>>S riMIIIIIMIIMMMlMfO 63
72 tlMMMMIIIIMIMIIIMI 72
8| IMOMMIMIMMMMMtlllt 81
40 (MltloalMiailMIMMMMIII 40
IDA [IMMMIMMlMMIIIMIIMMIUIMM 108
117
126
135 [MMMMMMMMMMMMMMMM l»ll«IMMMMM»ll 135
1 00
15)
162
i [|Ml|MtMM»MMBtal*lMMMlMMt»lMl**IIIMIMItlll 171
1*0
184
148
207 III MIMIMaMMflMlllMMMIIMMIir.il Ml Ml loMlltMIO*!! 2A7
216llMMfM«tMMMIMMlMllllMlMIM«MaMMIIMIIIIIMIIIIII 216
225 [tMIMMMMIMIMllMIMIIMMIIMIMMMIMIfOrilMMIItlllMI 225
2)0 (IMMMMMMMMIfMMMMIIMIMMIIMMMIM»MMMat«fllMI|ftlM| 230
203 (MIMMMMIMMMMIIIMMIMIMIIMMMMMIIIMMIMMMMMIMQIIMIMIMM 203
2S2IMtMtMIMIMtMIIIIMIII»MMMMMIMMIMMIItllllMIIIIIIMIIMIIIMIIIMI 252
270 riMMIMIMMM.MIIIMMMMIIIIMMIIMMIMfllMflllMMMIMMMMMIMMM 27n
279 t. , MIMMIMMMHIMIOIIIMIMIIIMMMIIMMIIIIMIIMMMMMMMMMMIMn 279
.•MtlllllMattMtlllllllllMIMItMIIMMIMMIMIMIIIIIIMMIIMO 288
,,,,,, ,', ,ni|lllll||MMatlMM«MMMIIIIM,'.MMMIMIIMMI*MO 297
306C. ..,.,.,,,»,»,*,. .. ,.,.,..,, ,MIIIII|M»MMM9HliHMMIMIIMM»Maitia 3"6
51Sr..'...,,,»,,,«»»»,,,.,. .,,....,.,. ,. .. '. •aMIflllltMMMIMMMMMMiaiso 315
lilt/..I',',',',', .O.U.,,,',)',!','.',!'.!..I!!'.!.' '. '.. . '. '. .MIIIIIIIM* 333
302( . . . 302
90 135 1*0 225 270 315 360 "OS 050 095 540 585 430 675 720 76*
VALUES PELO« 10 »»E BL»w«
10 35- 60 85 110 155 160 185 210
TO TO TO TO TO TO TO TO AND
30 59 40 104 )]• 159 180 209 UP
... ... »»» ooo a*a ••! 1*1 in in
,., .» »»» ooo **t ••• MI in in
... ... *»» 000 •*• ••• !•• Ill lit
Figure 46. Overprinted 10-level density slice.
104
-------
JU^C UtT CI« 5800 rr »"T «n «« It** SCfteO 4T 50
RP»S.. 1 TO J02 COLU-SS-- 6) TO 77«
£VE»» » «0»5 «"0 COlU»xS »»t S^OKM
SLICES OK 81X0 t
«0 115 110 225 270 115 J60 m) aSa ««S SUO 5«5 610 675 720 765
Of
90 t-v«it«'*»JO*'»J'*t*'L '0
1J5 [3N"»0|.>'f»0«0'.3flC1'00iOJJII'«LLI»0 US
1 80
< ^0"'."OlJ-L«""''*'Ol)0>.C'L1""-1''OOvO"OI>OILNOL''>'0-0«JO''LILfLOr)L"' 10" JLL1
27ft (SJ JI JJJtH»«PL\.^!*)***4««CP**onoiICM«*NPLO«OtQp%'C3»*0Mk**»CPscsP90»<»«*in.OsPS*«**MM««o\.'*l>*JOl«Mt 27ft
TO flCOVUV«iSLJ-«F*'Gp3
t 10 t OTCUWTVTZ
(III 2lSo«50iia5« 0«
(I 01J»«!«»&»6B57!8 01
lt5IJJ01J!»58kH«787HOU2 n'OMIIIlS JO
I 1 Si!»jaS«»?M79«S3?5aMO«JJU21 J2 2 1 I Z2'55vvy»TrS«JU II"»L"J»IS
!MUJ675<>B«8»»»»HO»l,6J5Jo5b06«»2Z J Zl « » J
I
•0 IIS 180 225 279 111 }60 «05 PP aaa »«• >J» TIT
rrr gee MHM in jjj no ILL MUM NNN ooo P»» ooe »•• ssj TTT
rrr tst HHN tit JJJ ««« LLL HMN NNN ooo P»* aoo ••• »ss TTT
tit 227 114 241 24» 255
TO TO TO TO TO 1X0
tit 2J1 140 J«T {5* UF
uuu »»» "•• >n Trr lit
uuu »»» »«• «> TTV tit
uuu vvv ••» m rrr HI
Figure 47. A 36-level density slice.
105
-------
0XOT R3*FS.TRAIN
TRAIN VERSION 2.1 — 09 JANUARY 1978
RUN AT 21105:57 ON 03/10/79
3HQU THE TRAINING SET PICTURE (YES OR NO)? —
NO
DATA-FILE IS
SEPTEMBER 1975 C.TR 5000 FT AMT SCANNED AT 50 MICRONS
NO PREVIOUS TRAINING SETS.
APEX LIST? —
420 224 -MO NXN240 •
MORE APEXES OR CLASS NAME? —
A WATER
CLASS 1—A WATER
UNCORRECTED APEX POINTS—
( -128 f 224) ( 440 r 240)
CORRECTED APEX POINTS—
( 42Sr 224) < 42S, 240) (
221 DATA POINTS FILED IN
440r 240)
FILE 20
( 440» 224)
APEX LIST?—
330 194 334 240
MORE APEXES OR CLASS NAME?—
A DIKE
CLASS 2—A DIKE
UNCORRECTED APEX POINTS —
( 330» 194) ( 334, 240)
CORRECTED APEX POINTS—
( 330 > 194) ( 330, 240) (
235 DATA POINTS FILED IN
334r 240)
FILE 20
( 334» 194)
APEX LIST?—
400 110 410 130
MORE APEXES OR CLASS NAME?—
A WATER-
CLASS 3 —A WATER
UNCORRECTED APEX POINTS —
( 400» 110) ( 410» 130)
CORRECTED APEX POINTS—
( 400r 110) ( 400, 130) (
231 DATA POINTS FILED IN
410r 130)
FILE 20
( 410, 110)
APEX
250 1
MORE
A C,D
LIST? —
10 260 120
APEXES OR CLASS NAME?—
CLASS 4—A C,D
UNCORRECTED APEX POINTS—
( 250, 110) ( 260. 120)
CORRECTED APEX POINTS—
( .250, 110) ( 250, 120) <
260? 120) ( 260» 110)
121 DATA POINTS FILED IN FILE 20
Figure 48. Listing provided by program TRAIN.
106
-------
Programs CLASSBAR, HSGRAM (histogram), and SCATTER (scatter diagrams)
are used to view and better define training set data. Program CLASSBAR
prints out bar diagrams for each spectral band in order to show the mean
spectral response of each class and the variance of the distribution. Mean
values are designated by a 0 and standard deviations by 1, 2, and 3
(Figure 49). These bar diagrams show where classes overlap in spectral
bands and which bands can best be used to discriminate classes. While bar
diagrams show how vegetation response patterns relate to one another or how
badly they may be confounded, they are not detailed enough to define
specific boundaries (Buchanan 1977).
Program HSGRAM plots histograms for any or all training sets filed
using Program TRAIN (Figure 50). The histograms plot the number of points
in a training set for the range of brightness values displayed. Histogram
output is important when using a maximum likelihood classifier (which
requires normally distributed data) as a histogram provides a visual check
on the normality of training set data. Separate histograms are produced for
each color band, permitting fairly exact spectral selection for training
sets. If multiple vegetation classes are included in a training set, a
histogram may show a bi-modal or multi-modal distribution. Sometimes a
bi-modal distribution stems from different illumination conditions within
the same training class. The resulting subclasses must be separated to get
a good classification.
Program SCATTER can help separate training sets that include two or
more vegetation classes. SCATTER produces scatter diagrams of training set
values, plotting two bands of brightness value data against each other.
Multiclass signatures often show up as separate ellipses in the scatter
diagram. Exposure value can be determined for all three bands by viewing
scatter diagrams (Figure 51). Use of the above programs allows the user to
determine the quality of the selected training sets.
Training sets can be modified by use of the programs MODSET
(modification of training sets), CLEANTR (cleans training sets), and CLSTRN
(class train). MODSET offers the following options: 1) several training
sets may be merged into one; 2) selected training sets may be deleted;
3) the name of a training set may be changed; 4) training sets may be
reordered; 5) training set reponse patterns may be restricted to within a
specified number of standard deviations; and 6) arbitrary upper and lower
bounds can be imposed on training sets whereby extraneous pixels are
removed.
Program CLEANTR was designed to provide a statistical clean-up of
training sets. This program allows the analyst to specify the number of
standard deviations beyond which data values will be deleted (Figure 52).
Program CLSTRN creates two matrices of all training sets selected for a
specific classification. This program runs a maximum likelihood test on
each training class* pixels as to whether they are most similar to
themselves or to other classes defined by other groups of pixels. If the
training sets are independent of one another, most of the points in any
training set are classified as belonging to that training set. The first
107
-------
Cl» *nno *T »«T SCIUMFO IT )o »icnnxs
BIND »
1 1 1 1 i222?22222222222222i2>2tiiii
i«2i-oi'< j
l.J-l..o-1-J-.J
«
10
12
U
15
I*
17
Ift
2«J
1 ..i-0-l-J'l
1«2»IO-1»«1
} <2>-l—>«»>1—2* >
) 2«l--n-l«2 3
I I lllllll 11 I 111 II Ml Ml IIIMIIII I I 12222222222222 222222222222222
I 1 < I t??2'?JJJ'luoooll<>5S5<6»6o»77TT7««»«fl9i»iQgnnonftll I 11 22222))M1<>«>o|i^^^^S6666«77777eit|(>ll««««*OOOI)Ot 111 122222 JJJJ5«««««»M
Figure 49. Bar diagrams generated by program CLASSBAR.
-------
?n
pj-n (,
0.
2f
i2r
10 r
161
J2t
2" r
2* r
jo»
'j?S oni--TS T' rt'TfE
P A M?* c
o.
? r
fl f
10»
tar
i u r
16 r
i st
?af
?*r
2M
50*
at
i- r
» r
| 0 t«
i6r
IM
r
r*
r.
110 f
1M
u4 r*
so»«
szt«
sa t«
56t«
58 [•
6n««
6? [•
70*
7ZI
Tar
76 [
78 f
80*
82t
80 f
»6t
88t
40*
»?r
9«r
too*
lit
a It
«?t
u 'j r
lot
us f
Sat
Sir
62 r
*
-------
SCATTER DIAGRAM FOR SEPTEMBER 1975 C I R 5000 FT »«T SCANNED AT 50 "1CRONS
ARE* * .W*TER _ .
111111111111111111111111111 li 111111111111111111111 il
8. ?**.???*»"»pOoqpqppqp 1 111 I I I I_I >?22?ZZ2222J3J_3J3_J_3JlVtHt|(_V|_<«H15|
'
61 C 1 _ _ CM
60 C I l" C 40
59 C I II S»
_ _ _ _ _.. _ _5B
57 C I I 135 I
54 C 1712
55 C | 2|MM<| | I _ . _
54"~C 1316I53SS f
S3 C 3H1I3| II
S2C_I323M _
"51" C |i21279"i """ " ""
SO C I 21 I |75| |
H9 C || H|32 II
ts c i iis
S7 C I 2122 I
H» C 2 212 _ I
S5~C 2 I 1 j 2
tl C l|
H3 C _ \
s*
ss
5)
52
50
tt
17
HI
H3
I I 1 I 1 I 1 t II I I I I I 1 11 1 1 I 1 1 I I 1 I I I t 1 I I I 1 II I I II I 1 1IH I I I II
8a99^99999290gO.OOOQOQO]J 1 I I l_» I I 122?2??222Z333333333jtt'<'"ttt')1H5i
89Q l 23^5678901 23^567890 i23HS478»6"l 23 HS»78»0 I 23 H5»7«»0 I 23"tS478»or
BAND 5
Figure 51. Scatter diagram created by program SCATTER.
110
-------
"CC"'" «CLE»>'TP» -- vfojmv 1.0 -- 7 nc', 1977
ITC* Ifl
Tl'lEi jr-ME"??* 1*75 Cl° S0"i Fi »**T SCi^En fci 5f> 1«1C9P**9
TTF». ^^.O'POTw^S -rams STi*JO»«nnEVHTI
At f\ li I? t« f> u 1 */ BA a« a«. a* • * f*4P A«.
fi\ >• ^*t »* pHj T
CL'SS li > »'TE»
1 oo* ooli n
CL'SS 21 ' ni«t
1 225 225 "
CL'SS li > c.n
1 217 ?OS 9
2 2"» 200 a
J 20« 201 J
a 201 201 n
CL'SS at t ft, C.n
1 3M J91 |0
2 «?9S 2^5 1 0
3 283 7M 0
CL'SS 5l t F'S
1 5nS 09T «
CUSS 61 ' «.l
1 61 63 0
Cl'SS 71 ' .«7£B
1 16a 36a !i
CL'SS *l i M,T,J
1 if* in9 n
CL'SS 9| i G.H
1 69 69 0
CL'SS ICI « -'Te»
1 56" ^21 07
t 521 505 16
1 50? 5"5 0
CL'SS III OJ«f '.",
1 155 155 0
CL'SS !2l ' or«E
1 23 23 0
CL'SS |3| • C,9.£,r
1 158 H» "
CL'SS iai i E (ivLr
1 »H 9» 0
CL'SS Hi f «,L.-
CL»SJ i»i i L,".".n
1 9|» »95 21
2 «9S ««J u
j M2 «7T 5
a 877 »7T 0
CL'SS 17i e "IOSTL'
1 830 *2I 9
2 "21 "20 1
1 "20 H2« 1
CL'SS 1M * L
1 377 3*7 10
> 167 165 ?
"3 "
'5.1
90.9
9? I1
1 n c '. »
123.9
aft. a
1 ?l . 9
53. T
5). »
ST."
*»
ITS.T
???. fl
<)• .0
1 ft 2 B t,
,« "
i on. 7
l<»*'.^
l nq.fl
••
1 ft * . *
ina,r
^ ft« . f
™~j
\ u*i S
150.2
172.2
75.7
1 76. •
la t it
"»«. *
9*1 ,4
?fl*.fl
?5fl. 7
>*») .2
)«*..».
t n ( *
I » 1 • 3
150.9
15".
15".
IS".
'53.
151.
151.
loa.2
on n f
«.l '.0
115.5 .1
12n.3 ,0
120.7 .0
ill.* ',e
133.2 .0
7.1 '.0
132.5 . n
? . 8 '. n
.0 ,n
.0 .0
155." '."
l»n.a '.o
'23,1 .0
119,1 '.0
4 9t L Ok A
I ^> O n O * 1*
\ 2ft, ^0
120. ."
1 ?rt , .0
12". .0
us. ,e
Ml, .0
1ST. ,rt
12T.S ."
n v
'"'.»
"'.5
0.0
J.n
I'.1?
'".1
a . 0
0.0
0.0
58.1
!.o
o.a
0.6
7.8
7.3
7.1
7.8
. J
.t
.1
•J
n-j ^o
?.k 13'.1
o.fc 6'.5
1.9 6.1
«,! a,»
l.T 0.2
l'. 1 V.l
12.1 12.)
? . o **„ ^
2.5 ".5
?.? ?. fl
?.5 .1
06'. 1 ?O.A
.! 2.5
o.a 3.5
a. 5 1.5
e.i O'.T
7, 0,6
7 0.4
7. 0.6
3 , a . A
1, a. 7
1. a. 7
3.5 «,3
." 3»
.0 216
.0 80
.0 8a
.0 12
.n 83
.0 90
.0 95
.0 !!6
.0 -a
.n no
.1 aa
.0 00
.0 oa
.0 30
,C 219
." 08
.» 91
.ft Al
.0 »2
.0 82
,0 83
.0 '3
.0 92
.0 92
.0 92
.0 83
273
l"l
106
ini
|ia
11!
US
115
131
• 5
1)0
1 ?i
1 C O
(.0
*0
60
321
205
110
Ita
li>5
121
119
118
1 1 A
118
117
117
107
B'10 5
90
132
112
113
133
135
1 16
1 i "
13*
102
102
160
-3
172
'
8*
a*
8*
93
?53
130
'35
131
130
131
131
111
105
105
1U5
135
136
255
155
I5a
153
15)
155
15S
166
166
166
190
15*
192
t Tfc
I f D
101
101
101
32)
25*
15?
15ft
\ 52
172
170
169
169
1*?
162
162
153
153
»-") 6 S'so 7
-2)
1*6
121
120
120
I 1 0
no
110
til
HI
125
-22
125
• • y
I ' *-
-17
-5
0
9)
170
IIS
110
119
109
109
109
109
122
122
122
117
117
02
22)
152
151
151
131
132
111
LSI
131
15)
152
152
101
19
100
| 91 L
1 1 •
20
*
0
217
187
132
129
15Q
t j *•
1 1)
132
132
132
las
145
ia5
138
139
0
ft
0
0
0
0
0
0
0
0
n
0
0
o
0
0
0
0
0
0
n
e
0
n
0
0
0
0
fl
n
0
0
0
0
n
0
0
0
0
n
ft
0
0
0
0
A
1
1
n
n
0
0
0
f)
ft
0
0
0
(t
n
n
0
p
Figure 52. Output from program CLEANTR.
Ill
-------
matrix lists how many pixels selected to define each class truly define that
class. The second matrix expresses this same information as percentages
(Figure 53). This program is a. particularly useful training set
modification program when used in conjunction with program CLASSBAR.
(CLASSBAR displays the range of brightness value for each training class for
each band (Figure 49).) Program CLSTRN suggests which classes might
successfully be combined while program CLASSBAR displays training set bounds
for each class, allowing the analyst to decide whether or not it is
reasonable to combine one class with another.
Classifying the Data—
The basic concept behind supervised classification programs is that the
spectral response signature of each pixel is compared to the spectral
response signatures of the training sets and classified for the best fit.
Three classifiers were used in this study—a parallelepiped, or box
classifier, an elliptical classifier and a maximum liklihood classifier.
When using a parallelepiped classifier, the lowest and highest
brightness values for each band that characterize the different classes are
entered into the computer (Figure 54). These values define a rectangular
area in a two band scatter diagram or in general for a three dimensional
parallelepiped, hence the name parallelepiped classifier (Figure 55). Each
pixel is classified according to the classes defined. If a pixel doesn't
fit any of the classes, it is put into an unknown category.
The parallelepiped classifier is a fast and efficient and therefore
inexpensive classifier. There is a problem with overlap of classes,
however, and if a pixel falls into one of these overlap areas, it generally
is arbitrarily placed in one of the two classes. To quote Lillesand and
Kiefer (1979), "Overlap is caused largely because category distributions
which exhibit correlations are poorly described by the rectangular decision
regions. Correlation is the tendency of spectral values to vary similarly
in two bands, resulting in slanting clouds of observations on scatter
diagrams." Where there is correlation, rectangular decision regions poorly
fit the class training data, resulting in confusion for the classifier. The
overlap problem within a parallelepiped classification can be helped however
by breaking the single rectangle that defines each decision region into a
series of rectangles whose stepped borders more closely describe the
distributions (Figure 56).
More sophisticated classifiers rely on the statistics of each training
set to classify a scene pixel by pixel. A maximum likelihood classifier
assumes that training set points are distributed normally and with airphoto
and multispectral scanner data, this is generally a reasonable assumption
(Lillesand and Kiefer 1979) (Figure 57).
When using a maximum likelihood classifier, training sets are evaluated
to determine their category spectral response pattern. This pattern can be
completely described by the mean vector and covariance matrix (which
112
-------
Ul
10 C M,l
CL*SS »i»"F srzf 'r»er.nii6r nf POINT* CL»ssiFifo «s cuss
i t r,f is"
? C I,J 166
j c i s«
a C J 166
f C C,M 1U5
6 C t,f aae
T C M la
» C «nSTl.» J 161
« C «,L 119
n e r,j.« in'
12 C f,6,M 11'
II C J,«,L 1ST
la CO 16)
15 C «< i«
\
'.ft
'.n
>
9.7
.0
3.1
>
2 *
p) I | t Q
V.ft 1?.7
O *.*
l> 1.1
'.o ,o
'.o '. «
> .0
u
rt
?fl c
10.0
1.2
*'.5
.0
'.o
s
.0
•'.1
'.0
'.0
.0
6 7
01 a
'.« ».J
:« ».»
!»'.' 2.1
.0 .0
l'.» .0
.0 .8
A
6.1
r.»
.0
.0
'.0
*
• 0
.0
'.0
',0
.0
.0
•.B
10
51.0
I2'.t ^0
'.0 1,2
5.1 .0
O
1.2
«.i
22.0
.0
77.7
.0
1.1
1«
.0
.0
.0
.0
i.a
'.0
M.I
.0
15
',0
.2
»'.»
.0
l.»
.0
«].!
Figure 53. Percentage matrix from program CLSTRN.
-------
TITLFt JU"£ 197? CI1* 50PP FT AMI itt\ -A" i.E^
FILE RESOLUTION! 1 PRT^T RESOLUTION! 2
AT SO MICRONS
CHAR»CTEP
w
2
•*
4
S
6
7
ft
9
i
8
C
f>
Z
F
F
G
M
w
I
J
f>
1
M
U
n
PF
A
A
A
A
A
A
«
A
A
A
n
n
R
A
R
R
r
r
r
r
r.
A
A
A
A
A
SOURCE
••IATER 0
6.7
.MOSTLY 7,s
«.c
SANO^LIw
n,R,S
E.r
G
A. 9, A
E,F,n
H,R
I.J
J.K
^IKF
F
TRSTHP
MOSTLV P
-------
PMf "§|
• «-«i» T a n
pj i*n TI l?fl, c **LM"^B 3"" T* 51"» "Tt.1 *t •*) tTf o E v(MV ^ o***, 4ift J C^Lr"*''S( US t *G 1
C* »*• Jcvf I'll C T* ^*fl' *' *"t »« •« tf^S *c*****?ft ** ^0 "ICB"**S
11 JttO *•» »ll T(» f
0au«if'*i«r-'-c''r^'i:r4»err*»r*r;?jccCCt.UCtCr
^j,*;„„_--,--_j?jv.«»>»,-wBa*«»f M-«J"«l •<«».* a«ern.t*??t"<»t*B<»h|
2-"'CrCCCC''"M"--?»iCL"'LCi?Cfc*'*FrC''-<'-g»'"'11111 lC"l?«C*?E*l!L''D4rr|
01l1"F»LLC«lLlLl.>*^"*»***4»(FCCff»* ! ")F "tt f ?rCCC'LCtCt"-k*<'??CCL -5??*»* B*^13C""ti*1 IL J) I 1 ?t?t£"? *••?•!'*'* I I It* I
Dill 3C»t.Le«"*"r«LB«'t*1|.4i«'«'':Cf CCCCCl !C*--lClCrrcC':cI5?w"l'-'CCCC?B?*a?CCa9**f*""*lltltani?lt I«-C«C£«2r4«ei« JCl-
»* I
f Fr»i--«cret "lu*: IF **•••»• *incr«:L
;f ^OCl
1L TJJLW-1"'*C'"'L* •***'•'"-lJe*f2*.^l3*3'*OCCSC?2*BT*313*7J''*ta"CC'5^(l**C5i''"B55";T?«P*('H(HPSaaJLl.C'4 »4dH*'4iB*i
8*«ViJLJJl,l'**CCC*'f*1*i'*3^3^'l'**Z^*8*€C**BttmsC3*"''tt44*«}*t«C"^l"*S**L€4'*S*fi^*?**ll.****llU*3BEO****'"'LJ**l1'l
•I
• I
Pi
II
039016-00080031— JMJJJJ-H.7L- ————^H--N-«— a i^CtC » «"t'M-»c--« *r 111DtMBlimi
DoCflan-IinTOtnaonflOBftO^noi—-*J7 »—«•»—•»•—»-—M[.TLCC>"'C--"CC<'<**<*l*-«9t 11 111 1111 I 11*1 1 33034«B444*DCI f
wjj«—.- —-^c-CCCCF.CfC«'"'-tCCCL —
tt&0&flOni«*''"4l«JJ,JJl.Lt««—«"'«»*H«Hit«»<
011111 ft 11 {MM 1111 i{t JMI ill* t inn it 111111 i"»-'"*"'«ooni8(ji«fi8iiiO'*'*-ooee«oooni---JJ»L>'-'*1*-
110 oimMiuiiiMMiiutinmiiiiiiiiMuiiijIIJMIIIHIIJ--J"-?--''!'''"!
lMM1Ml\11U111mil11M1\llll11MnVlMnUMilHlllUt'J-
9i 1111111 in 111 mi i ittni tin 111 in ii i jit iij 1111111 ii i H i' i nit ti ti i ii i mi
SMI Ml MM tMlllllMIMMtllllUMtMtllUIll t HMlllllMMIllll|llllllMMllllllllllllN-----nOODO(J()8800»
t«11111111 U11111111H111 Hi U U U11H1»111111M 11 M1 M H11111M l H1U11 \ i i 11 Ul 111 H1 H H U U11» H H«M«MMM| t
ei01Mniltl)lMMMMMl1|lll1IIMllHMIMUI t JMMMMM11IHHH M1MIIIM11|1|1UII11|I|I11IH tllHUlI'
• *. .1? .1* « - >».
Figure 55. Portion of printed out parallelepiped classification
created from program BOX4.
115
-------
w
0)
c
"re
5)
n
-Q
C
(13
m
iw w '
L "jw vv^
|W W
1 ww!
* II u
u u
u u u ..
u y u
ii U U
U U L
u u u u u
u u u u
1 1
u u u u
L U ,-J
i _ O _^
1 ^ Q
t 0 _.
1 o ^ S!
* 0 '
' O S '
J"'fl
• ccSJ
CCi
/- r '
1 C .V.-1...
i c. r LJ i
tf* /™*^^« i »• J
£C-j j H H
"H u
H H "
H "
H H H
HfcA
LJ
H ^- J
f i ^-*-
i |- -.
• Fp F FF;"'
iF F F- FF|
TF FCFF!""J
t F :
Band 4 digital number --
Figure 56. How a decision region can be broken into rectangles whose
borders closely describe the training «et distribution.
116
-------
0 EEF«C9a9787Fari 188
|9HlCaaja«F7CFa8778£HH"»El»EJ»iCE»«JFFIln nrfiwGGCl 6C7I
200J99»JJJ(>JJSF767E'»Fa a»E29«M£IHH9 5CJ8FZH nOIEE'tIZ? 1E» 200
l2H98Jaa8FGJJ«»29EMM»M999?91tME9J8Cf;9G»« ICC»FFEE79Ei
iG8F»EJJJ8G*9GFFJ299 H9C J90F«»w8F2a3«Cfla77Hnij Q HFTTfl
•|heaJJJjaF»«GJ9F9i"»J» "EFG?G>9C2ftaiC»»CS77780 00 F77EI
|FaaajJ4a8«C»2ECFSE9C<» »*«aJi(>C?OCJCCa8778EGtI I2CG7ai
220iC«85»JJaGa*»GFECCaaF JA92»r.JR9fl8«85f ""aFceFf G IGF7S7I 220
"|887GG6JJGC«JEGCG»GJ'"»*HI oaF8F809C 5F.a9CG7Pa*eGHrfM«aCl
l»aa»«JJJ999r?27JJ 8«8»E2?« G JGGU8291»EI»9822W» »E
J9H9FF?89C218CCJ« JHF7
J3?EE««7?C?C879778W9I
FuFaa9C»CC877U277CCGF
08999238CCC8'>77C7C77F
79J9 J«8F.8CCCE777C7777aai
G77CC!F289iCJCJCCCCJ 9C9fl8*8CCF8 JC»«EGFC7778I
GjnEa9laa7CSFF9CEFuc«2i:C77?928CFPT.7CE8JFJFaJ2£7G*GGFi
GDOJJ8FttF7C17FFC87ft2»8»777l»C8C»3E9<>8P8o3C»J9C87C7BEPl
280
jF8J«ajCCC77*7C«C 778 7 7789298 7FJCG 02»Ca«CC8a98FFFGF2EOI
'|99929a8Fa8FoCC8CCC77777«Cfl«CCHC78SCCC7CC1'92f8*77GGGC I
i6991ii999E«l«91J99aaftCC2t 2G ?C78a8C8»CaaF80777J9?HH J t
iNMO«NNNM JJJJ9J»*»*9»*»a99aaaFa
-------
describes the variance and the correlation of each training set). Figure 58
shows probability values plotted on a three dimensional graph of a scatter
diagram. The vertical axis expresses the probability of a pixel value being
a member of one of the classes. This creates a bell shaped surface called a
probability density function for each training set. These probability
density functions are used to classify an unidentified pixel by computing
the probability of that pixel's value belonging to each category. The
computer evaluates this probability for each category and places the pixel
in the category where it most likey belongs. A probability threshold can be
set so that pixels with a low probability of belonging in any class will be
put into an "unknown" category. Program MAXLIK is a maximum likelihood
classifier.
A third type classifier is an elliptical classifier. This classifier
relies on the statistics of each training set to classify a scene, pixel by
pixel. Training set data are analyzed to determine mean vectors, eigen
values, and eigen vectors, and covarlance matrices. A table, compiled from
statistics generated from the training sets can be used to classify the
scene by comparing each pixel against the look up values (Figure 59).
Program TABCLASS is an elliptical classifier. An elliptical classifier is
more expensive to use than a parallelepiped classifier but less expensive
than a maximum likelihood classifier if only 20 or fewer classes are used.
Steps Taken to Generate Computer-Assisted Maps
Figure 60, 61, 62, and 63 are the results of classifying data from four
scanned airphotos with a maximum likelihood classifier. Three of the four
airphotos were color infrared transparencies taken 25 September 1975,
24 July 1976, and 25 July 1977. The fourth airphoto is a color transparency
which was included in order to compare classification results using color
and color infrared film data. All photos used for this computer-assisted
mapping (scale 1:38,200) were scanned at 50 pm so that the ground resolution
of each pixel was 1.9 m.
The steps taken to reach film product renditions of these four classi-
fications were as follows: each photo was scanned using a 50 \im pixel size
through a blue, green, and red filter to extract blue, green, and red film
layer data. These data were corrected using programs DLOGE, CORRECT, and
FALLOFF. Only the July 1976 data were not corrected for falloff.
Once the data were corrected, a 26 character density slice was run on
each scene which was used to select training sets. Between 50 and 80
training sets were selected from the data in Bands 4, 5, and 6 for each
airphoto classified. Training set selection took 3 to 4 h for each scene.
Training sets were entered into a training set file using program TRAIN.
Programs HSGRAM, SCATTER, and CLASSBAR were run on each training set file to
select the "cleanest" training sets. Program MODSET was used to delete
training sets which appeared too broad or scattered to be useful. At this
point each training set file was copied three times. Program MODSET was
used to delete all classes except those selected from one band. Thus the
118
-------
Sand
Hay
Forest
Water
Figure 58. Probability values plotted on a three-dimensional graph of a scatter
diagram (from Lillesand and Kiefer, 1979).
-------
«T»BCl»SS CmSIFICATI3«i>i JUHf 1977 CI» 5000 FT »»T (in HM n»3 SCi^E') «T 58
100 320 . 300 Jk8 38« BOO .428. ««n «bft u«o .580
t.--.t->--t----*----<-*--*----«----«----*----»---*«->
IAD
Fll
I99HI
p
K
BIFF9£!ai 260
I 7COOL7 K99FFJIJ8CJFJFJFFFF|;0
I «S3«aaraI9r39IIFB9R2C8B999»FRFRlMCC9R8a3FBGCFB9F9B»II
28oi9999'F99999RFpAFFRF9a3FFF9FFCR2RBfl99XKKKO|
t7CC050CCCMrirncOacCa«8FF2l< 2 2F99a8F»8FaaIB»999RC2L "I
iEtESSSSEOOMOCDCO-riOOcnOOaCCaaalaaR f B -BFI I89FFI«F9'J|
«2 « " -2i Joo
I 0000000-
180 320 300 1*0 38A aoo a20 oao a*n aM) 100
Figure 59. Portion of a classification of the site using
program TABCLASS.
120
-------
Figure 60. Computer assisted map made from Figure 61.
CIR airphoto on September 15, 1975.
Computer assisted map made from
CIR airphoto on July 24, 1976.
-------
NJ
ISJ
Figure 62. Computer assisted map made from
CIR airphoto on June 25, 1977.
Figure 63. Computer assisted map made from
color airphoto on June 25, 1977.
-------
master file contained 81 training sets but the copied files were modified so
that they contained 27 training sets from Band 4, Band 5, and Band 6,
respectively. This made it possible to combine similar classes from the
same band. Classes were always cleantrained once they were combined to
discard any pixels beyond the 2.5 standard deviations of the mean values of
the training set. Once similar classes from each band were combined, the
same changes were made in the master file, and the copied files were
deleted. At this point programs CLSTRN and CLASSBAR were used to combine
similar classes selected from different data bands.
Once satisfactory training sets were obtained, the values could be
entered into program BOX4, a parallelepiped classifier. Inexpensive to run,
this classifier points out classes of marginal use. (If no pixels or very
few pixels are assigned to a particular class using program BOX4, that class
should be discarded before running classification programs MAXLIK or
TABCLASS.) Running BOX4 also displays any unclassified areas. Training
sets should be selected from any unclassified areas and put into the master
training set file so these areas will be classified thereafter. These new
training sets must be cleaned and displayed to see whether they will be
useful or redundant. Processing training sets to use in the TABCLASS or
MAXLIK classifiers took 4 to 6 h.
Using the September 1975 training sets (81 classes reduced down to 22)
the analyst ran the TABCLASS and MAXLIK programs on the corrected September
data. Since MAXLIK was less costly to use and provided as good a
classification as TABCLASS, program MAXLIK was used for all further
classification.
Next, program OUTLINE was used to outline the actual study site area
within the data-file. The accuracy of delineation of this new area can be
checked by running program SLICE which reveals the outline of the newly
created area. Program MAXLIK was run on the outlined data-file to create a
classification, followed by program SMOOTH which was run twice to make the
classification more readable. The July 1976 classification, the most
garbled of the four classifications, was smoothed three times. The next
step was to run program SLICE specifying the number of classes to be
displayed and the symbol to be assigned to each class. Several vegetation
classes had more than one spectral response pattern. Each of these patterns
was assigned the same symbol, reducing the 22 to 24 classes entered to the
eight to 10 classes which can be identified on the classifications.
Once the classification was labelled as appropriately as possible,
program MAXLIK was run on the corrected outline data-file at resolution 1,
putting the classification into a file without making a printout. The
classification was subsequently repeated the appropriate number of times.
Next, using program COLOR, selected color values were assigned to each
class, giving classes which defined the same resource the same color.
Lastly the analyst ran program FILM70 which writes these color values onto
computer tape such that the scanning microdensitometer can read and convert
them to film densities, band by band, which are written directly onto
film. The resulting three color separation film chips (blue, green, and
red) were put into the appropriate separation and combined onto a viewing
123
-------
screen. Figures 60 to 63 are photos of the color composites of each
classification created in this manner.
Computer-Assisted Mapping Results
Figure 60 to 63, color renditions of the four computer-assisted classi-
fications, display approximately the same ground area. Both the MAXLIK and
TABCLASS programs make it possible to select and enter an on-the-ground area
of pixel size and from this calculate the total area classified as each
vegetation class. On-the-ground pixel size was 3.6 m for each of the four
classifications in this study.
Since the areas considered in each classification were only approxi-
mately the same, the analyst returned to the original four data-files and
used program OUTLINE to outline an exact polygon (shown by the black
outlines in Figures 60 to 63). Total areas of each vegetation class with
time were compared to demonstrate change with time.
2
Table 25 lists area (in m ) and percent cover of the vegetation classes
which could be identified on each classification. (Not all classes could be
identified on all classifications.) The number of classes identified on
each scene is a function of: 1) which classes could be photointerpreted by
the analyst and 2) which classes could be defined using this method. (If
the analyst had selectively targeted the various communities, the number of
vegetation classes delineated using this method would be greater and more
consistent.)
o
Table 25 shows the area in m for each vegetation class for the
September 1975 classification, the July 1976 classification and the two June
1977 classifications. Total area classified/scene is about 246,000 m
(68,000 pixels).
On the September 1975 classification, eight vegetation classes could be
identified (Table 25) while on the July 1976 classification only six
vegetation classes could be identified. Seven vegetation classes were
identified on the June 1977 CIR classification; six vegetation classes were
identified on the June 1977 color classification.
Table 26 lists the agglomerated classes used to summarize computer-
assisted mapping results. The Carex lacustTris and Carex stricta classes
were combined into a Sedges class. The Spiraea/Sedges and Shrubs classes
were combined. Transition and Emergents classes were combined into a
Transition-Emergents class. The Other Disturbance class subsumes the Lemna
minor and Weedy Annuals classes. For 1975 no Degraded Sedges area could be
identified using computer-assisted mapping; on the June 1977 CIR
classification, no Sedges could be identified. Table 27 lists the color
assigned to each vegetation class using program COLOR throughout the four
classifications.
124
-------
TABLE 26. VEGETATION CLASSES USED TO SUMMARIZE COMPUTER ASSISTED MAPPING
Vegetation Classes
Shrubs
Spiraea /Sedges
Open Water
Tr ansi tion-Emergents
Sedges
Degraded Sedges
Other Disturbance
Dike
Unclassified
25 September
1975
9.7
18
0.5
22
39
-
1.8
6.2
2.8
24 July
1976
4.4
13
5.7
29
18
15
10
3.5
0.4
25 June
1977
CIR
6.7
9.6
8.1
17
13
21
18
1.9
3.9
25 June
1977
COLOR
7.6
17
6.1
20
-
24
20
4.8
1.2
TABLE 27. COLOR AND EXPOSURE VALUES ASSIGNED TO VEGETATION
CLASSES IDENTIFIED USING COMPUTER ASSISTED MAPPING
Exposure values
Vegetation class
Carex laaustris
Carex stricta
Spiraea
Transition
Emergents
Dike
Shrubs
Lenrna minor
Open water
Tr ansi tion-Emergents
Weedy Annuals /Floating Mat
Degraded Sedges
Spiraea/Sedges
Color assigned
Orange
Red
Dark magenta
Light green
Dark green
Olive
Brown
Yellow
Dark blue
Medium green
Cyan
Reddish brown
Magenta
Blue
0
0
200
0
0
120
0
0
190
0
225
0
255
Green
235
0
0
255
190
190
190
255
0
223
225
190
0
Red
255
255
200
0
0
0
230
255
0
0
0
255
255
125
-------
In the analyst's opinion, the September 1975 and June 1977 CIR
classifications are the most accurate of the four classifications. The
greatest number of vegetation classes were identified on these two scenes
when they were photointerpreted in conjunction with field data.
From field studies, the analyst knew that sedges were the predominant
species at the study site in 1975. The Sedges category includes the Carex
lacuetrie, Carex stricta, and Spiraea/Sedges classes. By 1977 sedges,
particulary Carex laaustris, had largely disappeared while large areas of
open water showed in places where previously there had been none. Much of
the Transition class had been destroyed by 1977 and large areas of Open
Water, Weedy Annuals and Lemna minor were appearing in its place.
In September 1975, the areas classified Carex etricta, Carex laauetrie,
and Spiraea/Sedges covered 143,000 m or 57% of the total area. By June
1977 the area classified as Carex etricta and Spiraea/Sedges (Carex
lacustris could no longer be identified) had dropped to 57,000 m or 23% of
the total area. Shrub area declined slightly. Combined area classified as
Transition and Emergents in 1975 came to 54,000 m2 or 22% of the total
area. By June 1977, although water levels were higher, encouraging growth
of this type of vegetation, Transition-Emergent area had shrunk to 42,000 m
(17% of the total area). Lemna area increased from 1.8% (4,500 m ) to 12%
(29,000 m2) from 1974 to 1977. Area classified as Weedy Annuals increased
from zero in 1975 to 5.9% or 14,000 m2 in 1977. Total area classified as
disturbance increased by 45% over the 21 months between September 1975 and
June 1977. Table 25 shows the amount of open water Increasing by a factor
of 18 or 13 (depending on whether June 1977 CIR or June 1977 color estimates
are used), from 1,100 m2 to 20,000 m2 or 15,000 m2.
The July 1976 CIR classification results are not as good as those of
September 1975 and June 1977 CIR for the following reasons. All the red
tones in this airphoto are very saturated, meaning that most of the vege-
tation was vigorous and strongly reflecting infrared energy. While these
red tones offer the advantage of being consistent across the scene (Meyer
1977), they are difficult to distinguish one from another. Training set
values in the blue data band were very broad for almost all training
classes. The analyst suspects the blue band data were not helpful in doing
this classification. Lastly the analyst did not correct this July 1976 data
for lens falloff. This may explain why the southern end of the study site
was adequately classified even before smoothing the data while the northern
end's data seemed badly garbled, even after smoothing. This caused fewer
classes to be defined, for many classes had to be combined into one class to
make any sense out of the classification results.
Classification results for the June 1977 color photo were less satis-
factory than results from the September 1975 and June 1977 color classifi-
cations but for different reasons. In this case the data were corrected for
lens falloff so the poorer classification results can be attributed to the
data being scanned from a color airphoto. On any color photo it is more
difficult to distinguish the various greens of the study site wetland
vegetation classes than to distinguish these same classes on a color
infrared airphoto. Water appears as a muddy green and is difficult to
126
-------
differentiate from the surrounding vegetation. Consequently the analyst has
least confidence in the accuracy of this classification done from color film
data of any of the four classifications.
Computer-assisted mapping as a method will be greatly improved when a
minicomputer, loaded with all the programs needed to do this mapping, is
available and connected to a color terminal. This will allow film exposure
data to be displayed on a terminal screen. Training sets can be selected
from the data and outlined and entered in the program TRAIN file by using a
cursor to outline them. This will greatly speed up this process.
A terminal screen display will also be invaluable for viewing a
classification in order to correctly label classes with appropriate symbols
or colors. The great advantage of the color display will be: 1) its lack
of distortion of the scene (which occurs with a computer printout) and
2) its color capability for differentiating classes. Using a terminal
screen display will allow the analyst to sit with a color print of the
scanned photo in hand and match the patterns on the photo with those on the
display screen.
A strong point of computer-assisted mapping is its ability to
consistently classify a scene. Error can be introduced, however, when
1) the analyst combines training sets (in order to have an adequate number
of pixels from which to derive satisfactory defining statistics) and 2) when
the analyst combines classes in an effort to make them meaningful. If the
training sets cannot be identified when they are selected, the analyst hopes
to identify correctly the pixels which were classified using those training
statistics. The analyst must assign the same symbol to those classes which
are the same resource.
Computer-Assisted Mapping Summary
Computer-assisted mapping offers consistent classification of vegeta-
tion classes and quantification of the area of each class. In addition this
method offers the most readable visual product in the form of computer
printed maps or color photo maps.
Considerable expertise is needed to use this method. An analyst should
expect to have 60 to 80 h experience with its programs before turning out a
classification in 20 to 30 h. As the computer-assisted mapping system now
exists, the analyst should have experience in magnetic tape and mass storage
file manipulation.
Computer-assisted maps are costly to generate and moderately time
intensive. Because computer-assisted mapping involves so many steps where
there is room for variation in results, the reliability (repeatability) of
this method may not be great. Lastly, use of this method requires
considerable training and experience.
127
-------
The advantages of this method as it exists now are 1) it consistently
classifies across an entire scene so that while the identity of a particular
class may not be known, it is known that most of the class is being
classified; 2) it quantifies the area mapped for each class in units of the
analyst's choice; and 3) it generates color photo maps of the classified
scene which are the most immediately readable products generated by any of
the mapping methods.
128
-------
SECTION 5
RESULTS
To evaluate Che nine methods discussed in this study, it is necessary
to consider 1) the kind of information each provides, 2) the training needed
to use the method, 3) capital equipment and material costs, 4) time require-
ments of each method; 5) total costs incurred using each method and 6) the
sensitivity and reliability of each method. Only then do the advantages and
disadvantages of using each method become clear.
Tables 28-41 were assembled in the process of evaluating these methods.
They should be referred to throughout the discussion. Table 28 catalogs the
vegetation classes identified with each of the eight classification methods.
Table 29 lists the expertise needed to use each method (botanical-ecological
knowledge, computer experience, computer-assisted image interpretation
skill, visual airphoto interpretation skill, and drafting skill). Table 30
enumerates capital equipment costs for data collection. Table 31 lists data
collection materials costs for one data set. Table 32 catalogs the capital
equipment used with the nine methods. Table 33 lists capital equipment
costs for data analysis. Table 34 lists data processing material cost for
one data set. Table 35 summarizes the data collection and analysis time
requirements of each method while Table 36 lists total time to collect and
analyze one set of data using each of the nine methods. Table 37 gives a
breakdown of data collection, processing and labor costs for each method
when used in a 172 ha area. Table 38 lists data collection, data processing
and labor cost/method for one data set and four data sets in a 172 ha area.
Table 39 rates each method's efficiency on the basis of time and cost
required to use the method. Table 40 evaluates method sensitivity based on
the number of vegetation classes defined and data type. Table 41 evaluates
method reliability in terms of data collection method, data analyst
interaction and quantitative or qualitative results.
This chapter continues with a discussion of method efficiency, sensi-
tivity and reliability. This is followed by an evaluation of each method in
terms of time and cost to use the method and the method's sensitivity and
reliability. (All costs quoted in sections 31, 32, and 33 are costs using
the in-house capability at the University of Wisconsin-Madison.) If a user
went to the private sector to contract to have the methods discussed in this
study used, the costs would be two to three times the costs quoted here
(Evans 1979). The six questions raised in this study's introduction are
addressed and recommendations are made as to combinations of methods to use
in various situations. Lastly, suggestions are made on ways to improve the
present study.
129
-------
TABLE 28. VEGETATION CLASSES IDENTIFIED USING THE EIGHT
CLASSIFICATION AND MAPPING METHODS3
Vegetation classes
H
Carex striata
Degraded Carex stricta
Carex lacustris
Degraded Carex lacustris
Transition
Degraded transition
Emergents
Degraded emergent s
Spiraea alba
Shrubs
Open water
Open- emergent s
Weedy annuals
Trans it ion-emergent s
Sedges and grasses
Grasslike
Tall-coarse
Grasslike-tall
Disturbed vegetation
Undisturbed vegetation
Degraded sedges
Shrubs and trees
Typha lati folia
Scirpus fluviatilis
Floating mat
Lemna minor
Trees
Spiraea 1 hedges
Spiraea /shrubs
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X X
X X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
XXX
X X
X
X
X
X
X X
X X
X
X
X
X X
X X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Mapping methods are A = Diversity index; B = Subjective classification; C =
Association analysis; D » Vegetation structure analysis; E = Airphoto
monitoring; F = Disturbance mapping; G = Airphoto interpreted; H = Computer
assisted mapping.
130
-------
TABLE 29. EXPERTISE NEEDED TO USE EACH METHODa
Ground Sampling Data
Diversity index X
Subjective classification X
Association analysis X X
Structure analysis X
Airphoto Data
Airphoto monitoring X X
Disturbance mapping X XX
Airphoto-Ground Sampling Data
Airphoto grid analysis X X
Vegetation mapping X X XX
Computer Assisted Mapping X X XX
aMapping methods are A = Diversity index; B = Subjective classification; C =
Association analysis; D = Vegetation structure analysis; E = Airphoto
monitoring; F = Disturbance mapping; G = Airphoto interpreted; H = Computer
assisted mapping.
TABLE 30. CAPITAL EQUIPMENT COSTS FOR
PHOTOGRAPHIC DATA COLLECTION3
Data type
Airphoto data
Equipment used
Hasselblad 500 EL/M
chrome body
40 mm Distagon lens
A70 Film magazine
Daylight filter
Minus blue filter
Filter holder
Camera mount
Cost
1620 x 2
2550 x 2
615 x 2
249 x 1
249 x 1
120 x 2
300 x 1
, $
=> 3420
- 5100
= 1230
= 250
= 250
- 240
= 300
10,790
quoted in April, 1979
131
-------
TABLE 31. DATA COLLECTION MATERIALS COST FOR ONE DATA SET
Data type Materials used Total cost,
Ground sampling Data sheets 140 x $0.02 = 2.80
data Transportation $0.10/mlle x 80 miles x 2 days 16.00
0.25 m2 Circular quadrat $0.20 x 2 = .40
62 metal stakes <§ $2.55 x 62 = 158.10
177.30
Airphoto data Airplane & photographer $150/hr x 1 = 150
Film-two 15 ft rolls $2.50/ft x 30 ft = 75
Film processing $1.00/ft x 30 ft = 30
Film mailing—UPS 2 Ib package, roundtrip 16
$271
TABLE 32. CAPITAL EQUIPMENT USED WITH THE NINE METHODS
Method Capital equipment
1. Diversity index - None
2. Subjective classification Computing capability
3. Association analysis Computing capability
4. Vegetation structure analysis None
5. Airphoto monitoring Stereoscope and light table
Camera equipment
6. Disturbance mapping Color additive viewer
Camera equipment
7. Airphoto grid analysis Stereoscope and light table
Computing capability
8. Airphoto interpreted vegetation mapping Color additive viewer
Computing capability
Camera equipment
9. Computer-assisted mapping Scanning microdensitometer
Computing capabilities
Camera equipment
TABLE 33. CAPITAL EQUIPMENT COSTS FOR DATA ANALYSIS*
Items Cost, $
Stereoscope and light table 12,000
Color additive viewer 16,000
Scanning microdensitometer 100,000
Computing capability
aCosts quoted in April, 1979.
132
-------
TABLE 34. DATA PROCESSING MATERIAL COST FOR ONE DATA SET
Data type
Method
Materials
Cost, $
Ground sampling data
Airphoto data
Diversity Index
Subjective classification
Association analysis
Structure analysis
Airphoto interpretative monitoring
Disturbance mapping
r
<-j Airphoto and ground data Airphoto grid analysis
Vegetation mapping
Computer-Assisted mapping
None
Data printout
Total run cost
Data printout
1 airphoto
1 airphoto
Mylar @ $.10/in. x
15"
1 airphoto
Grid to scale
Sampling station location map
Mylar @ $.10 in. x 15"
Airphoto
Sampling station location map
12" computer tape
Tape of scanned image
Airphoto
Computer processing time charge
27
27
27
2.66
2.66
1.50
2.66
3.75
3.75
1.80
2.66
3.75
6.00
20a
2.66
40
llf airphoto scanned at UW-Madison. The charge is $35/h.
-------
TABLE 35. TIME (h) FOR DATA COLLECTION AND ANALYSIS FOR A 33.5 HA SITE
UJ
•P-
1.
2.
3.
4.
5.
6.
7.
8.
9.
Method
Diversity
index
Subjective
classification
Association
analysis
Structure
analysis
Airphoto
interpretive
monitoring
Disturbance
mapping
Airphoto
grid analysis
Vegetation
mapping
Computer-
assisted mapping
Hours to
collect
data
16
32
16
16
4
4
36
36
2
First
data analysis
set up time
4
12
14 to 16
6
1
2
14
18
4
Subsequent
data analysis
time
4
-
6
6
.4
0.25
1.2
7
10
2
First data
processing
time
6 to 8
6
6
2
1
1.5 to 3
9
11 to 13
4
Subsequent
data processing
time
1
4
2
1.5
0.5
1.2 to 3.5
7
7 to 11
16
Total
time per
method
21
42
24
21
4.7
6.5 to 8.7
50
54 to 57
22
-------
TABLE 36. TIME (h) FOR COLLECTION AND ANALYSIS
FOR ONE DATA SET USING EACH OF THE NINE METHODS
1.
2.
3.
4.
5.
6.
7.
8.
9.
Method
Diversity index
Subjective classification
Association analysis
Structure analysis
Airphoto interpretive monitoring
Disturbance mapping
Airphoto grid analysis
Airphoto interpreted vegetation
mapping
Computer-assisted mapping
Total time/method
21
42
24
21
4.8
6.5 to 8.8
50a
54 to 57b
22
Total time
rating
Medium
High
Medium
Medium
Low
Low
High
High
Medium
aTotal time for airphoto grid analysis if association analysis were used to
analyze the ground data would be 32 h.
Total time for airphoto interpreted vegetation mapping if association
analysis were used to analyze the ground data would be 36 to 39 h.
Low time requirement is 4.8 to 8.8 h
Medium time requirement is 21 to 24 h
High time requirement is 42 to 57 h
135
-------
TABLE 37. A BREAKDOWN OF DATA COLLECTION AND PROCESSING COSTS AND LABOR
COSTS FOR ONE SET AND FOUR SETS OF DATA FROM A 33.5 HECTARE STUDY SITE
Diversity Index
Data collection — materials
Data col lection — labor
1st set up time
1st data processing time
2nd set up time each
2nd data processing time each
Total Cost
Subjective Classification
Data collection — materials
Data collection — labor
Data processing materials
1st set up time
1st data processing time
2nd set up time each
2nd data processing time each
Total Cost
Association analysis
Data collection — materials
Data col lection — labor
Data processing materials
1st set up time
1st data processing time
2nd set up time each
2nd data processing time each
Time (h)
8
6
4
7
4
1
16
16
12
6
6
4
8
8
15
6
6
2
Cost/h ($)
13
3.5
13
13
13
13
13
3.5
13
13
13
13
13
3.5
13
13
13
13
Cost/data set ($)
178
104
28
310
52
91
143
453»
178
208
56
27
469
156
78
234
703»
178
104
28
31
341
195
78
273
Cost/4 data sets ($)
238
416
112
766
52
91
143
156
39
195
1,104»*
238
832
224
108
1,402
156
78
234
234
156
2,026»»
238
416
112
125
891
195
78
273
234
78
Total Cost
614*
1,475»"
136
-------
Table 37. (Continued)
Time (h)
Structure Analysis
Data col lectlon--materlals
Data collection — labor
Data processing materials
1st set up time
1st data processing time
2nd set up time each
2nd data processing time each
Tota 1 cost
Airphoto Monitoring
Data col lection- -mater la Is
Data col lection— labor
Data analysis materials
1st data set up time
1st data processing time
2nd data set up time each
2nd data processing time each
Total Cost
Airphoto Disturbance Maps
Data col lectlon--materlals
Data collection — labor
Data analysis materials
1st data set up time
1st data processing time
2nd data set up time each
2nd data processing time each
Total cost
Airphoto Grid Analysis
Data collection — materials
Data collection — labor
Data processing materials
1st data set up time
1st data processing time
8
8
6
2
4
1.5
4
1
1
0.25
0.50
2
2
3.5
1
2
2
2
3
Cost/h {$) Cost/data set ($)
13
3.5
13
13
13
13
13
13
13
13
13 ,
13
13
13
13
13
13
13 .
13
178
104
28
27
337
78
26
104
441*
271
52
3
13
13
352*
271
52
4
26
45
398»
271
26
8
305
26
39
65
Cost/4 data sets <$)
238
416
112
108
874
78
26
104
156
60
216
1.192**
1,084
208
11
13
13
to
19
1,357»»
1,084
208
17
26
45
39
78
1,497**
1,084
104
32
1,220
26
39
65
137
-------
Table 37. (Continued)
Time (h) Cost/h ($)
2nd data set up time each t 13
2nd data processing time each 2 13
Grid Analysis
Subjective Classification
Tota 1
Vegetation Mapping
Data collection — materials
Data collection — labor 2 13
Data processing materials
1st data set up time 6 13
1st data processing time 7 13
2nd data set up time each 3.75 13
2nd data processing time each 5 13
Vegetation Mapping
Subjective Classification
Total
Computer-assisted mapping
Data collection — materials
Data collection — labor . 2 13
Data processing materials
1st data set up time 4 13
1st data processing time 40 13
2nd data analysis set up cost 2 13
2nd data processing time 20 13
Cost/data set ($)
370
703
1,073»
271
26
10
307
78
91
169
476
703
1,179»
271
26
79
52
520
948»
Cost/4 data sets ($)
39
78
1,402
2,026
3,428»»
1.084
104
41
1,229
78
91
169
146
195
1,739
2,026
3,765»»
1,084
104
317
52
520
78
780
2,935»»
•Cost for first data set.
**Cost for first data set plus three subsequent sets.
138
-------
TABLE 38. DATA COLLECTION AND PROCESSING AND LABOR
COSTS FOR ONE DATA SET AND FOUR DATA SETS FROM A 33.5 HA SITE
1.
2.
3.
4.
5.
6.
7.
8.
9.
Diversity index
Subjective classification
Association analysis
Structure analysis
Airphoto monitoring
Disturbance mapping
Airphoto grid analysis
Vegetation mapping
Computer-assisted mapping
Total cost
1 year's data
217
441
287.15
250.50
335.41
366.16
785.21
861.91
662.30
Total cost from
years 1, 2, 3a
1,104
2,026
1,475
1,192
1,357
1,497
3,428b
3,764C
2,935
Total cost
Rating
Low
Medium
Low
Low
Low
Low
High
High
High
aThis figure is arrived at from the costs totals for 4 yr of data found in
Table 37.
This figure was arrived at using subjective classification for the ground
data analysis. If association analyses were used to do the ground data
analysis, airphoto grid analysis cost for 1 yr of data would be $625 and
for 4 yr of data would be $2,327.
cThis figure was arrived at using subjective classification for the ground
data analysis. If association analyses were used to do the ground data
analyses, airphoto interpreted vegetation mapping would cost $712 for 1 yr
of data and $2,663 for 4 yr of data.
If these methods are contracted out with commercial firms, they will cost 2
to 3 times the amounts listed in this table.
TABLE 39. COMBINED TIME-COST (EFFICIENCY) RATING FOR EACH METHOD
1.
2.
3.
4.
5.
6.
7.
8.
9.
Method
Diversity index
Subjective classification
Association analysis .
Structure analysis
Airphoto monitoring
Disturbance mapping
Airphoto grid analysis
Airphoto interpreted
vegetation mapping
Computer-assisted mapping
Time
rating
Medium
High
Medium
Medium
Low
Low
High
High
Medium
Cost
rating
Low
Medium
Low
Low
Low
Low
High
High
High
Combined
rating
Medium-low
High-medium
Medium-low
Medium- low
Low- low
Low-low
High-high
High-high
Medium-high
139
-------
TABLE 40. METHOD SENSITIVITY RATING BASED ON
VEGETATION CLASSES AND DATA TYPE
1.
2.
3.
4.
5.
6.
7.
8.
9.
Method
Diversity index
Subjective classification
Association analysis
Structure analysis
Airphoto monitoring
Disturbance mapping
Airphoto grid analysis
Airphoto interpreted
vegetation mapping
Computer-assisted mapping
Classes
defined
—
High
Medium
Low
Medium
Low
High
High
Medium
Data
type
Medium
High
Medium
Medium
Low
Low
Low
Low
Low
Sensitivity
rating
a
High-high
Medium-high
Medium- low
Medium-low
Low-low
High-low
High- low
Medium-low
Vegetation classes defined—High
Medium
Low
12-19 classes defined
8-11 classes defined
3-5 classes defined
Data type—High
Medium
Low
Species stem counts data
Presence-absence data
Airphoto data
aThe diversity index cannot be rated as it could not be assigned a
vegetation classes defined .rating.
140
-------
TABLE 41. METHOD RELIABILITY BASED ON DATA COLLECTION REPEATABILITY
DATA-ANALYST INTERACTION, AND QUANTITATIVE OR QUALITATIVE RESULTS
Subjective Quantitative or
Data collection data analyst qualitative
Method repeatability interactions results
1.
2.
3.
4.
5.
6.
7.
8.
9.
Diversity index
Subjective classification
Association analysis
Structure analysis
Airphoto monitoring
Disturbance mapping
Airphoto grid analysis
Airphoto interpreted
vegetation mapping
Computer-assisted mapping
Medium
Low
Medium
Low
High
High
High
High
High
High
Medium
High
Medium
Low
Low
Low
Low
Low
Medium
Medium
High
Medium
Low
Low
Medium
Medium
High
Data collection repeatability—High airphoto data
Medium presence-absence data
Low species stem counts data
Data analyst interaction—High—Minimum subjective data-analyst interaction
Medium—Medium subjective data-analyst interaction
Low—Maximum subjective data-analyst interaction
Quantitative or qualitative results—High—Computer quantitative results
Medium—Analyst quantitative results
Low—Analyst qualitative results
141
-------
METHOD EFFICIENCY
For purposes of this study, an efficient method is defined as one
requiring little time and little cost to use. Table 36 lists the total time
in hours to use each method. Airphoto interpreted monitoring and distur-
bance mapping had the lowest time requirement. The diversity index,
association analysis, structure analysis, and computer-assisted mapping were
all rated as medium time intensive methods, taking between 21 and 24 h to
use. Subjective classification, airphoto grid analysis and airphoto
interpreted vegetation mapping were the most time intensive methods, taking
between 42 and 57 h to use. Indeed, it is interesting to note that there is
a tenfold difference between the least and most time intensive methods.
Table 37 breaks each method's total cost into: 1) data collection
equipment cost, 2) data collection materials cost, 3) data collection labor
cost, 4) first time data analysis set up cost, 5) first time data processing
cost and subsequent data analysis set up cost, and 6) subsequent data
processing cost. Total Cost 1 for one data set reflects the greater labor
cost incurred the first time a method is used. Total Cost 2 for one data
set reflects the reduced labor cost required once the analyst is familiar
with the method. Cost per four data sets is made up of the cost of one data
set at Total Cost 1 plus three data sets at Total Cost 2.
Table 38 uses figures from Table 37 to present a data collection,
processing and labor cost rating for each of the nine methods. The diver-
sity index, association analysis, structure analysis, airphoto monitoring
and disturbance mapping are low cost methods ranging from $217/data set to
$366/data set. Subjective classification is rated as a medium cost method
(costing $441.00) while airphoto grid analysis, vegetation mapping and
computer-assisted mapping are rated as high cost methods (ranging from $662
to $862). The cost difference among these methods is less dramatic than the
time difference. There is only a four-fold difference between the most and
least costly methods.
Table 39, a combined time-cost (efficiency) rating table, was assembled
from information presented in Table 36 and 38. The method with the greatest
efficiency is a low cost method requiring little time to use. The method
with the least efficiency would have a high time requirement and a high cost
rating.
The most efficient of the nine methods are airphoto monitoring and
disturbance mapping. The diversity index, association analysis and
structure analysis have a medium to high efficiency. Subjective classifica-
tion and computer-assisted mapping have a combined medium-high rating
indicating their medium to low efficiency. Airphoto grid analysis and
airphoto interpreted vegetation mapping were rated as high time and high
cost methods, indicating they are the least- efficient of the nine methods.
142
-------
METHOD SENSITIVITY AND RELIABILITY
A method sensitivity rating table and a method reliability rating table
(Tables 40 and 41) were assembled to provide ratings of all methods. Each
method was evaluated for sensitivity and reliability. The analyst would
prefer to evaluate each method in terms of accuracy, however, since no
detailed vegetation map of the study site was made prior to its disturbance,
there is no "truth" with which to compare the results of each method.
By sensitivity, the analyst means nuances of vegetation change which
each method is capable of showing. Reliability refers to the repeatability
of method results. A method is more reliable if it is less subject to error
and uses less analyst judgement. Method sensitivity and reliability are
difficult to evaluate since the variables which determine them are dependent
on one another.
The sensitivity evaluation is based on the following assumptions:
1) Number of vegetation classes defined: The more classes a method
defines, the more sensitive that method is. When a method defines
many classes, less obvious changes in the vegetation can be
detected earlier than would be possible using a method which
defines very few classes. Vegetation classes/method are defined as
all classes identified using all data sets available to the
method. Thus, airphoto grid analysis defined 19 classes (Table 28)
although 16 vegetation classes are the maximum number defined on
one photo at a time.
2) Data type affects method sensitivity. Species stem counts data are
considered more sensitive than either presence-absence data or
airphoto data as they are able to record change in species
densities. Presence-absence data are considered the next most
sensitive data type since they function at the species level,
recording change in species presence. Airphoto data are considered
the least sensitive of the data types since they record change at
the community level.
To assemble a sensitivity index, number of classes and data type were
recorded as follows: Methods defining 12 to 19 vegetation classes were
assigned a high sensitivity rating; methods defining 8 to 11 classes were
assigned a medium sensitivity rating; method defining 3 to 5 classes were
assigned a low sensitivity rating. Methods using species stem counts data
were assigned a high sensitivity rating; methods using presence-absence data
were assigned a medium sensitivity rating; and methods using airphoto data
were assigned a low sensitivity rating.
Table 40 lists the nine methods, together with each method's rating in
terms of the number of classes defined and the data type. These two ratings
are combined into a sensitivity rating. Subjective classification which
rated high for the number of classes defined and the data type, is the most
sensitive method. Association analysis, airphoto grid analysis and airphoto
143
-------
interpreted vegetation mapping all rated next with either high-low or
medium-high ratings. Structure analysis, airphoto monitoring, and computer-
assisted mapping were all assigned a medium-low sensitivity rating.
Disturbance mapping is the least sensitive method with a low-low rating.
A reliability rating was assembled based on the following assumptions:
1) Airphoto data collection repeatability is greater than either
species presence-absence data repeatability or species stem counts
data. This is because airphoto specifications (altitude, allowable
percent cloud cover, etc.) insure consistent high quality data.
Species presence-absence data are less exacting to collect than
species stem counts data and are therefore more repeatable.
2) Subjectivity of data-analyst interaction. The more opportunity for
subjective data-analyst interaction, the less reliable a method is
believed to be. All airphoto interpreted methods are subjective.
Computer-assisted mapping requires many subjective analyst
decisions which affect final results. Subjective classification
and structure analysis both quantitatively define each vegetation
class, removing some subjectivity. The diversity index is a purely
numerical method while association analysis is objective and
entirely repeatable once the analyst selects a species analysis
method and options.
3) Quantitative/qualitative results. Quantitative results are assumed
to be more repeatable than qualitative results. Quantitative
computerized results are assumed to be more repeatable than
quantitative analyst results. Quantitative-analyst results are
assumed to be more repeatable than qualitative-analyst results.
In this study, airphoto data were found to have a high repeatability;
presence-absence data, a medium repeatability; and species stem counts data,
a low repeatability. Methods with minimal subjective data-analyst
interaction were given a high repeatability rating; those with moderate
subjective data-analyst interaction were given a medium repeatability rating
and those methods with extensive subjective data-analyst interaction were
given a low repeatability rating.
Quantitative-computer methods were assigned a high repeatability
rating; quantitative-analyst methods were assigned a medium repeatability
rating. Qualitative-analyst methods were assigned a low repeatability
rating.
Table 41 lists each method and its data collection repeatability
rating, subjectivity of data-analyst interaction rating, and quantitative or
qualitative results rating. Association analysis with a medium-high-high
shows the highest reliability rating followed by the diversity index and
computer-assisted mapping. Airphoto grid analysis and airphoto interpreted
vegetation mapping were rated next with a high-low-medium reliability
rating. Subjective classification, structure analysis, airphoto monitoring
144
-------
and disturbance mapping were rated the least repeatable methods with low-
medium-medium or high-low-low ratings.
Looking at the reliability and sensitivity of each method, association
analysis had the highest repeatability rating (medium-high-high) and a
medium-medium sensitivity rating. Subjective classification had a high-high
sensitivity rating and low-medium-medium repeatability rating. Airphoto
grid analysis and airphoto interpreted vegetation mapping had a high-low
sensitivity rating and a high-low-medium repeatability rating. Computer-
assisted mapping had a medium-low sensitivity rating and a high-low-high
repeatability rating. Structure analysis and airphoto monitoring had a
medium-low sensitivity rating. Structure analysis had a low-medium-medium
repeatability rating while airphoto monitoring had a high-low-low rating.
Disturbance mapping had the lowest repeatability and sensitivity ratings
with a high-low-low reliability rating and a low-low sensitivity rating. As
the diversity index did not define classes, it could not be assigned a
sensitivity rating. However, it was assigned a medium-high-medium
reliability rating.
THE GROUND SAMPLING DATA METHODS
Training as a botanist-ecologist is required to 1) collect the ground
sampling data that insures correct identification of vegetation species; 2)
define the vegetation classes used in the subjective and structure analysis
classifications; and 3) label the clusters created using association
analysis. Computer experience is necessary when using association analysis,
Data collection materials costs totalled $177 (Table 31) for 62 metal
stakes, two circular quadrats, data sheets and transportation to and from
the site.
Diversity Index
The diversity index, based on presence-absence data, provides informa-
tion about the changes in numbers of species with time. This method only
indicates if the total number of different species varies with time; it does
not identify which species appear or disappear with time.
Twenty-one hours are required to carry out this method, placing it in a
medium time intensive class (Table 36). The total cost of processing one
set of data is $453; if four sets of data are processed, the total cost is
$1,104.
The advantages of this method are: 1) it offers a quantitative index
of species change; 2) it is easy to use in any size area; and 3) it is a
reliable, medium time intensive and low cost method. Its greatest disad-
vantage is that it offers only species presence information, not the
replacement of one species by another. Also, this method automatically
145
-------
interprets Increased or decreased diversity as either negative or positive
change.
Subjective Classification
The subjective classifcation, based on stem counts data, provides
information about changes in community location and trends in change which
the various communities are undergoing, year by year. When sampling station
classifications are mapped, they provide a rough visual presentation of
change in community location and area.
The subjective classification is one of the three most time intensive
methods, requiring 42 h. Total cost to collect and analyze one set of data
is $703. Total cost to collect and analyze four sets of data is $2,026,
making this the most expensive of the ground sampling data methods (Table
37).
Subjective classification, the most sensitive of the ground data
sampling methods, was given the highest possible sensitivity rating but only
a low-medium-medium reliability rating.
The disadvantage of this method is its time intensiveness which results
from the time necessary to collect stem counts data. Two hundred data
points (which take two people 4 to 5 days to collect and one person 18 h to
analyze) might well be the upper limit in using this method.
Association Analysis
Association analysis, based on presence-absence data, provides informa-
tion about vegetation changes in community type, community location, and
area and trends in change which communities are undergoing year by year.
This method, requiring 24 h to collect and analyze one data set, was
rated medium in time intensiveness. Total cost/data set is $614 and $1,476
for four data sets (Table 37).
The sensitivity of association analysis was rated medium-medium while
its reliability was rated medium-high. It was rated medium in terms of time
intensiveness and low in cost which indicates that it is a moderately to
highly efficient method.
An advantage of association analysis is the semi-objective and very
reliable classification that it offers. Once the data are collected and the
analyst has selected the species and the program options to be used, the
classification is done in a standard manner by the computer. . Use of
presence-absence data helps keep data collection time down. While data
analysis costs are related directly to data matrix size, this method can
handle large data sets in a short period of time.
146
-------
Structure Analysis
Structure analysis shows changes in vegetation structure based on the
percent of total cover area of individual vegetation classes. This method
provides information about change of community structure in terms of
location and area. It shows changes in community area and in community
location if classified data from all the sampling stations are mapped, point
by point, with time.
This method falls into the medium time intensiveness category requiring
21 h to collect and analyze one data set. Total cost for one data set is
$441. Total cost for four data sets is $1,192.
Structure analysis is less sensitive than the other classification
methods since it defines only five vegetation classes as the basis of
recording structure changes. Both species and community change could occur
within an area without a change occurring in structure and this method would
not record it.
Structure analysis was rated as low-medium in terms of sensitivity and
as low-medium-medium in terms of reliability. It was rated medium in terms
of time intensiveness but it was rated low in cost, meaning it has medium-
high efficiency.
The structure analysis method is only effective as a monitoring tool in
those situations where the change under study involves actual changes in
vegetation structure (e.g., a situation where shrubs/trees invade a grass-
land). Neither the amount of time this method takes nor its total cost
undercut the costs of the other two classification methods enough to make up
for the lesser quality of information it yields. Because this method can
only detect major change, it will tell what has happened in an area, but not
how change is happening—an important consideration in most monitoring
situations.
AIRPHOTO DATA ONLY METHODS
A botanist-ecologist with airphoto interpretation skills is best suited
to do airphoto interpretative monitoring and disturbance mapping (Table 29).
Data materials cost, including airplane rental, photographer fees, film and
film processing, come to $271/roll each of color and color infrared film.
Capital equipment costs for a two camera, 70 mm format system come to
approximately $10,790. Therefore the costs of data materials and data
collection equipment are far higher using airphoto data than using ground
sampling data (Tables 30 and 31).
147
-------
Airphoto Monitoring
Airphoto monitoring offers a visual record of change with time in the
form of a series of photos* This method does not require ground verifica-
tion. Results using this method may be presented as a written description.
Capital equipments costs (Tables 30 and 33) for this method are
approximately $23,000. Material cost is $274/airphoto analyzed. Airphoto
monitoring is a rapid method to carry out, requiring only 4.75 h from start
to finish. Total cost for analyzing one airphoto is $352. The cost of
analyzing four airphotos is $1,357, making this an inexpensive method (Table
37).
Airphoto monitoring potentially offers a detailed data base with which
to demonstrate change with time, however, since it does not define many
vegetation classes and uses only airphoto data, it was assigned a medium-low
sensitivity rating (Table 40). Airphoto monitoring was assigned a high-low-
low reliability rating because vegetation class data are not quantified, and
the method allows for many subjective analyst-data interactions. This
method takes the least time (Table 36) and is rated as a low cost method.
It has the highest possible combined time-cost (efficiency) rating.
Airphoto monitoring offers a quickly accessible visual interpretation
of an airphoto from which a written description can be made. This method
can be used to define obvious vegetation and land use classes in areas where
on-the-ground access is not possible (Hubbard and Grimes 1972). No
provision is made with this method for collecting ground verification
data. This keeps down costs but lessens the method's information content.
This method's greatest disadvantage is that information gathered is neither
quantifiable nor mapped; it can only be as good as the analyst's skill in
using it.
In summary, the primary advantages of airphoto monitoring are 1) if
changes that can be detected visually are taking place, they can be detected
using this method; 2) this method can be used to monitor areas that are
inaccessible on the ground; 3) this is an expensive method in terms of time
and costs (discounting capital equipment); and 4) it can be used to monitor
large sites.
Disturbance Mapping
Disturbance mapping offers a series of readable maps delineating
undisturbed and disturbed vegetation and open water. Disturbance mapping,
which defines only three classes, was rated the most insensitive method. It
has a high-low-low reliability rating.
Capital equipment costs (camera equipment and color additive viewer
costs) for disturbance mapping data collection and analysis is approximately
$27,000 (Tables 30 and 33). The cost for materials is approximately $275.
148
-------
A disturbance map takes between 6.5 and 8.75 h to generate (from start to
finish), depending on photo scale. Total cost for one map is $398
(discounting capital equipment); for a series of four maps it is $1,497.
Disturbance mapping has the highest possible efficiency rating.
The advantage of disturbance mapping is its provision of a highly
readable set of maps which are very sensitive in delineating generalized
disturbance. The disadvantage of this method lies in the small number of
vegetation classes it defnes. As a result of this, disturbance mapping
offers no information on how specific communities are changing.
AIRPHOTO GRID ANALYSIS, AIRPHOTO INTERPRETED VEGETATION MAPPING AND COMPUTER
ASSISTED MAPPING
A botanist-ecologist trained in airphoto interpretation is best suited
to use airphoto grid analysis and airphoto interpreted vegetation mapping.
Computer-assisted mapping requires an analyst with the skills listed above
plus training in computer-assisted mapping and experience in manipulating
data between computer mass storage and data tapes.
Data materials costs for these methods are the same as for the two
methods involving airphoto data exclusively. Film and film processing costs
come to $271 for a roll each of color and color infrared film. Two 70 mm
format cameras, including bodies, lenses, filters and film cans, cost
$10,800. Data acquisition time for each of these three methods is approxi-
mately 4 h. This includes loading film, taking cameras to and from the
airport and allowing 1 h flying time.
Airphoto Grid Analysis
Capital equipment cost (for camera equipment and a stereoscope and
light table) for airphoto grid analysis comes to approximately $23,000.
Total materials cost to analyze one airphoto is $279. Data collection and
analysis time required per airphoto is 50 h. Total analysis cost/airphoto
is $1,070; for four airphotos it is $3,420.
Airphoto grid analysis was assigned a high-low sensitivity rating along
with vegetation mapping. Its reliability is rated as high-low-medium.
Airphoto grid analysis is a time and cost intensive method. Due to its high
time and dollar costs, its overall efficiency rating is low-low.
Airphoto grid analysis is a sensitive method in terms of the number of
classes which can be identified using it. It is a quantifiable method
yielding community percent cover information. Performed on records obtained
with time, this method documents changes in each community. This method can
be reliable if the airphotos are consistent in scale and quality and if
interpretation is consistent. An airphoto interpretation key will help
149
-------
provide consistency among analysts. Frequent reference to ground sampling
data increases the sensitivity and reliability of the method.
Using airphoto grid analysis, percent cover change in area can be
detected for species which form large clones (such as cattail). Percent
cover changes in communities also can be detected. The method is well
suited for use in any monitoring situation where the work is done by one
analyst since this ensures consistency in interpretation.
This method is well suited to use with many airphotos and/or large
areas. It is the third most expensive in terms of the total cost for four
runs and it is the second most intensive in terms of time. It is one of
three quantifiable methods and ranks second in the number of classes it
defines.
Airphoto Interpreted Vegetation Mapping
Capital equipment costs for vegetation mapping (which requires the use
of camera equipment and a color additive viewer) are approximately $27,000.
Materials costs are $281/airphoto mapped. Total production time of one map
will take 54 to 58 h depending on the scale of the airphoto (larger scales
providing more detail take longer). (Forty-two of the hours required for
vegetation mapping stem from subjectively classifying the ground verifica-
tion data. This time can be reduced by using association analysis for
ground verification data.) Total cost to produce one map is $1,180; total
cost to produce four maps is $3,760.
Vegetation mapping was assigned a high-low sensitivity rating while its
reliability was assessed at high-low-medium. This method's highest cost and
time requirements give it the lowest possible efficiency rating.
Vegetation mapping is the most sensitive method in terms of numbers of
vegetation classes it can show with time. Its sensitivity and reliability
are made possible by frequent reference to ground sampling data.
This method shows changes in area and location of species such as Typha
latifolia and Sdr>pU8 fluviatilie which form large clones. Changes in
community location and area are recordable with time. Since the maps were
not drawn on a standard base map, they could not be overlaid and percentage
changes in community area could not be figured. If these maps had been
drawn on an orthophoto base map blown up to an appropriate scale, this would
have been possible. (A grid was laid over each map and numbers of
cells/vegetation class counted to determine relative percent of each class.)
Labelling each class with a letter made it difficult to keep track of the
changes revealed in a series of maps. A multiple pattern and gray tone
photographic process should be used to make detailed maps which are legible.
The great advantage offered by airphoto interpreted vegetation mapping
is the amount of visual detail it provides; detail showing change in area
and location of communities. It is the most time intensive and costly
150
-------
method. Its cost Increases with larger photo scale and area. Its strongest
points are the detail it offers and its potential overlay capability.
Computer-Assisted Mapping
Computer-assisted mapping offers consistent classification of vegeta-
tion classes and quantification of the area enclosed by each class. In
addition the method offers the most readable visual product in the form of
computer printed maps or color photo maps.
In-house capital equipment costs for computer-assisted mapping are
$110,000 (for camera equipment and scanning microdensitometer, Tables 30 and
33). This method rates medium in terms of intenslveness; 22 h are required
to generate one map. And in terms of cost, this method rates third out of
the nine methods. One map costs $948 to generate while a series of four
costs $2,935.
Considerable expertise is needed to use this method. An analyst should
expect to have 60 to 80 h experience with its programs before turning out a
classification in 20 to 30 h. As the computer-assisted mapping system now
exists, experience in key punching and magnetic tape and mass storage file
manipulation is necessary.
Computer-assisted mapping was assigned a sensitivity rating of medium-
low and a reliability rating of high-high-low. Its time-cost rating is
medium—high making it a moderately efficient method.
The sensitivity of this method is hampered by the confounding of
vegetation spectral response patterns. Such confounding prevents the
identification of a large number of vegetation classes. The inability to
define more classes in this particular study was due to the analyst's
inability to select training sets from areas of a known vegetation type.
Since no site specific targets were available, correct labelling of classes
could not be assured. These two problems could be overcome by putting out
permanent targets which would be visible on each airphoto.
A prime disadvantage of this method is its high cost if the method is
to be used in-house. Capital equipment costs exceed $100,000. Computer-
assisted maps are costly to generate and required medium time intensive-
ness. Since computer-assisted mapping involves so many steps where there is
room for variation in results, the reliability (repeatability) of this
method may not be great. Lastly the method requires considerable training
and experience to use it.
The advantages of this method are: 1) it classifies consistently
across an entire scene so that while a particular class is not known, it is
known that all or most of it is being classified; 2) this method quantifies
the area mapped for each class in units of the analyst's choice; and 3) this
method generates color photo maps of the scene classified which are the most
immediately readable products generated by any of the mapping methods. The
15L
-------
method's value is enhanced by the fact that similar computer-assisted
mapping methods are being developed and promoted for use in the U.S.
Department of Interior.
152
-------
SECTION 6
DISCUSSION
The questions raised in this study were:
1) Using these methods, can species changes be detected with time?
2) Which methods detect species changes most efficiently?
3) Using these methods, can community area and location changes be
detected with time?
4) Which methods detect community change most efficiently?
5) Can trends in vegetation changes be documented by these methods?
6) Which methods do so most efficiently?
For purposes of this study, an efficient method is defined as a low
cost method which takes little time to use. Table 39 rates the methods by
time and cost to use them, providing an efficiency rating.
Based on the results of this study, the diversity index recorded
species change most efficiently, however, it records change in numbers of
species only. Subjective classification and association analysis recorded
species change indirectly, through changes in the numbers of stations
classified as a particular community. (Community classifications changed as
species showed declines or increases in numbers or disappeared altogether.)
Subjective classification was based on species counts data and as such gave
the more sensitive analysis but association analysis is a less expensive
method to use.
Community changes in area and/or location were demonstrated by all
methods except the diversity index. All three classification methods showed
change based on community point location (for 62 sampling stations).
Airphoto methods showed changes in community area and location more clearly
than the ground sampling method. Grid analysis, assigned a high-low
sensitivity rating, recorded percent area change of communities. Although
this method records the grid cell location of each community, this informa-
tion was not mapped in any way. Airphoto monitoring displays changes on
airphotos of community location and area but does not visually record
manually or delineate them in any way. Both airphoto interpreted vegetation
mapping and computer-assisted mapping visually define communities and
quantify percent changes in area. Changes in community location can be
153
-------
inferred by looking at a series of maps generated with time. An overlay
capability could be built into these methods that would allow location
changes to be quantified. Photo interpreted vegetation mapping and
computer-assisted mapping are time consuming methods. If it is sufficient
to show only disturbed and undisturbed vegetation, disturbance mapping is
the most efficient method to show change in community location and area.
The methods which provided the most information on trends in community
change were subjective classification and association analysis. Subjective
classification is ranked medium-low for efficiency. Association analysis, a
more efficient classification method, is recommended as the second best
community trend demonstrating method. These two classification methods are
mentioned first because in classifying community data at a number of points
with time, they record the steps of deterioration which the various
communities move through going from an undegraded to a degraded state.
Structure analysis demonstrated trends in the physical structure of the
vegetation from grasslike vegetation to tall-coarse vegetation and open
water areas.
Most of the methods based on airphoto data show the community trends
taking place at the study site. Airphoto monitoring provides photos which
have the information but an analyst must interpret them to show change.
Airphoto grid analysis demonstrated the trend from grasses and sedges to
emergents, degraded vegetation and open water using percent community cover
information. Disturbance mapping with the highest possible efficiency
rating shows the increase in disturbed area and open water over 3 yr time.
Airphoto interpreted vegetation mapping and computer-assisted mapping
demonstrate community trends for classes ranging from sedges and grasses to
disturbed vegetation and open water using community percent cover estimates.
Of the two methods, airphoto Interpreted vegetation mapping is less
efficient.
USING METHODS TOGETHER
Ground data classification methods can be used to advantage with air-
photo data mapping methods because the classification will demonstrate the
stages in which change is occurring while the mapping methods will show
where change is occurring.
Subjective classification, a very time intensive method, can be used
with small data sets to demonstrate vegetation trends. With larger data
sets, it is better to use association analysis. Airphoto grid analysis,
airphoto interpreted vegetation mapping or computer-assisted mapping provide
percent cover change information. Airphoto interpreted vegetation mapping,
like subjective classification, is best suited to mapping smaller areas
(200 ha or less) whereas airphoto grid analysis and particularly computer-
assisted mapping can be used along with association analysis to monitor
large areas.
154
-------
Two compatible methods are airphoto grid analysis, which documents
percent cover change, and disturbance mapping, which visually shows the
areal extent of the change. The two methods are well suited for use in
large areas.
Airphoto monitoring and disturbance mapping can be used in areas which
are inaccessible on the ground because no ground verification data are
required.
Disturbance mapping is a quick, inexpensive, reliable method combina-
tion to show change in a disturbed area. Association analysis provides a
good supplemental analysis substantiating the disturbance mapping by showing
with changes in communities year by year.
If computer-assisted mapping is streamlined by using an interactive
graphics terminal for training set selection, classification verification,
etc., and if ground verification areas are targeted so known training sets
can be identified on imagery, this will be the ideal method to use on large
data sets in combination with association analysis. Association analysis
can be used to classify ground data and this Information can be tied into
surveyed target points to identify the mapped classes. Based on this study,
the visual displays and quantitative presentation of vegetation class area
offered by computer-assisted mapping make it the most valuable mapping
method.
155
-------
REFERENCES
An annotated bibliography on remote sensing for vegetation mapping is
presented in Appendix H.
Anderson, R. R., and F. J. Wobber. 1973. Wetlands mapping in New Jersey.
Photo. Eng. 39(4):353-358.
Andrews, C. 1977. Sensing and predicting thermal pollution of ground
water. In: Proceedings of the Fourth Joing Conference on Sensing of
Environmental Pollutants. American Chemical Society. Washington,
D.C. p. 105-115.
Bedford, B. L. 1977. Changes in wetland vegetation associated with leakage
from the cooling lake of a coal-fired power plant. M.S. Thesis.
University of Wisconsin-Madison, Madison, Wis. 37 p.
Bisset, J. 1978. Quantification, decision making and environmental impact
assessment in the United Kingdom. J. Environ. Management 7:43-58.
Brown, W. Wetland mapping in New Jersey and New York. Photo. Eng.
44(3):303-314.
Buchanan, W. J. 1977. Applications of digitized film analysis and
perceptions of possible users. M.S. Thesis. University of Wisconsin-
Madison, Madison, Wis. 124 p.
Bunce, R. G. H., and M. W. Shaw. 1973. A standardized procedure for
ecological survey. J. Environ. Management 1:239-258.
Civco, D. L., W. L. Kennard, and M. L. Lefor. 1978. A technique for
evaluating inland wetland photointerpretation: the cell analytical
method (CAM). Photo. Eng. 44(8):1045-1052.
Cowardin, L. M., and V. I. Meyers. 1971. Remote sensing for identification
and classification of wetland vegetation. J. Wild. Management.
38(2):308-314.
Dames and Moore, Inc. 1972. Environmental consultants. Chicago, II.
Dirschl, H. J., and D. L. Dabbs. 1972. The role of remote sensing in
wildland ecology and environmental impact studies. In: Proceedings of
the First Canadian Symposium. Ottawa, Ontario, p. 337-344.
156
-------
Eberhardt, L. L. 1976. Quantitative ecology and impact assessment. J.
Environ. Management 4:27-70.
Enslin, W. R., and M. C. Sullivan. 1974. The use of color infrared
photography for wetland assessment. In: Remote Sensing of Earth
Resources Conference. The University of Tennessee Space Institute.
Tullahoma, Tennessee. Vol. III. p. 697-720.
Evans, Kent. 1979. Personal communication. Enviro Control. Chicago.
Fischer, T. W., and G. S. Davies. 1973. An approach to assessing
environmental impacts. J. Environ. Management 1:207-227.
Fornes, A. 0., and J. R. Reimold. 1973. The estuarine environment:
location of mean high water—its engineering, economic and ecological
potential. Part II. In: Proceedings, Am. Soc. of Photogrammetry Fall
Convention. Lake Buena Vista, Florida, p. 938-978.
Frazier, B. E., and G. B. Lee. 1975. Effectiveness of a computer land use
planning system utilizing data. In: Proceedings of the Amer. Soc. of
Photogrammetry Fall Convention. Phoenix, Arizona, p. 754-777.
Gallagher, J. L., D. E. Thompson, and R. J. Reimold. 1972. Remote sensing
and salt marsh productivity. In: Proceedings, Am. Soc. Photogrammetry
Symposium on Coastal Mapping. Washington, D.C. p. 18-31.
Gammon, P. T., and V. Carter. 1979. Vegetation mapping with seasonal color
infrared photographs. Photo. Eng. 45(1):87-97.
Goodenough, D., and S. Shlien. 1974. Results of cover-type classification
by maximum likelihood and parallelepiped methods. In: Proceedings of
the Second Canadian Symposium on Remote Sensing. Guelph, Ontario.
p. 135-144.
Hough, A. F. 1965. A twenty year record of understory vegetational change
in a virgin Pennsylvania forest. Ecology 46(5,): 370-373.
Hubbard, J. C. E., and B. H. Grimes. 1972. Coastal vegetation surveys.
In: Environmental Remote Sensing: Applications and Achievements, Eric
C. Barrett and Leonard F. Curtis, eds. Crane and Russak,
New York. p. 129-141.
Jaeger, J. M. 1979. Predicting effects of an electrical generating station
on wetland passerine birds. Environmental Research Laboratory-
Duluth. U. S. Environmental Protection Agency. 57 p.
Johnson, B. G. 1974. Developing data for environmental impact studies. J.
of Tech. Ass. of the Pulp and Paper Industry. 57(9): 81-84.
157
-------
Kalman, L. S., and F. L. Scarpace. 1979. Determination of lens falloff
through digitized analysis of photographic imagery. In: Proceedings
of the An. Soc. of Photogrammetry 45th Annual Meeting, Washington,
D.C. p. 116-135.
Lillesand, T. M., and R. W. Kiefer. 1979. Remote sensing and image
interpretation. John Wiley & Sons. New York. 612 p.
Leopold, L. B., F. E. Clarke, B. B. Hanshaw, and J. R. Balsley. 1971. A
procedure for evaluating environmental impact. Geological Survey
Circular 645. Washington, D.C. 13 p.
McHarg, I. 1969. Design with nature. National History Press. Garden
City, New York. 197 p.
Meyer, Merle. 1977. Personal communication. University of Minnesota,
Minneapolis, Minnesota.
Muller-Dombois, D., and H. Ellenberg. 1975 Aims and methods of vegetation
ecology. John Wiley & Sons. New York. 547 p.
Niemann, B. J., Jr. 1979. Personal communication. University of
Wisconsin-Madison, Madison, Wisconsin.
Niemann, B. J. , Jr., X. A. Bonilla, S. R. Bruno, and R. A. Rose. 1975.
Rural landscape assessment: a comparative evaluation of high platform
remote sensors. Dep. of Interior, BOR, and College of Agriculture and
Life Sciences, University of Wisconsin-Madison, Madison, Wisconsin.
243 p.
Rainwater, C. 1971. Light and color. Western Publishing Company, Ind.
New York. 160 p.
Scarpace, F. L. 1979. Personal communication. University of Wisconsin-
Madison, Madison, Wisconsin.
Scarpace, F. L. 1978. Densitometry on multi-emulsion imagery. Photo. Eng.
44(10): 1279-1292.
Scarpace, F. L., L. T. Fisher, and B. Quirk. 1978. Programs for digital
processing. Remote Sensing Group, Institute for Environmental
Studies. University of Wisconsin-Madison, Madison, Wisconsin.
Scher, J. S., and P. T. Tueller. 1973. Color aerial photos for marshlands.
Photo. Eng. 39:480-499.
Shima, J. L. 1973. Wetland vegetation mapping using aerial, color infrared
photography. M.S. Thesis. The American University. Washington,
D.C. 34 p.
158
-------
Shima, J. L., R. R. Anderson, and V. P. Carter. 1976. The use of aerial
color photography in mapping the vegetation of a freshwater marsh.
Chesapeake Science 17(2):74-85.
Siegal, S. 1956. Nonparametric statistics for the behavioral sciences.
McGraw-Hill, Inc. New York. 312 p.
Smith, D. W., R. Suffling, D. Stevens, and T. S. Dai. 1975. Plant
community age as a measurement of sensitivity of ecosystems to
disturbances. J. Environ. Management 6:27-42.
Swain, P. H., and S. M. Davis, eds. 1978. Remote sensing: the
quantitative approach. McGraw-Hill. New York. 376 p.
Thompson, D. E. 1972. Airborne remote sensing of Georgia tidal marshes.
In: Operational remote sensing: an interactive seminar to evaluate
current capabilities. Am. Soc. of Photogrammetry. Houston, Texas.
p. 126-139.
Wacker, A. G., and D. A. Landgrebe. 1972. Minimum distance classification
in remote sensing. In: Proceedings of the First Canadian Symposium on
Remote Sensing. Ottawa, Ontario, p. 577-599.
Warner, M. L., and D. W. Bromley. 1974. Environmental impact analysis: a
review of three methodologies. University of Wisconsin-Madison.
Institute for Environmental Studies. Departments of Forestry and
Agriculture Economics and Water Resources Center, Technical Report.
65 p.
Whitman, R. I., and K. L. Marcellus. 1973. Textural signatures for wetland
vegetation. In: Proceedings of the Am. Soc. of Photogrammetry Fall
Convention. Lake Buena Vista, Florida, p. 979-992.
Williams, W. T., and J. M. Lambert. 1959. Multivariate methods in plant
ecology. I. Association analysis in plant communities. J. Ecol.
47:83-101.
Williams, W. T., and J. M. Lambert. 1960. Multivariate methods in plant
ecology. II. The use of an electronic computer for association
analysis. J. Ecol. 48:689-710.
Wishert, D. 1970. Clustan IA User Manual. Computer Laboratory, University
of St. Andrews, St. Andrews, Scotland, p. 118.
159
-------
APPENDIX A
SPECIES CODE LIST
TABLE A-l. SPECIES CODE LIST
Species Code
Species Code
No.
Speclea
Species Code
No.
Species
No.
Species
1. Acorns calamus
2. Calamagroatie ganadeneia
3. Carex aqua.ti.li8
4. Carex emoryi
5. Carex haydenii
6. Carex laauetria
7. Carex laeiocarpa
8. Carex roetrata
9. Carex etriata
10. Dryopterie thelypterie
31. Aealepia inaarnata
32. Aster sp.
33. Aeter luaidulue
34. Aster puniceps
35. Aeter simplex
36. Bidene aermua
37. Bidene aoronata
38. Biehmeria aylindriaa
39. Campanual aparinoideo
40. Cardamine bulboea
41. 'Carex lanuginoea
42. Carex eartuellii
43. Carex atipata
44. Carex veaiaaria
45. Chenopodium album
46. Ciouta bulbifera
47. Eleoeharie aaiaularie
48. Eleoeharia compreaaa
49. Eleoaharie paluetrie
50. Equieetum arvenee
51. Equistum fluviatile
VISUAL DOMINANTS
11. Eupatorium manoulatum 21.
12. Eupatorium perfoliatum 22.
13. Helianthue groaaeeaerratua 23.
14. Iria ahrevii 24.
15. Learaia oryzoidee 25.
16. Lemna minor 26.
17. Onoalea eeneibilie 27.
18. Polygonum coecineum 28.
19. Polygonum natana 29.
20. Rumex orbiaulatua 30.
NUMERICAL DOMINANTS
52. Go.li.um tinctorium 72.
53. Junaua braahyoephalie 73.
54. Lyaopue ap. 74.
55. Lyeopue omeriaonua 75.
56. Lyaopue uniflorua
57. Lyaopue. virginiaua 76.
58. lyaimochia ap. 77.
59. Lyeimaehia terrietrie 78.
60. Lyeimaahia thyaiflora 79.
61. Hentha artteneie 80.
62. Mentha epiaata 81.
63. Phlox piloea 82.
64. Piloa pumila 83.
65. Polygonum ep. 84.
66. Polygonum hydropiper- 85.
hydropiperoidea 86.
67. Polygonum eagittatum 87.
68. Potent-ilia paluetria 88.
69. Stum gauve 89.
70. Differ 8
-------
Appendix A (continued)
Species Code
Species Code
No.
Species
Species Code
No.
Species
No.
Species
92. Cerastium vulgatum 138.
93. Cicuta sp.
94. Cirsium vulgare 139.
95. Convolvulus septum 140.
96 Convolvulus spithameus 141.
97. Conyza conadensia 142.
98. Cornus obliqua 143.
99. Cornue racemose 144.
100. Cornus etolonifera 145.
101. Cueouta gronovii 146.
102. Cyperue sp. 147.
103. Eahinochlea pungens 148.
104. Elaocharis obrusa 149.
105. Epilobium coloration 150.
106. Epilobium leptophyllum 151.
107. Equisetum hyemale 152.
108. Ereehtitee hieraaifolia 153.
109. Erigeron sp. 154.
110. Erigeron philadelphiaus 155.
111. Feetuaa elatior 156.
112. Festuaa ovina 157.
113. Geranium maculatum 158.
114. Gerardia paupercula 159.
115. Ceum aleppiaum var. 160.
etrictum . 161.
116. Heleniim autumale 162.
117. Hypericum ka.lmia.num 163.
118. Hypericum ma jus 164.
119. Impatiena biflora 165.
120. Ipomoea biflora 166.
121. Iris virginicim 167.
122. Juncus nodoeus 168.
123. Lathyrus paluetrie 169.
124. Laportea canadensie 170.
125. Lepidium aampeetre 171.
126. Lepidium virginicum 172.
127. lespadesa capitation 173.
128. Liatrus pyanoetachya 174.
129. Lobelia kalmii 175.
130. lobelia eiphilitica 176.
131. Lidvegia palustris 177.
132. ii/8-tmatffcia eiliata 178.
133. Marchantia sp. 179.
134. Medicago lupulina 180.
135. Medicago sativa 181.
136. Melilotue officinalis 182.
137. Mentha piperata 183.
Osmunda regalia var. 184.
speatabilis 185.
Panicwn capiZZare 186.
Panicum flexile 187.
Panicum praecocius 188.
Penthorum eedoides 189.
Phalaris arundinaaea 190.
Phleum pratense 191.
Poa canaieneia 192.
Pi?a compressa 193.
Pax pratensis 194.
Polygonum convolvulus 195.
Polygonum lapathifolium 196.
Polygonum norvegiaa 197.
Polygonum pengylvanioum 198.
Pontederia cordata
Populue ep. 199.
Populus deltoides 200.
Populus tremuloidee 201.
Pycnantheumu virginianum 202.
Ranunculus flabellaris 203.
flaniinculus longirostris 204.
Riccia flutane 205.
Riaaioaarpus 206.
Rorippa islandica 207.
ffoea paZuetr*ts 208.
ffubue ep. 209.
Rubus hispidus 210.
SaHz ep. 211.
Salix bebbiana 212.
Salix Candida 213.
Salix discolor 214.
Salix interior 215.
Salix nigra 216.
Salix petiolaris
Saponaria officinalis 217.
Scirpus ameriaanus 218.
Scirpus atrovirons 219.
Scutellaria galericulata
Scittellaria lateriflora 220.
Setaria viridis
Smilacina stellata
Solomon dulcamara
Solanum nigrton
Solidago sp.
Solidago altiasima
Sporobolus cryptandrus
Stachys hispida
Stachye palustris
Stellaria longifolia
Taraxacum officinale
Thlaspi arvense
Triadenum virginicum
Trifolium pratense
Trifolium repens
Typha anguetifolium
Vaccinum angustifoloium
Verbena hastata
Verbena striata
Veronica fasiculata
Veronica ecutellata
Veroniaaetnm
virginicum
Viaia americana
Vicia villosa
Viola ep.
Viola cucullata
Viola papilionacea
Viola aoforia
Vitus volpina
Urtica dioica
Utricularia Bulgaria
Unknown
Lilum euperbum
Oxalis sp.
Allium canadense
Cardamine pennsylvanica
Ceratophyllum sp.
Ribes ameraanum
Carex sp.
Cephalanthus
oaaidentalis
Chelone glabra
Ringens mimulus
Epilobium
anguetifolium
Open water
.161
-------
APPENDIX B
TABLE B-l. STATION AND TRANSECT NUMBERS*
Station
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Transect
No.
18
20
21
22
23
24
26
26
27
27
27
28
28
28
29
29
29
29
30
30
30
Station
No.
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
Transect
No.
30
31
31
31
31
32
32
32
32
33
33
33
33
33
34
34
34
34
35
35
35
Station
No.
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Transect
No.
35
35
36
36
36
36
37
37
37
37
38
38
38
38
39
39
39
40
40
40
162
-------
APPENDIX C
TABLE C-l. DIVERSITY INDEX
Numbers of Species Found at Each Sampling
Station
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
1974
10
9
8
6
6
5
5
7
7
7
6
10
6
7
6
6
8
6
5
3
5
5
9
6
3
9
5
4
5
6
7
5
1975
7
5
4
5
5
6
7
7
7
6
6
5
5
6
8
6
5
9
6
5
5
7
3
4
5
7
6
5
5
6
11
6
1976
6
4
2
11
2
7
6
4
6
3
3
6
3
6
6
5
5
5
3
5
4
3
4
2
5
5
6
3
4
5
6
4
1977
9
8
2
7
2
3
6
7
5
5
4
5
3
5
3
5
2
4
3
4
2
3
3
1
3
4
3
3
3
3
0
2
Station
No.
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
1974
4
5
6
5
7
5
6
6
9
5
5
5
6
7
5
6
5
9
6
8
5
6
5
5
6
6
5
6
7
6
9
Station
1975
5
4
5
7
8
4
4
7
9
3
3
3
5
6
7
5
9
8
6
6
7
5
6
6
6
5
5
4
3
6
7
1976
4
3
4
4
6
6
5
5
6
4
2
3
6
3
8
4
5
5
8
5
2
5
8
4
6
6
6
6
2
6
6
1977
3
3
4
3
5
3
4
6
8
2
2
2
8
5
5
4
5
9
6
6
3
10
7
4
7
7
8
7
0
8
6
(continued)
163
-------
Appendix C (continued)
1974 1975 1976 1977
379/62 = 6.11 357/62 = 5.76 296/62 - 4.78 266/62 = 4.29
1974 to 1975 1975 to 1976 1976 to 1977
6.11 5.76 4.78
-5.76 -4.78 -4.29
0.35 0.98 0.49
1.82 overall difference
164
-------
APPENDIX D
TABLE D-l. SUBJECTIVE CLASSIFICATION3
Station
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1974
CL
Sp
E
Sp
CL
CL
CS
CS
CL
CS
CS
Sh
T
Sh
T
T
CL
Sh
Sh
T
E
E
Sh
T
T
E
Sh
T
T
CL
CL
1975
CL
Sp
E
Sp
0
CL
CS
CS
CL
CS
CS
Sh
DT
Sh
T
T
DCL
Sh
Sh
T
E
E
Sh
T
E
E
Sh
T
T
T
CL
1976
WA
Sp
E
Sp
0
CL
CS
CS
DCL
DCS
DCS
CS
0
Sh
DT
T
DCL
Sh
Sh
DT
E
E
Sh
0
E
E
Sh
T
DT
DT
DCL
1977
DCS
SP
0
Sp
0
CL
CS
CS
DCL
DCS
DCS
CS
0
Sh
DT
DT
DCL
Sh
Sh
DT
E
E
Sh
0
E
E
Sh
E
E
E
0
Station
No.
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
1974
CL
E
T
E
CS
T
E
E
CL
CL
E
E
E
CS
CL
CL
T
CS
CS
CS
CL
CS
CS
SP
T
CS
CS
CL
CL
CS
CS
1975
CL
E
DT
E
CS
T
E
DE
CL
WA
E
E
E
CS
CL
CL
E
CS
DCS
CS
CS
CS
CS
SP
T
CS
CS
CS
CL
CS
CS
1976
DCL
E
E
DE
DCS
T
E
DE
DCL
WA
DE
E
E
CS
DCL
CL
E
CS
DCS
CS
SH
CS
CS
SP
T
CS
CS
CS
CS
CS
CS
1977
0
DE
0
DE
DCS
DT
DE
DE
WA
WA
0
DE
E
CS
DCL
CS
DE
CS
T
CS
0
DCS
DCS
SP
T
DCS
DCS
DCS
DCS
CS
CS
aCS = Carex stricta; CL = Carex lacustris; E = Emergents; T =• Transition;
Sp = Spiraea; Sh = Shrubs; 0 = Open; DCS = Degraded Carex strieta; DCL =
Degraded Carex laouetris; DE = Degraded emergents; DT = Degraded
transition; WA = Weedy annuals.
165
-------
APPENDIX E
TABLE E-l. ASSOCIATION ANALYSIS3
Station
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1974
CL
Sp
E
CS
CL
CL
CS
CS
CL
SL
CS
CL
CL
SL
CL
CL
CL
Sh
Sh
T
T
E
Sh
CL
T
E
Sh
T
T
CL
CL
1975
CC
CS
E
Sp
CL
CL
CS
CS
T
CS
CS
T
T
T
T
T
T
Sh
T
T
E
E
Sh
T
E
E
Sh
T
T
E
CL
1976
WA
SP
E
Sp
0
CL
CS
DCS
T
DCS
DCS
DCS
0
T
T
T
T
Sh
T
T
E
E
Sh
0
E
E
Sh
E
T
E
DCS
1977
WA
CS
E-O
SP
E-O
DCL
DCS
DCS
DT
DCS
DCS
DCS
DCS
DT
DCS
DT
E-O
Sh
E-O
DT
E
D
E-O
E-O
E
E-O
Sh
E-O
E-O
E
DCS
Station
No.
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
1974
CL
T
T
E
CS
T
T
T
CS
CL
E
E
E
CS
CL
CS
CS
CL
CL
CS
CS
CS
CS
SP
CL
CS
CS
CL
CS
Sp
CS
1975
T
T
E
E
CL
CS
T
E
DCL
CL
E
E
E
CS
CL
CL
CS
CS
CS
CS
Sh
CS
CS
Sp
CS
CS
CS
CS
CS
CS
CS
1976
T
E
E
E
DCS
DCS
DCS
E
0
CS
E
E
T
CS
CS
CS
CS
DCS
DCS
DCS
Sh
CS
CS
CS
"T
CS
CS
CS
CS
CS
CS
1977
DLC
E
E
E
DCS
T
E-O
DE
WA
WA
E-O
E-O
DCS
T
CS
E-O
T
T
T
DCS
Sh
DCS
CS
CS
Sh
DCS
DCS
DCS
CS
WA
CA
aCS = Carex striata; CL = Carex lacustris; T = Transition; E - Emergents;
0 = Open; Sp = Spiraea', Sh «* Shrubs; WA = Weedy annual; DCS = Degraded
Carex striata; DCL = Degraded Carex lacustris; DT = Degraded transition;
DE = Degraded emergents; E-O = Emergents-open.
166
-------
APPENDIX F
TABLE F-l. AIRPHOTO INTERPRETATION KEYS FOR AIRPHOTO GRID ASSESSMENT
Grid Interpretation Keys
June 4, 1972 CIR Scale 1:120,000
1. Sedges and grasses—bright pink tone; smooth texture.
2. Transition—light blue tone; fine texture.
3. Emergent—deep blue tone; fine texture.
4. Spiraea—deep pink tone; very coarse texture.
5. Open water—blue black tone; smooth texture.
7. Shrub carr—deep pink-magenta tone; smooth texture.
July 31, 1974 CIR Scale 1:120,000
1. Sedges and grasses—pinkish red tone; fine texture.
2. Transition—mixed greenish and red tones; fine texture.
3. Emergents—dark greenish tones; fine texture.
4. Spiraea—deep pinkish-red tones; medium texture.
5. Shrubs & trees—dark magenta tone; coarse texture.
6. Open water—none visible in study area on this image.
September 25, 1975 CIR 1:38,200
1. Carex lacustris—light lime green tone; fine texture.
2. Carex stricta—reddish pink; fine texture.
3. Transition—rich medium-toned green; fine texture.
4. Emergents--dark green to black with coarse red areas mixed in; fine
texture.
5. Spiraea—medium reddish-green tone; coarse texture.
6. Shrubs & trees—all shades of mottled reds and green tones; definite
rounded shapes; very coarse texture.
7. Lemna minor—eight whitish pink areas; smooth texture.
8. Open water—dark, black green; smooth texture.
July 24, 1976 CIR 1:19,100
1. Carex lacustris—red tones with greenish tinge to them; pitted texture.
2. Degraded Carex lacustris—red and green mottled tones; coarse textured
and patches of Lemna minor mixed in.
3. Carex stricta—bright red tones; fine texture.
4. Transition—deeper green tones with scattered red; coarse pitted
texture.
167
-------
Table F-l. Continued
Grid Interpretation Keys
5. Emergents—very green tones; clonal shapes of Sairpus fluviatilis and
Typha latifolia; coarse textured.
6. Spiraea—deep red-brown tones; coarse texture.
7. Shrubs and trees—deep red tones; definite round shape; very coarse
texture.
8. Lerrma minor—whitish pink tone; smooth texture.
9. Open water—black tone; smooth texture.
September 24, 1976 CIR 1:19,100
1. Degraded Carex lacustris—red tussocks surrounded with pinkish white
Lemna minor, medium texture.
2. Carex striata—bright red; fine texture.
3. Degraded Carex etriota—red tussocks surrounded with pinkish white
Lemna minor; medium texture.
4. Transition—deep green tones mottled with red; pitted texture.
5. Emergents—deep green tones; coarse texture, some areas of distinctive
clones.
6. Typha latifolia—deep red-brown colored clones; coarse textured.
7. Spiraea alba—greenish red tones; medium tones.
8. Shrubs and trees--green and pink tones; very coarse textured.
9. Lemna. minor—pinkish white tones; smooth texture.
10. Open Water-rdeep green, blue-black tones; smooth texture.
June 25, 1977 Color 1:38,200
1. Sedges and grasses—deep green tone; smooth, fine texture.
2. Degraded Carex striata—brownish areas; fine texture.
3. Transition-emergent—deep green and brown tones; more interspersion
evident than in sedge-grass areas.
4. Typha latifolia—deep green clones; fuzzy texture.
5. Spiraea—green areas with coarse texture.
6. Shrubs and trees—dark green, lumpy textured area.
7. Lemna minor—bright, chartreuse green; smooth texture.
8. Open water—deep green or muddy brown tones.
June 25, 1977 CIR 1:38,200
1. Degraded Carex lacustris—deep red and rust tones with light pinkish
tones of Lemna minor mixed in; fine texture.
2. Carex striata—bright pink tone; fine texture.
3. Degraded Carex etricta—open areas (blue-black) with bright red clumps
(tussocks) and scattered pinkish-white tones (Lemna minor) pitted
texture.
4. Transition—pink toned areas dissected with water and scattered Lemna
minor .
5. Emergents—deep red-brown tones; medium texture.
168
-------
Table F-l. Continued
Grid Interpretation Keys
6. Typha latifolia—bright deep red tones; coarse texture.
7. Spiraea alba—brownish red tones; coarse texture.
8. Shrubs and trees—bright pink, reddish-brown tones; very coarse
texture; round shape.
9. Weedy annuals—light pink toned areas containing very sparse
vegetation.
10. Lerma minor—light-bright pink tones; smooth texture.
11. Open water—blue-black tones; smooth texture.
October 3, 1977 CIR 1:11,500
1. Carex striata—visible greenish tussocks.
2. Degraded Carex striata—green toned clumps (Tussocks) surrounded by
pinkish toned Lemna minor.
3. Sedges and grasses—green and red tones; fine texture.
4. Degraded sedges—sedge areas pock-marked with areas of open water.
5. Spiraea alba—red tones; fine to medium texture.
6. Typha latifolia—bright blue green; medium texture.
7. Emergents—blue green and green tones; clone shape with coarse texture.
8. Lemna minor—bright pink areas; fine texture.
9. Open water—deep blue tones; smooth texture.
10. Shrubs and trees—pink, red and green tones; definite round crown
shapes; very coarse texture.
169
-------
APPENDIX G
TABLE G-l. AIRPHOTO INTERPRETATION KEYS FOR VEGETATION MAPPING
Interpretation Keys
June 4, 1972 1:120,000 CIR
1. Sedges and grasses--bright pink tone with a smooth texture.
2. Transition—light blue tone with a fine texture.
3. Emergents—deep blue tones; fine texture.
4. Spiraea—deep pink tone with a coarse texture.
5. Shrubs--magenta tone; very coarse texture.
6. Shrub carr—deep pinkish-magenta tones with a very coarse texture.
7. Open water—blue black tone with a smooth texture.
8. Trees—bright, deep magenta tone; very coarse texture.
July 31, 1974 1:120,000 CIR
1. Sedges and grasses—pinkish red color; fine texture.
2. Transition—greenish and red tones mixed; fine texture.
3. Emergents—dark greenish tones; fine texture.
4. Spiraea.—deep red color and medium texture.
5. Shrubs and trees—dark magenta tones with coarse texture.
6. Open water—black-blue tones with a flat texture.
September 25, 1975 CIR Scale 1:38,200
1. Carex lacustrie—bright lime green; fine texture.
2. Carex striata—reddish pink; fine texture.
3. Degraded Carex etricta—greenish red clumps in pink matrix of Lerma
minor*.
4. Transition—rich medium toned green; fine texture.
5. Degraded transition—greenish areas with interspersion; fine textured.
6. Emergents—dark green to black with coarse red areas mixed in; fine
textured along stream.
7. Scirpus fluviatilis—bright white green area; coarse texture at back of
site in the southern end of the site.
8. Spiraea alba—medium reddish-green tone; coarse texture primarily in
the southern end of the site.
9. Shrubs—very bright pink red tone with a coarse texture.
10. Lerma minor—very bright whitish pink area with a flat texture.
11. Open water—dark black green with a smooth texture.
170
-------
Table G-l. Continued
Interpretation Keys
12. Shrub carr—coarse greenish-reddish area at southern end of study
site. Individual shrub crowns visible.
13. Trees—very red tones; coarsest vegetation on the photo; all tree
crowns visible.
July 26, 1976 CIR 1:19,100
1. Carex lacustris—red tones with greenish tinge to them; pitted texture.
2. Degraded Carex lacustris—red and green mottled tones with patches of
Lernna minor mixed in; coarse textured.
3. Carex stricta—bright red tones; very fine texture.
4. Degraded Carex stricta—red clumps surrounded by Lerma minor (flat
textured, pinkish toned vegetation).
5. Transition—deep green tones with scattered red tones; coarse pitted
texture.
6. Degraded transition—greenish red tones with much interspersion; fine
textured.
7. Bnergents—very green tones; coarse textured; often clonally shaped.
8. Typha latifolia—most deep, intense, true red on airphoto with a medium
coarse texture. Usually areas of open water nearby.
9. Scirpus fluviatilis—rosy red tones, with medium coarse texture located
along the forest on the far side of the stream.
10. Spiraea alba—deep red brown tones; a definite tuft texture.
11. Shrubs—deep red tones; a definite round shape creating a very coarse
texture.
12. Weedy annuals—pinkish red areas with Lerma mixed in; medium texture.
13. Lerma minor—very bright pinkish white areas; smooth textured.
14. Open water—deep green, black tones.
15. Shrub carr—bright red tones; very coarse tufted texture, southo f the
outflow channel.
16. Trees—bright and deep red tones with a very coarse texture.
September 24, 1976 CIR Scale 1:19,100
1. Carex lacustris—pinkish green areas along dike; fine texture.
2. Degraded Carex lacuatris—blue-green and red dots surrounded with
pinkish white Lernna minor; medium texture.
3. Carex stricta—deep red; fine textured clones along dike and out in the
center of the study site south of the keyhole well.
4. Degraded Carex stricta—red tussocks (dots) surrounded with pinkish
white Lernna minor; medium texture.
5. Transition—deep green tones mottled with red; pitted texture.
6. Degraded transition—greenish areas with much interspersion; medium
coarse textured.
7. Emergents—deep green tones; coarse texture; some distinctive clones.
8. Typha latifolia—deep red-brown colored clones; coarse textured areas.
9. Scirpus fluviatilis—brown-red area against forest; coarse textured.
171
-------
Table G-l. Continued
Interpretation Keys
10. Spiraea alba—very fine tufted texture; mottled red-khaki tones.
11. Shrubs—pink and red tones; very coarse textured; visible round crowns.
12. Weedy annuals—pinkish white areas with scattered sparse vegetation.
13. Floating mat—whitish, flat textured areas with sparse vegetation.
14. Lemna minor—pinkish white areas; a smooth, solid appearing texture.
15. Open water—deep green-blue black tones with a smooth texture.
16. Trees—rosy red tones with coarsest texture on the photograph.
June 25, 1977 CIR Scale 1:38,200
1. Carex lacustrie—reddish, fine textured; sparse scattered clones in
south end of study site.
2. Degraded Carex laauetris—deep red and rust tones with light pinkish
tones of Lemna minor mixed in; fine texture.
3. Carex etriata—deep rosy tone; dense vegetation; fine texture.
4. Degraded Carex striota—open areas with bright red clumps and scattered
pink-white tones; pitted textures.
5. Transition—pink toned areas dissected with water and scattered Lemna
minor.
6. Degraded transition—greenish areas with much interspersion and Lemna*
7. Emergents—deep red-brown tones; medium texture.
8. Typha latifolia—bright deep red tones; medium texture.
9. Sairpue fluviatilis—very intense pink areas with medium coarse texture
against forest behind stream.
10. Spiraea alba—brownish-red tones with a coarse tufted texture.
11. Shrubs—deep red-pink; distinct rough shape; scattered throughout study
site, along dike and along overflow channel.
12. Weedy annuals—light bluish pink toned areas containing sparse
vegetation.
13. Floating mat—blue-greenish white areas along areas of open water.
14. Lemna minor—light pink toned areas containing very sparse vegetation.
15. Open water—blue black tones; smooth texture.
16. Shrub carr—coarse textured, deep magenta tones in area south of
overflow channel near south knoll.
17. Trees—deep red magenta tones; most coarse texture in scene; tree
crowns very distinctive.
June 25, 1977 Color Scale 1:38,200
1. Carex laoustris—deep green tone; smooth fine texture.
2. Degraded Carex laoustris—large mottled tannish areas; fine texture.
3. Carex stricta—medium green; fine textured.
4. Degraded Carex stricta—brownish areas; fine texture.
5. Transition—deep green and brown tones with much interspersion.
172
-------
Table G-l. Continued
Interpretation Keys
6. Emergents—mottled deep green toned areas with a medium coarse texture
at back of site.
7. Typha latifolia—deep green clones; fuzzy textured.
8. Scirpus fluviatilis—intense deep green tones with medium texture;
along forest.
9. Spiraea alba—green areas with a coarse texture.
10. Shrubs—coarse textured green areas—difficult to see unless a number
of them exist together.
11. Weedy annuals—mottled green-white areas, fine textured.
12. Floating mat—brownish green areas adjacent to open water in center of
scene.
13. Lemna minor—bright chartreuse green; smooth textured.
14. Open water—deep green and muddy brown tones.
15. Shrub carr—coarse textured green area near the south knoll.
16. Trees—very deep, forest green tone with coarsest texture on the
airphoto.
October 3,. 1977 Scale 1:11,500
1. Carex laauetris—deep greenish pink toned clones in south end of study
site; fine textured.
2. Degraded Carex laoustris—whitish pink areas with much interspersion;
fine textured.
3. Carex stricta—visible greenish tussocks.
4. Degraded Carex striata—green toned tussocks surrounded by Lemna minor.
5. Transition—green toned areas mottled with red; some interspersion;
texture medium.
6. Emergents—blue-green and green tones; clone shaped areas; coarse
textures.
7. Degraded emergents—mostly water cress with sparse emergent vegetation.
8. Scirpus fluviatilis—-large, medium coarse textured reddish-green clone
to back of study site.
9. Spiraea alba—coarse textured, deep russet-red areas out along the
overlow channel.
10. Shrubs—quite reddish tones; individual round crowns visible throughout
study site.
11. Weedy annuals—water and interspersed sparse vegetation.
12. Floating mat—large scattered pink tones; areas with coarse texture and
much interspersion.
13. Lemna minor—light pink toned, flat textured areas.
14. Open water—deep blue tones; smooth texture.
15. Shrub carr—area south of overflow channel-brown-green red tones with
coarse texture.
16. Trees—pink-red tones; coarsest texture; individual tree crowns
visible.
173
-------
APPENDIX H
ANNOTATED BIBLIOGRAPHY
REMOTE SENSING—GENERAL INFORMATION
Ashley, M. 0., W. W. Knapp, and J. Rea. 1974. Phenological Data From the
ERTS-1 Satellite. In: Proceedings of the 2nd Canadian Symposium on
Remote Sensing. Guelph, Ontario, p. 662-667.
The analysis, procedures, and results of research utilizing Earth Resources
Technology Satellite multispectral scanner data to study phenological or
seasonal changes in forest, crop, and range vegetation are discussed. This
study was undertaken by a multidisciplinary group of researchers working
with data from several areas of the United States. Visual imagery
interpretation, band-to-band density ratios and computer generated band
ratio parameters are used to show local and regional phenological events for
the 1972-1973 fall recession and spring progression of vegetation
development. Image density measurements put in ratioform using band 5 (red
wavelengths) and band 7 (near infrared wavelengths) Imagery correlate well
with forest vegetation changes documented by ground observation photography.
This ratio T,an. , ~ _ j- , is lowest with leaf off and steadily increases
Band 5 + Band 7 , J
with leaf development. Typical values are 0.08 for leaf off to 0.26 for
full leaf in mid-summer. Band ratio parameters calculated from reflectances
recorded on the scanner's computer computable tapes are also shown to be
well correlated with these developments. This ratio nan. , . nan. e follows
f Band 7 + Band 5
the same pattern as the density ratio. Visual interpretations of positive
transparencies indicated a decreasing red reflectance and increasing
infrared reflectance with vegetation development. Conclusions are that
phenological events for crops, forest, and range land can be predicted
through visual interpretation, image density measurement and computer
calculated reflectance methods.
Aldred, A. H. 1972. Decisions on Combining Data From Several Sensors.
In: Proceedings of the 1st Canadian Symposium on Remote Sensing.
Ottawa, Ontario, p. 681-690.
The purpose of the paper is to set up a cost effectiveness model for
evaluating the efficiency of combining information collected from satellite
imagery, large-scale photos and ground measurements. A forest inventory
problem is used to illustrate the approach and to indicate the cost
advantages of using more than one imaging medium for certain problems. Some
criteria are given for deciding when to use multiple sources of image data.
Allen, W. A., H. W. Gausman, A. J. Richardson, and C. L. Wiegand. 1970.
Mean Effective Optical Constants of Thirteen Kinds of Plant Leaves.
Applied Optics 9:2573-2577.
174
-------
Plant leaves grown in a greenhouse and leaves collected from the field have
been analyzed to obtain mean effective optical constants based upon diffuse
reflectance and transmittance measurements taken over the 0.5-2.5- spectral
range. These optical constants are used in a generalized flat-plate model
to describe the phenomena of leaf reflectance. The analytical procedures
that were developed led to measurement of the amount of water and
Intercellular air spaces in the leaves. Over the 1.4-2.5 spectral range,
the adsorption spectra o leaves are not statistically different from that
of pure liquid water. Leaf reflectance differences among the plant leaves
over the 0.5-1.4 range are caused principally by Fresnel reflections at
external and internal leaf surfaces and by plant pigment absorption.
Reflectance over the 1.4-2.5 ) range results largely from Frenel reflections
and adsorption by water. Data are presented in the form of dispersion
curves with 95% confidence bands and tabulated plant leaf t adsorption
spectra. The dispersion curves were assumed to be cubic equations of the
form a , (1=0, 1, 2, 3), where is wavelength. Reflectance
measurements at 1.65 have been associated with the equivalent water
thickness and the intercellular air spaces in the leaf. Accuracy of the
plate theory based upon a cubic dispersion curve is shown to be within
experimental error.
Anson, Abraham. 1966. Color Photo Comparison. Photo. Eng. 32:286-297.
This article is the result of action by the Color Photography Committee of
the American Society of Photogrammetry in which panchromatic, color and
Ekcachrom IR photography of the same area were compared under essentially
identical conditions. The study includes the Identification and
interpretation of drainage, vegetation, soils, and map features such as
roads, railroads, and buildings. As a control the same features were
identified on the ground. In addition, the photoInterpreters were required
to identify 42 selected photopoints that appeared in photographs. On the
basis of the limited study, Ektachrome IR photography proved to be superior
to both color and panchromatic photography for mapping, vegetation, and
drainage. Color photography was found to be superior to panchromatic and
Ektachrome IR for mapping soils and culture.
Ballard, R. J., and L. F. Eastwood, Jr. 1977. Estimating Costs and
Performance of Systems for Machine Processing of Remotely Sensed Data.
In: Fourth Annual Symposium on Machine Processing of Remotely Sensed
Data. LARS, West Lafayette, Indiana, p. 208-214.
This paper outlines a method for estimating computer processing times and
costs incurred in producing information products from digital remotely
snesed data. The method accounts for both computation and overhead, and it
may be applied to any serial computer. The analysts apply the method to
estimate the cost and computer time involved in producing Level II Land Use
and Vegetative Cover Maps for a five-state, midwestern region. Their
results show that the amount of data to be processed overloads some computer
systems, but that the processing is feasible on others.
175
-------
Barry, F. K., and J. A. Smith. 1977. An Overview of Vegetation Canopy
Modelling for Signature Correction and Analysis. In: Fourth Annual
Symposium on Machine Processing of Remotely Sensed Data. LARS, West
Lafayette, Indiana, p. 194.
Modeling of the interaction of solar radiation with vegetation canopies
offers a tool for sensor design, signature extension, and relating intrinsic
scene parameters to composite scene response. Theoretical approaches
include both the deterministic solution of a system of simultaneous
differential equations and Monte Carlo Modeling which treats the canopy as
consisting of layered statistical ensembles of foliage elements against a
soil background. In this paper the authors discuss several applications of
canopy modeling to the general problem of understanding and correcting
signature variations. Discussion will emphasize a Monte Carlo model that
was originally developed to Investigate the bidirectional reflectance
character of natural grasslands. Subsequently, as part of the Large Area
Crop Inventory Experiment, the model was used to analyze wheat reflectance
dependence on both diurnal and crop development variation. LANDSAT response
was simulated by interfacing the canopy reflectance model with an
atmospheric radiation transfer model. The combined model predictions were
used to develop correction coefficients for sun angle effects in wheat and
to investigate signal variations induced by soil brightness. Research into
the feasibility of utilizing model-derived data to infer intrinsic scene
variables through divergence classification was also conducted. The model
is currently being modified for forest canopies to study scene mixture and
sun angle effects in this context.
Bauer, Kenneth G., and John A. Outton. 1962. Albedo Variations Measured
From an Airplane Over Several Types of Surface. J. of Geophys. Res.
67(6):2367-2376.
Albedo values for four different types of terrain were observed at
semiregular intervals over south-central Wisconsin from October 1959 to July
1960 with instruments mounted on a light twin-engined aircraft. It was
found that albedo has only two basic seasonal vaues—snow or no snow.
Values between 10 and 20 percent were observed over agricultural land areas
in snow-free seasons. With snow, the albedo values were as high as 80
percent over a frozen lake and as low as 50 percent over wooded hills. The
instrument installation is discussed -and it is shown how a "beam" system may
be calibrated against a hemispherical system. Measured data from both
systems then agreed over land. Within 1,000 ft of the ground, the measured
albedo decreased less than 5 percent. Data on cloudedge effects, on albedo
changes observed during descent through a cloud layer, and on the albedo of
a lake preceding the spring breakup of the ice cover are presented.
Beckett, P. H. T. 1972. The Statistical Assessment of Resource Surveys by
Remote Sensors. In: Environmental Remote Sensing: Applications and
Achievements. Eric C. Barrett and Leonard F. Curtis, eds. Crane &
Russak, New York. p. 11-27.
Published discussions on the use of remote sensing procedures for obtaining
information about natural resources or the environment contain many
unsupported value judgments (e.g., "better than", "more information than").
176
-------
The utility of remote sensing procedures should be assessed on the truth and
precision of the information (statements) they provide, and the costs of
obtaining it. This paper lists the kinds of statement which may be required
from remote sensors, and offers a preliminary review of methods for judging
the success of remote sensors in providing diem. The same methods may be
used for the quality control of resource surveys by remote sensors,
performed under contract.
Blanchard, Bruce J., and Ross W. Learner. 1973v Spectral Reflectance of
Water Containing Suspended Sediment. In: Remote Sensing and Water
Resources Management, Proc. No. 17. p. 339-347.
A spectral radiometer, measuring radiation in the visible and near-infrared
portion of the spectrum, was used to examine (1) different concentrations of
red, black, and gray clay particles in water, and (2) several samples of
natural pond water containing sediment. Four of the pond samples had algae
present. Density measurements were made of color and color infrared film
photographs of the sample exposed at the same time as the radiometer
measurements. Reflectance curves in the near-infrared region show very
little change caused by changing sediment concentrations. However,
reflectance curves in the visible portion of the spectrum are sensitive to
very low concentrations (less than 200 ppm suspended solids) of sediment
with similar characteristics. The samples containing algae showed a good
possibility of detecting algae by ratioing the reflectance near wavelength
570 nm with the reflectance at 630 nm. Response at 570 nm appears to be
related to the suspended sediment. Results using the film density
measurements also show promise for use in estimating low sediment
concentrations; however, the best combination of film and filter are not
known for the different sediment characteristics. (KEY TERMS: Sediment,
remote sensing, spectra, reflectance, visible light, near-infrared, water
quality.)
Bogard, Jacqueline A. 1974. A Comparative Analysis of Remote Sensing
Imagery for Vegetation Studies. 50 p. For: GEE 552, University of
Wisconsin.
Remote sensing has been recognized as a valuable tool for vegetation
studies. A considerable amount of research has been done to determine which
sensing system is optimal for particular plant formations. Although every
type of imagery reviewed in this paper possesses certain inherent
advantages, color and color infrared have the widest range of capabilities.
I found, however, that the inexpensive panchromatic photographs (scale
1:20,000) from the surveys of the Agriculture Stabilization and Conservation
Service provided a valuable data source for monitoring land use and
vegetation patterns over time.
Buchanan, Warren J. 1978. Applications of Digitized Film Analyses and
Perceptions of Possible Users. M.S. Thesis, University of Wisconsin-
Madison, Madison, Wisconsin. 147 p.
70mm color infrared transparencies of upland and lowland resource complexes
were digitized with a drum-type scanning microdensitometer. The effects of
seasonal change and resolution on digital signature behavior were
177
-------
investigated along with signature extendability, classification accuracy,
time and cost efficiency of resource classification. Seasonal studies
indicated that, for digital analysis, at least some signatures were
confounded in all seasons, but these confounded signatures shifted through
the growing season such that a combination of seasonal data would
successfuly separate nearly all resources. As the resolution cell increased
in size from 0.25 m to 14.59 m , the resource signatures decreased in width
and became more discrete; however, linear resources such as roads were
unresolved with the largest resolution cell. Signatures appeared to be
extendable from image to image if images were: 1) from the same roll of
film; 2) were processed the same; 3) were exposed within a short time span;
4) were corrected and calibrated the same; and 5) were scanned at the same
time. The best classification accuracy achieved was 85% until these
advantages are repeatedly verified, demonstrated, and documented, the growth
of this technology will probably be gradual at best. Problems such as
unfamiliariy or inaccessibility can be overcome simply by increasing
awareness of existing facilities. The acceptance of digitized film analysis
will always behindered by its perceived disadvantages: expense and
inaccuracy.
Carter, P., and W. E. Gardner. 1977. An Image-Processing System Applied to
Earth-Resource Imagery. In: Environmental Remote Sensing. E. C.
Barrett and L. F. Curtis, eds. p. 143-162,
The Harwell Image Processing System (HIPS) has been adapted for processing
earth-resource imagery in either film or tape format. Data from film is
obtained using a computer-controlled flying-spot scanner. Local rapid
interactive processing is based on a PDP 11/20 mini-computer which has
suitable display facilities for immediate visual appraisal of results and
also a fast data link to an IBM 370/168 computer complex. An extensive
subroutine library is being assembled for data preprocessing and feature
extraction. This chapter includes a discussion of the basic principles of
image analysis, a description of the HIPS system, and finally, for
illustrative purposes, a description of several simple software routines.
Clegg, Robert H. 1975. Accuracy, Resolution and Cost Comparisons Between
Small Format and Mapping Cameras for Environmental Mapping. In:
Proceedings of the American Society of Photogrammetry, 41st Annual
meeting. Washington, D.C. p. 663-691.
Successful aerial photography depends on aerial cameras providing acceptable
photographs within the cost restrictions of the job. For topographic
mapping, where the ultimate accuracy is required, only large format mapping
cameras will suffice. For mapping environmental patterns of vegetation,
soils, or water pollution, 9-inch cameras often exceed accuracy and cost
requirements, and small formats may be better. In choosing the best camera
for environmental mapping, relative capabilities and costs must be
understood. This study compares resolution, photo interpretation potential,
metric accuracy, and cost of 9-inch, 70 mm, 35 mm cameras for obtaining
simultaneous color and color infrared photography for environmental mapping
purposes.
178
-------
Clous ton, John G. 1950. The Use of Aerial Photographs in Range Inventory
Work on the National Forests. Photo. Eng. 16:320-331.
The use of aerial photographs in making range inventories is described. The
author believes that use of airphotos greatly increases the accuracy of such
inventories since they clearly define range types and make type checking
eas ier.
Collins, S. H. 1972. The Block Adjustment of Colour in High-Altitude
Photography. In: Proceedings of the 1st Canadian Symposium on Remote
Sensing. Ottawa, Ontario, p. 659-575.
This paper describes a complete program of density calibration for color and
multispectral high-altitude photography. The central technique is a "block
adjustment" of colour, considered as a multi-dimensional variable, over a
complete block of frame photography; in a manner analogous to the block
adjustment of point location in analytical photogrammetry. The images of
selected ground areas are located in the overlaps between frames and between
flight-lines. The colours of these "pass points" are used to determine
correction functions for colour variations that occur within and between
frames throughout the roll. The part of the sun-angle effect which is due
to the nature of the terrain is suggested as a powerful discriminant for
automatic photointerpretation. A method is also described for the absolute
calibration of the whole block for ground radiance. The method is
photographic, and it ties the radiance values to a great variety of known
terrain and cover types on a regional basis. The value of this work in
providing comprehensive ground truth for satellite imagery over a large
region is discussed.
Colewell, R. N., and D. L. Olson. 1964. Thermal Infrared Imagery and Its
Use in Vegetation Analysis by Remote Aerial Reconnaissance. In:
Proceedings of the Third Symposium on the Environment. Ann Arbor,
Michigan, p. 607-621.
In recent years it has become acutely apparent that from both the military
and civil standards, there is a need for some rapid, accurate and economical
means of analyzing vegetation. This paper will consider the various kinds
of useful information which military and civil experts can obtain regarding
vegetation from the use of a thermal infrared mapping system which operates
in the 7-15 micro band. Special emphasis will be placed on the value of
this Imagery when used in conjunction with Imagery obtained in the visible,
near infrared, and near ultraviolet portions of the electromagnetic spectra.
Colwell, John E. 1974. Vegetation Canopy Reflectance. Remote Sensing of
the Environment 3:175-183.
Possible cause-effect relationships in producing vegetation canopy
reflectance are discussed. Hemispherical reflectance and even bidirectional
reflectance measurements are shown to be inadequate for predicting or
understanding vegetation canopy relfectance in many situations. Among the
additional important parameters necessary for prediction and understanding
of vegetation canopy reflectance are leaf hemispherical transmittance, leaf
area and orientation, characteristics of other components of the vegetation
canopy (stalks, trunks, limbs), soil reflectance, solar zenith angle, look
179
-------
angle, and azimuth angle. The effects of these parameters on vegetation
canopy bidirectional spectral reflectance are described.
Conn, Jeffery S., Kenneth C. Foster, and William G. McGinnies. 1975. The
Nature of Spectral Signatures in Native Arid Plant Communities. In:
Proceedings of the Am. Soc. of Photogrammetry Fall Convention, Phoenix,
Arizona, p. 876-883.
Radiometric data in ERTS bands 5 and 7 of spectral signature components were
compared to the overall signatures obtained from an airborne radiometric
data collection system flown at low altitude. Results indicate that due to
the low density and low vigor of the vegetation, vegetation has little
effect on the overall signature, thus making differentiation of desert plant
communities on the basis of spectral signature extremely difficult.
Cooper, Charles F. 1964. Potential Applications of Remote Sensing to
Ecological Research. In: Proceedings of the 3rd Symposium on Remote
Sensing of the Environment. Ann Arbor, Michigan, p. 601-606.
Field investigations of the behavior of natural plant communities require
knowledge of physical and biological characteristics integrated over areas
of a few square feet to several square miles. Important properties
potentially measurable by remote sensing techniques, singly or in
combination, include leaf area, volume, weight, and chlorophyll content of
vegetation; heat budgets of vegetated surfaces; qualitative and quantitative
local differences in water vapor and carbon dioxide fluxes; water content of
soils and vegetation; and depth and density of snow. Some Implications of
these measurements for understanding of ecological processes are
discussed. Close collaboration between instrumentation engineers and field
biologists is essential if best results are to be obtained.
Cox, T. L., H. C. Hitchcock, and S. G. Weber. 1975. Processing of Remotely
Sensed Data for Dimensional Analysis. In: Symposium Proceedings of
Machine Processing of Remotely Sensed Data. LARS, West Lafeyette,
Indiana, p. 18-37—18-44.
Forest inventory data was interpreted from color IR photography, transferred
to base maps, and digitized for machine processing. The data was registered
to geodetic coordinates providing the capability to perform several types of
dimensional analysis. Processing data by this technique allowed: (1)
spatial or single variable analysis, (2) overlay or composite analysis (in
combination with other variables), and (3) temporal analysis. Information
derived from this procedure was input for a land management decision system
used to construct a forest management plan for 25,000 acres in east
Tennessee.
Cummings, Robert, and Robert R. Jayroe, Jr. 1973. Unsupervised
Classification Techniques as Components of a Data and Information
System. In: Am. Soc. of Photogrammetry Symposium Proceedings, Mngt. &
Utilization of Remote Sensing Data. Sioux Falls, S.C. p. 248-256.
The phenomenal increase in the amount of data and information being
generated by remote sensing systems is stressed. A total system design
approach as a solution to this problem is discussed with specific reference
180
-------
to the data and information system needs for Sortie Lab—a multiple use
payload for the Shuttle. The development of a raultispectral data processing
system as a needed component of such a system is reviewed with emphasis on
unsupervised classification techniques developed and presently in use at
Marshall Space Flight Center.
Curtis, L. F. 1977. Ground Monitoring for Airborne and Space Sudies of
Land Use and Soil Conditions. In: Environmental Remote Sensing. E. C.
Barrett and L. F. Curtis, eds. Edward Arnold, London, p. 192-215.
Published discussions on the use of remote sensing procedures for obtaining
information about natural resources or the environment contain many
unsupported value judgments (e.g., "better than", "more information than").
The utility of remote sensing procedures should be assessed on the truth and
precision of the information (statements) they provide, and the costs of
obtaining it. This paper lists the kinds of statements which may be
required from remote sensors, and offers a preliminary review of methods for
judging the success of remote sensors in providing them. The same methods
may be used for the quality control of resource surveys by remote sensors,
performed under contract.
Derenyi, Eugene E. 1972. Geometric Considerations in Remote Sensing.
In: Proceedings of the 1st Canadian Symposium on Remote Sensing.
Ottawa, Ontario, p. 547-550.
In order to judge the potentials of Image-forming remote sensors properly,
both the spectral characteristics and the geometric aspects must be
considered. With respect to the latter one, two factors are of primary
importance: (1) Geometric or spatial resolution which, in the case of an
optical mechanical scanner, is a function of the instantaneous angular field
of view and for a radar system depends on the accuracy of the time
measurement, on the slant range and on the beam width. (2) Geometric
fidelity, which is influenced by the inherent distortions of the sensor,
distortion characteristics of the photographic material, terrain and
environmental conditions, and by fluctuating in the attitude, altitude and
velocity of the airborne vehicle carrying the sensor. Both factors are
discussed in detail. From a geometric point of view, the performance of
unorthodox image forming sensors is inferior to that of modern frame camera
systems. Results of a theoretical investigation and of a test conducted on
infrared scanner Imagery are presented as a proof. The use of stabilized
platforms and analytical or analogue image restitution is suggested to
improve the geometric fidelity of dynamic Imagery.
El-Baz, F. 1978. The Meaning of Desert Color in Earth Orbital
Photographs. Photo. Eng. 44(l):69-75.
The color of desert surfaces as seen in Earth orbital photographs is
indicative of soil composition. Apollo-Soyuz photographs of the Sturt and
Simpson Deserts of Australia confirm that sand grains become redder as the
distance from the source increases. Reddening is caused by a thin iron-
oxide coating on individual sand grains and can be used, in some cases, to
map relative-age zones. Photographs of the Western (Libyan) Desert of Egypt
Indicate three distinct and nearly parallel color zones that have been
181
-------
correlated in the field with: (1) arable soil composed of quartz, clay, and
calcium carbonate particles; (2) relatively active sand with or without
sparse vegetation; and (3) relatively inactive sand mixed with dark (desert-
varnished) pebbles. The youngest sands are in the form of longitudinal
dunes, which are migrating to the south-southeast along the prevailing wind
direction. Some of the young dune fields are encroaching on the western
boundary of the fertile Nile Valley.
Friederichs, G. A., and F. L. Scarpace. 1977. A Method of Determining
Spectral Dye Densities in Color Films. In: Proceedings of the Am. Soc.
of Photogrammetry, 43rd Annual Meeting. Washington, D.C. p. 257-279.
A straight forward method for the user of color imagery to determine the
spectral density of the dyes present in the processed imagery is
presented. The method involves exposing a large number of different color
patches on the film. The number of different patches necessary to span the
gamut of the film's imaging capabilities has been investigated. From
integral spectral density measurements at sixteen different wavelengths, the
unit spectral dye curves for each of the three dyes present were
determined. The spectral density measurements were subjected to a
characteristic vector analysis which determined a set of eigenvalues and
eigenfunctions for the set of exposed color patches. The best linear
combinations of the eigen vectors fit to the published spectral dye curves
were determined to be the spectral dye densities of the film after
processing. A discussion of the use of these spectral dye densities to
determine the transformation between integral density measurements and
analytical density is presented.
Fritz, Norman L. 1967. Optimum Methods for Using Infrared-Sensitive Color
Films. Phot. Eng. 33:1128-1138.
Considerable interest has currently been expressed in the potential of Kodak
Ektachrome Infrared Aero Film, Type 8443, as a remote sensor for
applications as diverse as aerial reconnaissance and the detection of
diseases and pests in agricultural crops. The results obtained with this
film can be optimized through a knowledge of some of its special
characteristics, and by using photographic techniques which take advantage
of its unique properties. Consideration of the typical scene
characteristics indicates that the principal applications at the present
time involve the photography of foliage. By observing appropriate methods
for storing, exposing and processing, one Is assured of obtaining
photographs having the highest information content.
Gausman, H. W., et al. 1978. Distinguishing Succulent Plants from Crop and
Woody Plants. Photo. Eng. 44(4):487-491.
The analysts compared laboratory spectrophotometrically measured leaf
reflectances of six succulents (peperomia, possum-grape, prickly pear,
spiderwort, Texas tuberose, wolfberry) with those of four nonsucculents
(cenizo, honey mesquite, cotton, sugarcane) for plant species
discrimination. Succulents (average leaf water content of 92.2 percent)
could be distinguished from nonsucculents (average leaf content of 71.2
percent) within the near-infrared water adsorption waveband (1.35 to
182
-------
2.5 m). This was substantiated by field spectrophotometric reflectances of
plant canopies. Sensor bands encompassing either the 1.6- or 2.2- m
wavelengths may be useful to distinguish succulent from nonsucculent plant
species.
Gausman, H. W. 1977. Reflectance of Leaf Components. Remote Sensing of
Environ. 6:1-9.
The reflectance of leaf components was evaluated over the 370 to 1100 nm
wavelength interval. Kodak high speed, black-and-white infrared photographs
at 850 nm showed that: (1) Leaf epidermises of Elodea (Anacharis
canandensis, Planch.) and Lemna L. diffused incoming infrared light; (2)
infrared light was reflected from surfaces inside leaves of Rhoeo discolor
Hance, through stomatal apertures; and (3) crystals and chloroplasts in the
expressed sap of Zebrina pendula Schnizl. reflected infrared light. Scans
(370 to 1100 nm) showed that reflectance from complex cell walls of Aqavae
americana L. compared with that of the adjacent cytoplasm was significantly
greater (p = 0.01) than the reflectance of simpler cell walls of Heliconia
humile L. compared with that of the adjacent cytoplasm. The reflectance of
Vicia faba L. nuclei was larger (significant, p * 0.10) than that of
adjacent cell areas. Results show that refractive index discontinuities in
leaves cause the reflectance of near-infrared light.
Goodenough, David, and Seymour Shlien. 1974. Results of Cover Type
Classification by Maximum Likelihood and Parallelepiped Methods. In:
Proceedings, 2nd Canadian Symposium on Remote Sensing. Gueleph,
Ontario. Vol. I. p. 135-164.
This paper describes the results of automatic ground cover classification
utilizing the spectral intensities of ERTS-1 images. The methodology used
for the interactive software classifier has been described in the preceding
paper (Shlien and Goodenough 1974). The classifier was used to distinguish
crops and different types of vegetation and water in Manitoba and Ontario.
The results are presented in the form of colour photographs showing regions
before and after classification. The effects of ratioing and radiometric
calibrations on the classifications are also visually presented. The
accuracies of the classification are discussed. The lowest classification
accuracies occurred with crop identification.
Grinnell, H. Rae. 1972. The Economics of Remote Sensing of Forest Land.
In: Proceedings of the 1st Canadian Symposium on Remote Sensing.
Ottawa, Ontario, p. 691-696.
The output of remote sensing systems are discussed in terms of economics,
early development and effective use in Canada. Opportunities to increase
benefits from current systems depend on clear objectives for the multiple
use of a limited number of image resolutions derived at specific time
intervals. The frationated incomplete photo coverage of Canada and the low
unit cost of standardized scales are suggested as ample reasons to bring
about some rationalization of the current multi-view approach to resource
surveying in Canada.
183
-------
Gustafson, T. D., and M. S. Adams. 1974. Remote Sensing of Myriophyllum
Spicatum L. in a Shallow, Eutrophic Lake. In: Remote Sensing and Water
Resources Mngt., Proc. No. 17, Am. Water Resources Assoc. p. 387-391.
An aerial 35 am system was used for the acquisition of vertical color and
color infrared imagery of the submergent aquatic macrophytes of Lake Wingra,
Wisconsin. A method of photographic interpretation of stem density classes
is listed for its ability to make standing crop biomass estimates of
Myriophyllum spicatum. The results of film image density analysis are
significantly correlated with stem densities and standing crop biomass of
Myriophyllum and with the biomass of Oedogonium mats. Photographic methods
are contrasted with conventional harvest procedures for efficiency and
accuracy. (KEY TERMS: Myriophyllum; Oedogonium; Aquatic Macrophytes;
Biomass; Water quality; Eutrophication; Limnology; Ecology; Color, color
infrared aerial photography.)
Hallert, Bertel. 1970. Calibration and Tests of Hasselblad EL-Data Camera
No. 12123. In: 1970 International Symposium on Photography and
Navigation. Columbus, Ohio. p. 309-327.
The new Hasselblad-Hallert camera, used during the latest moon expeditions,
has been carefully tested and calibrated in accordance with I.S.P.-
resolutions and recommendations. The photographs in black-and-white as well
as in color have been found to have excellent geometrical qualities. The
quality of the final results of analytical and analogue stereoscopic
restitution is in full agreement with theoretically expected data as
determined from the basic quality of image coordinates and y-parallax
measurement. The camera can be expected to be of the greatest value for
terrestrial and aerial photogrammetry.
Hansen, Jack H. 1973. Color and Color Variation of a Hardwood Forest as
Imaged on Color Infrared Film. In: Proceedings of the Am. Soc. of
Photogrammetry, 39th Annual Meeting. Washington, D. C. p. 326.
Techniques for measuring and methods of describing color and color
differences of imaged objects on color transparencies are explained. The
measurement of the transmittance of imaged objects is done on a Leitz MPV
microscope photometer equipped with an in-line monochrometer. From these
measurements the internal transmittance of the three dye layers is
calculated at 10 nanometer intervals from 380 to 720 nanometers. CIE
(Commission Internationale de 1"Eclairage) approved methods are used to
define the Imaged object's color by three approved systems and the paired
color differences between all objects are calculated. A multiple comparison
test yielded predictions that are correct for 119 out of 120 pairs oft
comparisons possible for the 16 objects studied on the 1/3,000 scale
infrared transparencies.
Hardy, Rolland L., and J. V. Taranik. 1973. Geodesy, Photogrammetry,
Photointerpretation and Remote Sensing—Getting It All Together in Earth
Mapping Education. In: Proceedings of the Am. Soc. of Photogrammetry
Symposium on the Management and Utilization of Remotely Sensed Data.
Sioux Fall, South Dakota, p. 501-511.
184
-------
Iowa State University and The University of Iowa now offer a coordinated
interdisciplinary program in imagery interpretation, photogrammetry,
geodesy, and remote sensing. This interdisciplinary program has evolved
because of the rapid development of space technology and sensor
instrumentation. There is a growing need for engineers and scientists to
have a broad background in earth mapping technology. Photogrammetrists now
use Imagery from multispectral scanners and video imagers mounted in
spacecraft. Photo interpreters must now evaluate imagery from radar imagers
and line scanners, and to understand the taxonomic characteristics of
imagery and to properly evaluate imagery artifacts, they must understand the
spectroradiometric characteristics of the energy path from source to
detector. Scientists engaged in remote sensing programs must understand how
to mensurate and interprete Imagery, as well as design remote sensing and
analysis procedures. Educators at Iowa State University and The University
of Iowa have recently joined forces to offer a combined course of study for
students with interests in mapping the earth from aircraft and spacecraft.
Prior to the new program, photogrammetry and geodesy were taught in the
Civil Engineering Department at Iowa State University, remote sensing in the
Geology Department at The University of Iowa, and photointerpretation in
different departments at both universities. Often a student with interests
in earth mapping outside his major department would have difficulty taking
courses in other departments or other universities. Under the new program,
students and faculty are exchanged between universities and departments, and
those engaged in research can utilize facilities and equipment at either
university. The recent development of a state remote sensing laboratory,
which serves the needs of 24 state and federal agencies operating in Iowa,
provides Impetus for the interdisciplinary educational program because it
repeatedly brings together basic researchers from the academic community,
analysts of imagery and those applying the analyses to the practical
problems of natural resource management, land utilization planning, and
environmental control in Iowa.
Heller, R. C. 1978. Case Applications of Remote Sensing for Vegetation
Damage Assessment. Photo. Eng. 44:1159-1166.
The assessment of vegetation damage by remote sensing has reached a fairly
sophisticated level. This paper identifies the advantages, pitfalls,
current practical applications, and future possibilities of the use of
remote sensing for this purpose. Advantages include: (1) the use of many
parts of the electromagnetic spectrum; (2) the saving of time, money, and
manpower; (3) the ability to cover large areas; and (4) the use of
successive remote sensing surveys to follow damage trends. Some pitfalls
included: (1) the overselling of remote sensing techniques without adequate
quantitative data showing errors of estimate at pre-defined confidence
limits; (2) using very expensive remote sensing systems on a transitory
phenomenon; (3) the Inability of some Landsat users to recognize that
reflectant values are relative subject to atmospheric attenuation, and
amplified signals; (4) the poor design of Landsat wavebands for vegetation
damage assessment (a yellow-orange waveband, 0.58 to 0.62 m is needed); (5)
a need for better statistical techniques to check classification accuracies;
and (6) using color or color infrared films to obtain previsual detection of
coniferous tree damage. Current practical applications for assessing
185
-------
vegetation damage include: (1) visual observation techniques (sketch
mapping and strip recording); (2) color and color infrared (CIR) photography
(both large and very small scale) when properly matched with damage symptom,
host type, and atmospheric conditions; (3) multi-stage sampling; and
(4) risk rating systems using aerial photos to define factors such as
aspect, slope, elevation, and stand density that contribute to
susceptibility of vegetation to damaging agents. Future remote sensing
possibilities predicted are (1) increasing standardization of color and CIR
photography and greater use of small-scale CIR (1:32,000); (2) the
availability of lightweight, inexpensive radar and laser altimeters together
with better electronic guidance systems for repetitive flights; (3) faster
service for receipt of Landsat data products which will be geometrically
corrected and enhanced; (4) improved Landsat computer classified Images with
accuracy statements; (5) better resolution available on Landsat D (thematic
mapper) with narrower wavebands which should Improve class ificatory
procedures and accuracies; and (6) improvements in other sensors such as
side-looking radar, charge coupled detectors, and microwave imagers.
Hoffer, R. M., P. E. Anuta, and T. L. Phillips. 1972. ADP Multiband and
Multiemulsion Digitized Photos. Photo. Eng. 38(10):989-1001.
Automatic data processing (ADP) techniques using a digital computer for data
handling and analysis have allowed quantitative examination of aerial
photography. Scanning microdensitometer techniques were utilized to
digitize both multiband and multiemulsion photography. These digital
density data from 1:120,000-scale aerial photos were spatially registered by
computer and then analyzed, using statistical pattern recognition
algorithms. The feasibility for automatic recognition of several cover
types is indicated. Similar results were obtained from the digitized
multiband and multiemulsion photographic data.
Hornung, R. J., and J. A. Smith. 1973. Application of Fourier analysis to
multispectral/spatial recognition. In: Proceedings of the Am. Soc. of
Photogrammetry Symposium on Management and Utilization of Remotely
Sensed Data. Sioux Falls, South Dakota, p. 268-283.
One approach for investigating spectral response from materials is to
consider spatial features of the response. This might be accomplished by
considering the Fourier spectrum of the spatial response. The Fourier
Transform may be used in a one-dimensional to multi-dimensional analysis of
more than one channel of data. The two-dimensional transform represents the
Fraunhofer diffraction pattern of the image in optics and has certain
invariant features. Physically the diffraction pattern contains spatial
features which are possibly unique to a given configuration or
classification type. Different sampling strategies may be used to either
enhance geometrical differences or extract additional features.
Hostrop, Bernard W., and T. Kawaguche. 1971. Aerial Color in Forestry.
Photo. Eng. 37:555-563.
The most significant Improvement in photogrammetry in recent years has been
the introduction of aerial color photographs. At the same time, electronic
printing of aerial negatives had enhanced the information content in
186
-------
reproducing both black-and-white and color materials. Color photographs
play an important role in the development of areas for new timber harvest
methods, soil and watershed studies, development of range and reforestation
area, planning land development, and streambed and fish habitat studies.
Color provides a signficant increase in accuracy of species identification,
identification of dead trees, and facilitates the identification of property
corners. Image-point selection and identification are improved by as much
as 60 percent.
Howard, John A. 1970. Aerial-Photo Ecology. American Elsevier Publishing
Co., New York. 325 p.
The reader is first introduced to the most important physical aspects of
aerial photography, including the reflection of light; and is then given an
adequate background of photogrammetry before proceeding to the general
principles of photo-interpretation of the natural environment. Topics are
discussed which will interest the agriculturalist, archaeologist, ecologist,
entomologist, forester, game manager, geographer, geologist,
geomorphologist, pathologist, physiologist, soil scientist and zoologist.
The extensive bibliography from the nineteenth century until the end of the
international congress on aerial photography, photogrammetry and
photointerpretation in 1968 will be useful to workers in almost all fields
where aerial photographs are used.
Hoyer, B. E., G. R. Hallberg, and J. V. Taranik. 1973. Seasonal
Multispectral Flood Inundation Happing in Iowa. In: Proceedings of the
Am. Soc. of Photogrammetry Symposium on the Management and Utilization
of Remotely Sensed Data. Sioux Falls, South Dakota, p. 130-141.
Evaluation of multispectral imagery from three floods occurring at different
times of the year in Iowa has indicated methods of mapping different times
of the year in Iowa has indicated methods of mapping flood inundation
several days after flood waters have returned to the main river channel.
Cooperative study by the Iowa Geological Survey, Remote Sensing Laboratory
and the U.S. Geological Survey, Water Resources Division, on flooding in
three seasons, suggests that color infrared film would provide flood
inundation data having the highest multiplicity of possible uses for
floodplain management-planning in Iowa. Characteristics of infrared
radiation, including the adsorption of photograhic infrared wavelengths by
water, the reduced infrared reflectance of wet soils and stressed plant
species, and the different reflective properties of snow and ice at infrared
wavelengths, account for this film's wide application to midwestern flood
mapping. Winter floods may be mapped by identifying ice remaining after
flood recession. Most photographic imagery appears adequate for mapping
winter floods, but color infrared Imagery appears somewhat superior.
Evaluation of imagery from mid-spring floods indicates that significant
flooding may be mapped for at least five days following flood recession.
Conventional photographic imagery is adequate for interpretation in bare
fields, but flood inundation of immature crops, pasture, or forest is most
adequately interpreted on infrared imagery. Late summer floods may be
mapped for at least seven days following flood crest using color infrared
imagery. Best flood inundation mapping was accomplished by multispectral
color-additive viewing utilizing the blue and infrared bands. ERTS-1
187
-------
satellite data has supported some of the basic conclusions of the low-
altitude studies. The satellite imagery also allowed the rapid appraisal of
the area! extent of flood inundation on a regional scale.
Humiston, Homer A., and G. E. Trisdale. 1973. A Peripheral Change
Detection Process. In: Proceedings of the Am. Soc. of Photogrammetry
Symposium on the Management and Utilization of Remotely Sensed Data.
Sioux Falls, South Dakota, p. 413-426.
A "Quick-Look." change detection function is discussed. The method
incorporates an on-line image registration capability which is not only
independent of relative orientation and scale but also eliminates the need
for identifiable control points within the images. The presentation will
include experimental results produced by a pilot model system. Television
input and output functions for this system incorporate self-calibrations,
enabling precision control of scan linearities as well as photometry.
Hyde, R. F., S. W. Bowand, and P. W. Mausel. 1977. ISURSL Levels
Classification: A Low Cost Approach to Multispectral Data Analysis.
In: Proceedings of the Symposium on Machine Processing of Remotely
Sensed Data. LARS, Lafayette, Indiana, p. 322-331.
The Indiana State University Remote Sensing Laboratory (ISURSL) recognized
that the promise of low-cost earth resource evaluation through machine-
assisted processing of multispectral (MS) data has not been fully
realized. In response to this problem the ISURSL has developed a complete
low-cost system of processing MS data which minimized analysis time for both
man and computer while simultaneously maximizing utilization of the data.
The basis of the ISURSL classification algorithm, designated LEVELS
CLASSIFIER, is identification of numeric boundaries located in
multidimensional feature space which differentiate features of interest.
Land use classes of interest to an analyst are described by the range of
radiance (relative spectral response) levels which define these
boundaries. The identification of levels boundaries which accurately
defines an earth surface feature is accomplished through sophisticated
single and multidimensional histogram terrain analysis. This approach to
multispectral data analysis has been shown to be cost effective and accurate
in several applied research projects at ISURSL.
Kalensky, L., and D. A. Wilson. 1975. Spectral Signatures of Forest Trees.
In: Proceedings of the Third Canadian Symposium of Remote Sensing.
Edmonton, Alberta, p. 190-205.
Described are the field measurements of daylight radiation reflected upwards
from the crowns of six tree species in visible and near-infrared frequencies
of the electromagnetic spectrum. A portable mast and two permanent towers
provided platforms for a spectroradiometer at a height of 3 to 5 m above the
tree canopy. Each site was measured on at least two different dates between
late June and September to account for variations in species phenological
stages during the summer season. In addition, some of the species were
measured in two different locations to account for differences in site
conditions. Described are the instruments used for the measurement of
incident and reflected daylight radiation, the field measurement technique
188
-------
and computational procedure. Presented are the reflectance data of tree
species and their variations calculated from the field spectroradiometric
data measured in the 1974 season. Their relevance for multispectral remote
sensing and image interpretation is discussed.
Kasvoud, T. 1972. Can We Teach Computers to See? In: First Canadian
Symposium on Remote Sensing. Ottawa, Ontario, p. 551-667.
The number of pictures obtained by satellites, from aerial photography, from
bubble chambers, microscopes and so on is nearly unlimited. Ideally we
would like this picture data to be analysed by computers. On specific but
limited problems fair success has been achieved. Thus, bubble chamber
pictures are processed in large numbers, several types of micro-biological
objects are recognized by computers, attempts have been made to classify
fingerprints, to recognize objects on aerial photographs, etc. By pooling
the experience gained from successes as well as failures, the author tries
to outline the requirements for a fairly general pattern recognition system
which can be taught to recognize given objects in the pictures.
Kau, E. P., D. L. Ball, J. P. Basu, and R. L. Smelser. 1975. Data
Resolution Versus Forest Classification and Modeling. In: Proceedings
of the Symposium Machine Processing of Remotely Sensed Data. LARS, West
Lafayette, Indiana, p. 1B-24—1B-36.
This paper examines the effects on timber stand computer classification
accuracies caused by changes in the resolution of remotely sensed
multispectral data. This Investigation is valuable, especially for
determining optimal sensor and platform designs. Theoretical justification
and experimental verification support the finding that classification
accuracies for low resolution data could be better than the accuracies for
data with higher resolution. The increase in accuracy is construed as due
to the reduction of scene inhomogeneity at lower resolution. The computer
classification scheme was a maximum likelihood classifier.
Kie, Soon T., and 0. J. Lewis. 1974. Optical Film Density Values From
Color IR Photography for Wetland Soils Mapping. In: Proceedings of the
1974 Fall Meeting of the Am. Soc. of Photogrammetry. Washington, D.C.
p. 323-331.
An attempt is made to analyze characteristics of optical film density from
small scale color IR film for the purpose of soils mapping. Four hundred
and one (401) sample spots were selected, and transmittance diffuse density
values were measured for four filters using a Macbeth Transmission
Densitometer. A CRD analysis of variance, a factor analysis and two
different cluster analyses were conducted. The results indicate that 1) the
three known soil groups cannot be discriminated using the cluster analyses,
2) only one common factor is retained by the factor analysis and it
contributes approximately 85 percent to the total communality, and
3) classification of soil groups may be feasible by an adjusted computer
mapping procedure.
139
-------
Kiefer, Ralph W. 1970. Effects of Date of Photography on Airphoto
Interpretation Using Color and Color-Infrared Films. In: Proceedings
of the International Symposium on Photography and Navigation.
p. 100-117.
More than 3000 exposures of color and color-infrared film on 35 mm format
were taken of selected sites in Southern Wisconsin during 1969 from
elevations of 2000 to 8000 feet above terrain. The subject matter includes
rural terrain (cropland, grazing land, and woodland), lakes (showing weed
and algae growth), and river flood plains (showing river flooding and
subsequent crop damage). Certain intensive study sites were photographed on
20 different dates during the year. The striking changes in week-to-week at
the intensive study sites are illustrated in this paper. Relevant ground-
truth data of selected sites are included. The results of this research
show that there are certain optimum dates during the year for the
procurement of aerial photography for interpretive uses. The results also
show that the optimum date of photography may not be the same for different
interpretive uses.
Kiefer, Ralph W. 1969. Airphoto Interpretation of Flood Plain Soils. J.
of the Surveying and Mapping Division. 93:119-139.
The use of. aerial photographic interpretation techniques to estimate flood
plain soil conditions has been described herein. Problem areas requiring
futher study include: (1) The use of airphoto interpretation to aid in
estimating flood frequency; and (2) the effects of the date of photography
on the airphoto appearance of flood plain soils. New techniques of airborne
remote sensing can be useful for flood plain soil studies. Imaging sensors
such as infrared photography, color photography, infrared Imagery, and side-
looking airborne radar, as well as a number of nonimaging sensors, may prove
very useful. In order to plan successfully for land use on flood plains,
all available planning tools must be utilized. Planners and developers
should be aware that much valuable information about flood plain soils can
be obtained through aerial photographic interpretation studies.
Kirby, C L., D. Goodenough, D. Day, and P. Van Eck. 1975. Landsat Imagery
for Banff and Jasper National Parks Inventory and Management. In:
Proceedings of the 3rd Canadian Symposium on Remote Sensing. Edmonton,
Alberta, p. 107-225.
Computer assisted classification of Lands at digital magnetic tapes of Banff
and Jasper National Parks were done using the General Electric "Image-100"
of the Canada Centre for Remote Sensing in Ottawa. Themes of pine, spruce
and popular-shrub forest, water, snow and meadows were classified by their
spectral signatures. From 70 to 80 percent of the four areas studied were
classified with 80 to 90 percent accuracy using a supervised parallelepiped
classification. Small training areas (50-100 km ) in each Lands at image
were classified using this method as well as 1200 km areas on two Landsat
Images classified successfully at full resolution. The classifications
produced were geometrically correct in color at a scale of 1:250,000 on an
electron beam Image inventory and in a National Park public education
program. A limited number of Landsat photo-maps of Banff and Jasper Parks
at a scale of 1:500,000 on color, and with national topographic map
190
-------
information on water resources and transportation, are available on request
to the Northern Forest Research Centre.
Knipling, E. G. 1970. Physical and Physiological Basis for the Reflectance
of Visible and Near Infrared Radiation from Vegetation. Rem. Sens.
Environ, p. 155-159.
Knowledge of how solar radiation interacts with vegetation is necessary to
interpret and process remote sensing data of agricultural and many natural
resources. A plant leaf typically has a low reflectance in the visible
spectral region because of strong adsorption by chlorophylls, a relatively
high reflectance in the near-infrared because of internal leaf scattering
and no adsorption, and a relatively low reflectance in the infrared beyond
1.3 because of strong, adsorption by water. The reflectance of a plant
canopy is similar, but is modified by the nonuniformity of incident solar
radiation, plant structures, leaf areas, shadows, and background
reflectivities. Airborne sensors receive an integrated view of all these
effects and each crop or vegetation type tends to have a characteristic
signature which permits its discrimination. When disease and physiological
stresses directly affect the reflectance properties of individual leaves,
the most pronounced initial changes often occur in the visible spectral
region rather than in the infrared because of the sensitivity of chlorophyll
to physiological disturbances. The primary basis for the detection of
stress conditions in a crop of other plan community by aerial remote sensors
often, however, is not a change in the reflectance characteristics of
individual leaves, but a reduction in the total leaf area exposed to the
sensors. This reduction can result from a direct loss of leaves, a change
in their orientation, or an overall suppression of plant growth. In such
cases the total infrared reflectance tends to be decreased relatively more
than the visible reflectance because of a reduction in the infrared
enhancement due to fewer multiple leaf layers and because of an increase in
background exposure.
Krumpe, P. P., H. R. Deselm, and C. C. Anderson. 1971. An Ecological
Analysis of Forest Landscape Parameters by Multiband Remote Sensing.
In: Proceedings of the Seventh Symposium on the Environment. Ann
Arbor, Michigan, p. 715-130.
This study, part of multidisciplinary research under contract with the U.S.
Department of Defense through Project THEMIS, tests remote sensor
utilization and application in environmental studies. The purpose is to
provide a means for predicting distributional and statistical parameters of
vegetation in areas devoid of ground control. Ground truth studies used
one-fifth acre targeted plot areas and point samples in which Southern
Appalachian landscape attributes were quantitatively measured, relatively
scaled or qualitatively classed. These included forest tree crown size,
stand canopy closure, tree density and size, phenology, and other
characteristics by species. Autumn and winter-flow 70 mm imagery obtained
in different spectral bands at large and small photographic scales, were
examined to assess the limits of this method in ground truth verification.
Differing vegetation types characterized by nearly two dozen canopy size
Deciduous Forest tree species, necessitated preparation of a dichotomous key
to aid in species identification and the separation of community types. The
191
-------
key was based on stereoscopic examination of a 2000 tree sample, among which
33 crown foliage characteristics were documented and many designated Munsell
color classes differentiated. Comparative studies between ground truth and
film acquired data yielded predictive regression equations expressing their
interaction. The validity of visual vegetation type mapping was tested by
comparison with comprehensive ground control data. Preliminary use of the
Tech/Ops Scandig Model 25 microdensitometer indicated the potential
reliability of extending visual-manual species recognition techniques to
semi-automated inventorying and mapping of Southern Appalachian natural area
forest resources.
Lewis, Anthony J., and H. C. MacDonald. 1973. Radar Geomorphology of Coast
and Wetland Environments. In: Proceedings of the Am. Soc. of
Photogrammetry Fall Convention. Lake Buena Vista, Florida.
p. 992-1003.
jJide Looking Airborne jladar (SLAR) Imaging systems are of special interest
to the coastal and wetland geomorphologist. Continuous strip presentation
of the land-water interface of at least 16 kilometers wide and hundreds of
miles long is advantageous for the study of the relatively narrow coastal
zone. In addition, the near all-weather, 24-h imaging capability is a
particular asset in coastal and wetland environs commonly obscured by cloud
cover. A variety of coastal environments have been imaged with commercial
radar mapping systems during the past 10 yr. Some of these coastal areas
include the Arctic Coast of Alaska, the Gulf Coast of Louisiana and Texas,
the California and Oregon Coasts, Chesapeake Bay, and the Atlantic and
Pacific Coasts of Central and South America. The wetland environment of the
Atchafalaya Basin and the coastal swamp and marsh region in Louisiana have
also been imaged. This study summarized the past work in radar coastal
morphology by the authors and their co-workers but primarily focuses on
recently complete research in the wetland environment of Louisiana and the
coastal environment of Oregon.
Mairs, Robert L., and Denis K. Clark. 1973. Remote Sensing of Cstuarine
Circulation Dynamics. 39:929-938.
Multispectral and color aerial photography and infrared imagery of naturally
occurring water color boundaries and/or dye tracer implants have been used
successfully in the study of temporal coastal and estuarine circulation
dynamics. Sequential photography and high-contrast enhancements of color
imagery of fronts such as foam lines, current shears, etc., along with point
and line sources of fluorescent dye are used to calculate and plot
displacements and velocity vectors of water masses along the North Carolina
coast and in the Patuxent River estuary, Maryland. Techniques have been
developed for incorporation of remotely sensed data which are collected on a
temporal scale ranging from minutes to hours, with extensive surface truth
measurements to describe further the complex nature of estuarine flow.
Malhotra, R. C., and M. C. Rader. 1975. Locating Remotely Sensed Data on
the Ground. In: Remote Sensing Energy Related Studies, T. Nejat
Veziroglu, ed. Hemisphere Publishing Co. Seattle, Washington.
p. 432-436.
192
-------
This paper briefly discusses techniques for identifying the precise ground
location represented by a specific set of remotely sensed data. Automated
mathematical procedures using navigation (ephemeris) data and/or ground
control points to identify the ground location of remotely sensed data sets
by parametric modeling, "speculative" polynominal adjustment, and a method
combining modeling and polynomial adjustment are described. Data from the
NASA Skylab and the Earth Resources Technology Satellite (ERTS-1) projects
are used in the examples given, but the basic methods apply to remotely
sensed data collected from any aircraft or Earth-orbital satellite.
Markham, B. L., W. R. Philipson, and A. E. Russell. 1977. Airphoto
assessment of changes in aquatic vegetation. In: Proceedings of the
43rd Annual Meeting of the Am. Soc. of Photogrammetry. Washington, D.C.
p. 504-516.
Large scale, multi-year, color and color infrared aerial photographs were
used to evaluate changes in aquatic vegetation that have accompanied a
reduction in phosphorus inputs to a phosphorus-limited, eutrophic lake in
New York State. The study showed that the distribution of emergent,
floating and submersed vegetation could be determined with little or no
concurrent ground data; that various emergent and floating types could be
separated and, with limited field checks, identified; and that different
submersed types are generally not separable. Major vegetation types are
characterized by spectral and non-spectral features, and a classification is
developed for compiling time-sequential vegetation maps.
Marshall, J. R., and M. P. Meyer. 1978. Field Evaluation of Small-Scale
Forest Resource Aerial Photography. Photo. Eng. 44(l):37-42.
Economic considerations prevent most forest land managers from obtaining
conventional black-and-white medium-scale (circa 1:15,000-1:20,000) forest
aerial photography at adequate intervals. Were smaller-scale photos
comparably useful, the savings in procurement and interpretation costs could
be used for more frequent overflights. Forested portions of Minnesota were
flown with black-and-white infrared at scales of 1:15,840; 1:24,000; and
1:31,680 and with color infrared at scales of 1:1,680 and 1:80,000. Trained
cooperators who analyzed the photographs under field-use conditions with
high quality viewing equipment considered black-and-white forest photography
at a scale of 1:24,000 marginally acceptable at best, and judged scales
smaller than 1:24,000 unacceptable for the resource management applications
involved. Overall, good quality summer black-and-white infrared 1:51,840
scale photography was preferred, but many user-cooperators were enthused
about the potential of small-scale coverage as a supplement to, not a
replacement for, conventional medium-scale photography. Color infrared
transparencies provided more information than black-and-white prints of
equivalent scale, but were considered overly cumbersone for day-to-day use
under existing field office conditions.
McDowell, David Q. 1973. Determination of Spectral Reflectance Using
Aerial Photography. In: Proceedings of the Am. Soc. of Photogrammetry
Fall Convention, Lake Buena Vista, Florida, p. 408-423.
193
-------
Many groups Involved in remote sensing are concerned with the spectral
reflectance of specific terrain elements. In some applications the spectral
differences detectable visually with color film are adequate; for others a
more detailed knowledge of the spectral reflectance curve is required. A
technique has been proposed, using characteristic vector analysis, whereby a
relationship can be developed between the spectral reflectance of a limited
class of objects and the densities that these objects produce on a color
film. This report describes a field test program conducted to evaluate the
feasibility of this technique using low-altitude aerial photography of
agricultural crops typical of upstate New York and in situ measurements of
spectral reflectance. Reflectance curves were predicted over the 400- to
690-nm region with an average standard error of 0.0061 for samples of five
different field crops that comprised the class of objects studied.
Reflectance curves for two additional agricultural crops outside the basic
group studied were also predicted with somewhat larger standard error.
McLaurin, John D. 1975. Information System for Aerial Photographers.
In: Proceedings of the 41st Annual Meeting of the Am. Soc. of
Photogrammetry. Washington, D.C. p. 154-161.
The National Cartographic Information Center has designed a summary record
information system for aerial photographs. The system data base contains
information on the geogrpahic extent of coverage and general characteristics
of aerial photographs; first emphasis is on coverage at scales of 1:40,000
and smaller. In addition to information on current holdings the data base
also contains information on planned photo acquisition. Participating
agencies will regularly provide NCIC with the information in digital form
for direct input to the data base. Each summary record includes the agency
name, date of coverage, photo scale, film type, extent of coverage to the
nearest 7.5-min quadrangel or by state and county, agency project code, and
status (planned, in progress, or complete). One can search the data base
using any of these parameters and obtain a computer-generated graphic
indicating the extent of available coverage. Work is underway to expand the
interactive capabilities of the system to provide different types of
output. For the first time aerial photography users can readily keep
informed on photo holdings and acquisition plans. The system's value
depends directly on the agencies providing the essential input data. Thus
NCIC is continually negotiating with organizations that collect aerial
photographs, with the aim of providing complete information.
Mo11ay, Martin W., and V. V. Salomonson. 1973. Remote sensing and water
resources U. S. space program. In: Remote Sensing and Water Resources
Management Proceedings. No. 17. p. 6-38.
Since the launch of TIROS I in 1960, the utility of remote sensing from
orbit for monitoring the earth's weather has been conclusively
demonstrated. The past decade has also seen progress in applying remote
sensing to the observation of terrestrial features. As in meteorology,
networks of ground instrumentation are essential; their data may be relayed
by satellite for calibration of orbital measurements. Variations in snow
and ice cover, surface water, river and lake turbidity, and other
hydrological features are now being accurately observed from orbital
altitudes by the Earth Resources Technology Satellite (ERTS-1), NQAA-2, and
194
-------
Nimbus 5. Satellite visible, infrared and microwave measurements will be
continued over the next few years—with improved spatial and spectral
accuracy—by Skylab, ERTS-B and Nimbus F. Delineations of soil and snow
moisture variations, thermal patterns in lakes and estuaries, and regions of
heavy precipitation are among the results anticipated. Operational earth
resources survey programs are expected to evolve within the user agencies of
the United States Government. Route repetitive, quantitative observations
over watersheds as large as the Mississippi River are ultimately
anticipated. These developments will provide input to regional and global
numerical models that better predict, define, and manage the components of
the hydrological cycle.
Holland, D. 1975. An Integration of Different Aerial Remote Sensors and
Map Data in Making Engineering and Resource Studies. In: Proceedings
of the Third Canadian Symposium of Remote Sensing. Edmonton, Alberta.
p. 413-419.
Over 1,800 aerial remote sensing studies of environmental and resource
projects have been reviewed. All studies, covering the period 1955 to 1975,
involved the stereoscopic interpretation of panchromatic airphotds, mostly
at the scale range 1:80,000 to 1:10,000. Airphoto mosaics scaled
approximately 1:125,000 and contact prints scaled approximately 1:63,360
were used on all major hydroelectric and transportation route studies.
Interpretations of several kinds of available maps and aerial and ground
reconnaissance were carried out on the larger projects. Since 1972 Landsat-
1 imagery has been used selectively on large corridor-type projects. Part
of the review entailed identification of projects where the interpretation
of Lands at-1 imagery either was or would have been signficantly helpful,
marginally helpful, of questionable assistance, and of no assistance. The
number and type of more common remote sensing studies carried out over 20 yr
are listed. The types and scales of remote sensors and their integration
with map and field work for a typical northern route corridor study are
described. In addition, ways of increasing remote sensing expertise,
especially for beginners, are briefly discussed.
Murtha, P. A. 1978. Symposium on Remote Sensing for Vegetation Damage
Assessment. Phot. Eng. and Rem. Sens. 44:1139-1145.
A Symposium on Remote Sensing for Vegetation Damage Assessment was held in
Seattle, Washington, U.S.A., on February 14, 15, and 16, 1978. Four
invited and 27 presented papers were delivered during the Symposium. The
papers dealt with (1) the theory of vegetation damage detection and
assessment, (ii) the technologies involved, (ill) case studies, and (iv)
economics and current applications. Resolutions were called for and
submitted during the Symposium. The resolutions reflected the moods,
present needs, and future concerns of the scientist and managers at the
meeting. The resolutions asked for (i) ASP and ISP-Comm. VII support and
encouragement of research into vegetative dysfunction relative to remote
sensing; (ii) an international study on "previsual" or extravisual damage
detection; (iii) more precise definition of "damage" and damage classes;
(iv) coding of forest damage types in chronic vegetation damage situations;
(v) quality control through use of defined confidence levels and statements
of errors of estimates; and (vi) more effective technology information
195
-------
transfer at symposia, and at government or institutionally sponsored local
area workshops.
Murtha, P. A. 1978. Remote sensing and vegetation damage: a theory for
deflection and assessment. Photo. Eng. 44:1147-1158.
This paper discusses the philosophical and technical aspects of remote
sensing for vegetation damage assessment. Answers are presented for these
questions: (1) What constitutes remote sensing evidence of vegetation
damage? (2) How is vegetation damage interpreted from remote sensed data?
and (3) How can the damage be assessed? The answers to these questions are
discussed in detail relevant to normal color and color-infrared aerial
photography. Consideration is given to details of film reaction to
variations in spectral reflectance patterns. Damages showing morphological
or physiological changes are discussed relative to spectral reflectance
changes and presented as a means to code damage types. An hypothesis for
quantitatively monitoring forest damage is presented.
Nielson, U. 1972. Effects of Spectral Filtration and Atmospheric
Conditions on Aerial Photography Obtained in 1970 and 1971. In:
Proceedings, 1st Canadian Symposium on Remote Sensing. Ottawa, Ontario.
p. 411-416.
Photography obtained by the Airborne Sensing Unit provided material which
illustrates many of the basic concepts of spectral reflectance, atmospheric
attenuation and spectral filtration discussed in the paper. These concepts
have to be understood and the presently very limited amount of information
concerning spectral reflectance and effect of atmospheric factor has to be
vastly increased if specifications for aerial photography are to be
optimized.
Newhally, N. R., and R. E. Witraer. 1970. Remote Sensing for Land-Use
Studies. Photo. Eng. 36:449-453.
The most common problems associated with interpreting land-use data are
(1) incompatible and inconsistent use of terminology, and (2) developing
useful and comparable classification systems. The interpretation and
classification system proposed and tested here has two basic parts: land-
use interpretation in as great detail as possible; and devising a
classification system using the interpreted data which is specifically
suited to the problem at hand. Sixteen photo interpretors participated in
an experiment to test the validity and utility of the proposed system. They
were divided into a control group which used any interpretation and
classification system, and an experimental group which used the proposed
system described in this report. Preliminary sampling analysis of these
interpretations indicate that the members of the experimental group had the
most detailed interpretations, produced more specific land-use data with
less ambiguity, had fewer non-use classes, and employed more compatible
classification systems.
196
-------
Olson, C. E., and R. E. Good. 1961. Seasonal Changes In Light Reflectance
from Forest Vegetation. Photo. Eng. 27:492-493.
During the 1960 growing season light reflectance from foliage of nine tree
species was mesured weekly with a General Electric recording
spectrophotometer for the wave-length range from 400 to 700 millimicrons.
Sampling was begun in early May and continued through the fall color change
terminating in November. Four replications were obtained for each species
in each sampling period and all foliage sampled was taken from the south
side of the upper quarter of the tree crown. All samples were picked
between 10 a.m. and 2 p.m. local standard time, and reflectance measurements
were completed within one hour of the time the foliage was picked. It was
found that hardwood foliage reflected more light than pine foliage in almost
all wave-lengths during all parts of the growing season. Differences in
reflectance between hardwood and pine foliage decreased steadily from May to
the beginning of the fall color change in hardwoods. Date of initiation of
the fall color change varied with species but, in all hardwood species, the
color change was characterized by increasing reflectance. When percent
reflectance at 550 millimicrons was plotted over date, a seasonal pattern of
changing reflectance seemed apparent in all hardwood species. Reflectance
decreased during the early weeks of the growing season, remained nearly
constant until mid- or late-summer, and then rose rapidly during the fall
color change. A distinct, but short-lived, decrease in reflectance occurred
in all hardwood species several weeks after the beginning of the fall color
change. This pattern was also apparent when percent reflectance was plotted
over data for 600 millimicrons.
Owen-Jones, E. S. 1977. Densitometer Methods of Processing Remote Sensing
Data with Special Reference to Crop Type and Terrain Studies. In:
Environmental Remote Sensing, E. C. Barrett and L. F. Curtis, eds.
Crane and Russak, New York. p. 101-124.
The film portion affecting densitometrie measurements are discussed in
relation to general requirements for satisfactory results. Next, the
principles of densitometry are reviewed, including a discussion of flat-bed,
rotating-drura and flying-spot scanners. Applications of classification
techniques to agricultural and natural terrain areas are also discussed.
These include crop-type and terrain analysis from false colour photography
obtained by the Skylark sounding rocket over Argentina and aircraft surveys
in Australia. It is concluded that both supervised and unsupervised
classification methods have their own particular merits. At the present
time the state of the art does not preclude an element of subjective
judgement.
Pakrekak, A. J., U. Sawka, and R. K.. Schmidt. 1974. Analysis of Nesting
Habitat of Canada Geese Using Remote Sensing Imagery. In: Second
Canadian Symposium on Remote Sensing. Guelph, Ontario, p. 336-371.
The purpose of this study was to evaluate nesting habitat of Canada geese
(Branta can ad ens is interior) in the Little Seal River Area of Manitoba using
recent remote sensing imagery. Five different sets of imagery, all taken in
August 1972, were carefully examined to determine which films best
represented vegetation and landform features. Two types of color infrared
197
-------
photography proved to be most suitable and were used In delineating
vegetation-landform-(habitat) units on a study area map. These units were
then compared with goose nesting data collected in spring, 1970. Canada
geese nested in 6 of the 8 designated units but showed a marked preference
for birch-willow and gravel ridge habitats. In general, results suggested
that remote sensing imagery could be used to describe habitats in other,
largely inaccessible goose breeding areas. Such an approach, if applied on
a large enough scale, would provide new and relatively inexpensive ways of
estimating annual production of Canada geese in the Eastern Prairie
Population.
Polcyn, Fabian C., and D. R. Lysanga. 1973. Multispectral Sensing of Water
Parameters. In: Remote Sensing and Water Resources Mngr. Proceedings
No. 17. p. 394-403.
With the development of the multispectral scanner, Improved techniques for
mapping temperature gradients, turbidity, water color, and alga
concentrations over large areas have been demonstrated. Where lake water
transparency is sufficiently clear to detect light reflections from the lake
floor a remote calculation of water depth is possible. Depth to 20 ft has
been measured in the nearshore zone of Lake Michigan and near the Little
Bahama Bank. Maps showing relative chlorophyll concentrations have been
made for a portion of the shoreline areas near Port Sheldon, Michigan.
Examples will be shown for the mapping of the terminal bar in Lake Michigan,
river discharges, and the nearshore environment. Spectral characteristics
related to chlorophyll concentrations were investigated for test samples
across the thermal bar taken during the spring formation of the bar.
Rohde, W. G., and G. E. Olson, Jr. 1972. Multispectral Sensing of Forest
Tree Species. Photo. Eng. 38:1209-1215.
Computer recognition of forest tree species at the NASA-Ann Arbor Forestry
Test Site has been accomplished using data collected in six spectral regions
between 0.4 and 1.0 micrometers. The six wavelength bands used were
selected on the basis of laboratory reflectance data previously collected by
the authors. Data obtained with The Unversity of Michigan C-47 aircraft and
processed with the University of Michigan Spectral Analysis and Recognition
Computer (SPARC), provided successful separation of coniferous and broad-
leaved trees. Specific recognition and separation of sugar maple, black
walnut, black locust, red oak, and white oak were also successful.
Discrimination among conifers was not so successful as for broad-leaved
species, but spruce were consistently separated from pine.
0
Sayn-Wittgenstein, L., and Z. Kalensky. 1974. Interpretation of Forest
Patterns on Computer Compatible Tapes. In: Proceedings of the 2nd
Canadian Symposium on Remote Sensing. Guelph, Ontario. Vol. 1:267-276.
The identification of spatial patterns should receive more emphasis in the
interpretation of ERTS imagery. Spatial patterns can be recognized by
involving such concepts and techniques as serial correlation, central
tendency, periodicity and spectral analysis. The best results are obtained
by using computer compatible tapes, rather than photographic Images.
198
-------
Scarpace, F. L. 1978a. Densitometry on Multi-Emulsion Imagery. Photo.
Eng. 44(10):1279-1292.
Basic concepts of color densitometry and film calibration procedures are
reviewed with special emphasis on the specific application to the remote
sensing investigator. The differences between spectral, broad band,
specular, diffuse, integral, and analytical densities are discussed and the
instrumentation necessary for each type of measurement is described. An
explanation of equivalent neutral density and methods of determining this
type of density are presented. Methodologies of using analytical densities
for the remote sensing community are detailed. The use of analytical
densities in the construction of characteristic curves is discussed. Also
included are comments made on reasons for using analytical densities in the
analysis of film Imagery and on proper application of the exposure values
derived from the characteristic curves.
Scarpace, F. L., and G. L. Friederichs. 1978b. A Method of Determining
Spectral Analytical Dye Densities. Photo. Eng. 44(10):1293-1301.
A straightforward method for the user of color Imagery to determine the
spectral analytical density of dyes present in the processed imagery is
presented. The method involves exposing a large number of different color
patches on the film which span the gamut of the film's imaging
capabilities. From integral spectral density measurements at 16 to 19
different wavelengths, the unit spectral dye curves for each of the three
dyes present were determined in two different types of color films. A
discussion of the use of these spectral dye densities to determine the
transformation between integral density measurements and analytical density
is presented.
Scarpace, F. L. 1977. Densitometry on Color and Color IR Imagery. In:
Proceedings of the 43rd Annual Meeting of the Am. Soc. of
Photogrammetry. Washington, D. C. p. 301-318.
Basic concepts of color densitometry and film calibration procedures are
reviewed with special emphasis on the specific application to the Remote
Sensing investigator. The differences between, and the instrumentation to
measure the spectral, broad band, specular, diffuse, integral and analytical
densities are discussed. An explanation of equivalent neutral density and
methods of determining this type of density are presented. Methodologies of
using analytical densities for the Remote Sensing community are detailed.
The use of analytical densities in the construction of characteristic curves
are discussed. Comments are made on reasons for the use of analytical
densities in the analysis of film imagery and on proper application of the
exposure values derived from the characteristic curves.
Scarpace, F. L., and P. R. Wolf. 1972. Convenient Atmospheric Refraction
Equations. University of Wisconsin-Madison Remote Sensing Program.
Madison, Wisconsin. 21 p.
Methods have been formulated for computing atmospheric refraction
corrections to be applied to measured plate coordinates. The corrections
may be calculated based upon any combination of flying height, ground
elevation, atmospheric pressure, atmospheric temperature, and measured plate
199
-------
coordinates. The mathematical derivation of the equations is simple and the
parameters used in the equations are readily measured. A theoretically
exact method is presented first, followed by other methods which are
approximate and simplify the calculations. Tests have shown that the
approximations consistently yield results that agree with the theoretically
exact values to within one micrometer.
Scherz, James P. 1974. Errors in Photogrammetry. Photo. Eng. 40:493-500.
To teach and work effectively with photogrammetry, one should have a basic
understanding of the sources and relative magnitude of errors inherent in
aerial photographs. Exact calculus approaches are often so complicated that
they cause one to want to forget about errors entirely and pretend they do
not exist. The approach described herein equates all source error effects
to a percentage, and as long as the mathematical manipulations are
multiplication and division, the same percentages can be applied to the
final answer to ascertain its probable error. The method described provides
estimates of errors identical to those obtained using calculus, but the
described method is much easier. The method provides students and users
with a ready and quick method both for analyzing errors and for obtaining a
feeling for the relative magnitudes of errors in photogrammetry work.
Scherz, J. P., and S. S. Ramchandra. 1973. A Practical Indexing and
Retrieval System for Remote Sensing Data. In: Preceedings of the Am.
Soc. of Photogrammetry Symposium on Management and Utilization of
Remotely Sensed Data. Sioux Falls, South Dakota, p. 528-538.
Remote sensing imagery and support data is very valuable for any school or
agency involved with mapping and environmental monitoring. However, this
remote sensing data is all but useless unless it can be found and retrieved
when needed. An interdisciplinary remote sensing library in which remote
sensing data is index, catalogued, and filed by methods not entirely foreign
to library methods used for books. The method of access in this program is
a card catalogue system which contains the necessary pertinent data and a
cross reference procedure which allows access to the data by users from
various fields of interest. The system has proven very practical and
usable, and can be adapted by any agency with a minimum of special
personnel, training, and cost.
Schulte, 0. W. 1951. The Use of Panchromatic, Infrared and Color Aerial
Photography in the Study of Plant Distribution. Photo. Eng. 17:688-714.
(1) It is dangerous to generalize from one region to another in the
selection of films. (2) Of the three films—panchromatic, infrared, and
color—used in recognizing plant species of southeastern Canada, infrared is
most preferable, and panchromatic and color about equal. (3) The infrared
caused some confusion in distinguishing among dark conifers, water, cloud
shadows, rock outcrops, and certain soils that print black. Ground details
are obscured in shadow areas. There is poor distinction between all low
herbaceous plants and dry or wet meadows and brushland. It is poor when the
density count of a stand is desired. However, these disadvantages are
outweighed by the superior results of infrared in delineating forest types,
especially conifers and hardwoods, and by its ability to penetrate haze.
200
-------
When such factors as size, location, association, tone, and texture are
known, even the conifers and hardwoods can be broken down into smaller
groups, or even species, more readily than on panchromatic. (4) Color
photography is the least used because of cost. It is preferable for special
problems, such as disease-control in summer, because it renders the leaves
which have turned brown or yellow very evident. Any other disturbance of
vegetation and consequent browning is rendered more visible on color
photography. For a limited area it is better than panchromatic or infrared
film for distinguishing hardwoods by color contrast in spring and autumn.
During the summer, it is inferior to infrared for species identification.
(5) Panchromatic is generally poorer in the east because of its failure to
make the major distinction between conifers and hardwoods in the summer
condition. Regardless of filters, the tones are often unreliable, and the
textures not sufficiently differentiated to rate it over infrared.
Panchromatic gives better detail for low shrubby areas, grasses, rocks,
soils, roads, paths, houses, etc. The density of a forest can be estimated
better on panchromatic. In the tundra of the North and the grasslands of
the Plains where the vegetation is all shrubby (more or less), panchromatic
is better than infrared since it produces more tonal differences and
detail. (6) The film-filter-scale combination is a local problem and is
dependent on the nature of the vegetation and the specific objectives
desired. (7) In regard to keys, it is almost necessary to make one for each
forest condition, and modified more or less for each flight. (8) The
nearest approach to a practical key for recognizing species is a complete
file for every recognizable species with stereoscopic ground and aerial
photographs for all possible conditions. (9) The techniques of using
hetero-stereo3copic pairs affords no real advantage over any single type.
However, when all three types of film are available for a given area, then
what is lacking on one film, can possibly be obtained on the other.
Slater, P. N. 1975. Basic Differences in the Quality of Analog and Digital
Imagery from Photographic and Solid-State Array Remote Sensing
Systems. In: Proceedings of the Am. Soc. of Photogrammetry Fall
Convention. Phoenix, Arizona, p. 139-153.
An analytical study has been made to compare the imagery from a solid-state
array camera and a photographic film camera operating under the same
conditions. The two cameras were chosen to be of the same size and to yield
digital imagery of the same effective instantaneous field of view (EIFOV).
The comparison covered both digital Imagery, in which the film was scanned
by a microdensitometer, and analog imagery, in which the output from the
array was recorded on film. The digital imagery was evaluated with regard
to signal-to-noise ratio (SNR) and minimum detectable ground reflectance
difference (Ap); the analog Imagery was evaluated with regard to visual
resolution limit. The effects on the Imagery of various atmospheric
conditions and of different ground scene contrast ratios were also
investigated. The most interesting result of the study is that the digital
imagery from the solid-state array camera has a higher SNR than that from
the film earnera-microdensitometer system. Thus the array camera imagery is
preferred for automated scene classification purposes. On the other hand,
the visual resolution limit is at a higher spatial frequency for the imagery
from the film camera than that from the array camera-film recorded system.
201
-------
Thus the film camera la preferred for cartographic and mensuration
applications.
Souto-Maior, Joel. 1973. Applications of Thermal Remote Sensing to
Detailed Ground Water Studies. In: Remote Sensing and Water Resources
Mngt. Proc. No. 17. p. 238-298.
Three possible applications of thermal (8-14 microns) remote sensing to
detailed hydrogeologic studies are discussed in this paper: 1) the direct
detection of seeps and springs, 2) the indirect evaluation of shallow ground
water flow through its thermal effects on the land surface, and 3) the
indirect location of small volumes of ground water inflow into surface water
bodies. An investigation carried out with this purpose in an area
containing a complex shallow ground water flow system indicates that the
interpretation of the thermal imageries is complicated by many factors,
among which the most important are: 1) altitude, angle of view, and
thermal-spatial resolution of the sensor; 2) vegetation type, density, and
vigor; 3) topography; 4) climatological and micrometeorological effects;
5) variation in soil type and soil moisture; 6) variation in volume and
temperature of ground water inflow; 7) the hydraulic characteristics of the
receiving water body, and 8) the presence of decaying organic material.
Despite these limitations, the thermal remote sensing method can provide an
array of hydrogeologic data not easily obtained by ground-based
techniques. (KEY TERMS: thermal remote sensors; ground water to surface
water temperature relationships; soil temperatures; landfills; ground water
flow systems.)
Specht, M. R., P. Weedier, and N. L. Fritz. 1973. New Color Film for
Water-Penetration Photography. Photo. Eng. 39:359-369.
The need for a film which will provide maximum information about underwater
detail and water characteristics from photographs made from the air has led
to the design of a special film for this purpose. A study of the
transmittance characteristics of water shows that a film with two layers
having peak sensitivities at about 480 and 550 nm will provide maximum
penetration of water with various amounts of organic matter present. It
will also allow some estimate of the amount of such material. Maximum
detectability in the processed film is accomplished by providing that the
dyes formed in the two layers are the complementary colors magenta and
green, and that the contrast of the film be high. Aerial photographs made
with an experimental film designed with the above characteristics show its
superiority over regular color film for delineating underwater detail, and
the distinct superiority of both over a film made by omitting the blue-
sensitive layer from a regular color film.
Steiner, Dieter, et al. 1974. Digital Processing of Image Data for
Automatic Terrain Recognition. In: Proceedings of the 2nd Canadian
Symposium on Remote Sensing. Guelph, Ontario, p. 59-75.
This paper reports on selected aspects of more recent work carried out
within a project dealing with the establishment of a digital Image data
processing capability. The objective is the development of techniques for
the automatic or semi-automatic extraction of data on ground features and
202
-------
parameters. The topics covered include Image data accessing, noise removal,
geometrical image registration, the matching of map to image data,
multispectral pattern recognition, spatial analysis and shadow analysis for
the purpose of deriving building parameters in urban areas.
Steiner, Dieter. 1974. Digital Geometric Picture Correction Using a
Piecewise Zero-Order Transformation. Remote Sensing of Environment
3:261-283.
This paper describes a procedure for the digital geometric registration of
digitized air photographic data. Match points are located visually and used
to formulate a global linear conformal transformation for each slave
picture. Each transformation then serves to segment the corresponding image
into regions such that, for the Implementation of the correction, all pixels
within a region can be shifted by the same number of rows and columns. In
other words, the correction Is achieved by a series of local translations
(zero-order transformations). Remaining problems are then the filling of
holes in the output pictures and the finding of a common submatrix. The
image registration which results, provides a preprocessing operation needed
to combine data from multitemporal and multispectral photography.
Steiner, Dieter. 1972. Multispectral-Multitemporal Photography and
Automatic Terrain Recognition. In: Proceedings of the 1st Canadian
Symposium on Remote Sensing. Ottawa, Ontario, p. 601-609.
During the past few years progress has been made in the automatic
recognition of terrain features on the basis of multispectral data gathering
and pattern recognition methods. Particularly notable is research performed
at the University of Michigan and at Purdue University (see, for example,
Marshall and Kriegler 1971, and Laboratory for Agricultural Remote Sensing
1970) with sophisticated hard- and software, involving multispectral
scanners and digital and/or analogue computer processing. The author of
this paper, more interested in methodological aspects than in an operational
system, started working in this area some years ago in Switzerland when at
the Department of Geography, University of Zurich. These initial
investigations were based on simple manual desitometric spot measurements on
conventional aerial photography and their subsequent classification on a
digital computer (Steiner et al. 1969). Since then, work has been continued
at the University of Waterloo with somewhat more sophisticated techniques.
The purpose of this paper is to report on progress made and future plans
within a project started two years ago. The different phases of the program
are summarized in Table 1.
Strong, Alan E. 1974. Remote Sensing of Algal Blooms by Aircraft and
Satellite in Lake Erie and Utah Lake. Remote Sensing of Environment
3:99-107.
During late summer, when the surface waters of Lake Erie reach their maximum
temperature, an algal bloom is likely to develop. Such phenomena, which
characterize eutrophic conditions, have been noticed on other shallow lakes
using the Earth Resources Technology Satellite (ERTS-1). The concentration
of the algae into long streamers provides additional information on surface
circulations. To augment the ERTS Multispectral Scanner Subsystem (MSS)
203
-------
data of Lake Erie an aircraft was used to obtain correlative thermal-lR and
additional multiband photographs. The algal bloom is highly absorptive in
the visible wavelengths but reverses contrast with the surrounding water in
the near-IR bands. The absorption of shortwave energy heats the dark brown
algal mass, providing a hot surface target for the thermal-IR scanner. A
large bloom of Aphanizomenon flos-aquae observed in Utah Lake together with
recent bloom history in Lake Erie is used to verify the Great Lakes bloom.
Tarnocai, C. 1972. The Use of Remote Sensing Techniques to Study Peatland
and Vegetation Types, Organic Soils and Permafrost in the Boreal Region
of Manitoba. In: Proceedings of the 1st Canadian Symposium on Remote
Sensing. Ottawa, Ontario, p. 323-335.
Multispectral imagery obtained in northern Manitoba was analyzed to
determine the usefulness of remote sensing techniques in studying peatlands
and permafrost. Dependable differences were found in the multispectral
response patterns obtained from thermal infrared, near infrared color,
color, panchromatic black and white and near infrared black and white
photographs of the various peatland types. These differences made possible
the separation and mapping of the peat landforms, vegetation, organic soils
and permafrost. The cyclic nature of permafrost was also monitored using
remote sensing data obtained in 1946, 1968, and 1971 and it was found that
the area of permafrost decreased at a rate of 1 percent per year over the
25-year period studied.
Taylor, M. M. 1972. Perceptual Principles Related to Remote Sensing. In:
Proceedings of the 1st Canadian Symposium on Remote Sensing. Ottawa,
Ontario.?. 497-503.
Remote sensing systems may be regarded as extensions of man's natural
senses. As such, they should be governed by the same principles that govern
natural perception. Perception is described as a means whereby information
useful for action is separated from the enormous mass of useless
information, and encoded in a way suited to rapid evaluation of potential
behavior. This functional viewpoint leads to the idea that the "attention"
of a "central processor" must be devoted at any one time to a small region
of the environment and that a behaving organism should be provided with a
large number of feature detectors which continually monitor the environment
for items that deserve the attention of the central processor and send
"alarms" when such features are detected. Attention should continually
shift except when called by these alarms. The vigilance decrement is
perhaps due to an inappropriate requirement that attention be deployed on a
single display for continuous periods of time, and might be averted by
transforming the display in such a way that targets appear in a manner to
which natural "alarm" detectors are suited. Similarly, the provision of
hardware feature detectors should be a major part of remote sensing used in
searches for known target types.
Thie, Jean. 1972. Application of Remote Sensing Techniques for Description
and Mapping Forest Ecosystems. In: Proceedings of the 1st Canadian
Symposium on Remote Sensing. Ottawa, Ontario, p. 149-169.
204
-------
Remote sensing provides environmental studies with a true third dimension
(accurate area distribution of parameters) and adds a fourth dimension,
time. Careful interpretation of remote sensing Imagery for individual
elements of a forest ecosystem, like relief, drainage, vegetation, living
organisms and time, supplies new and better information to describe and map
ecosystems and to understand the interaction of its elements. Existing and
potential applications of remote sensing techniques to analyse and describe
these elements are discussed and examples given in three case studies.
Thiessen, H. W. 1972. A Proposed Organization for the Efficient
Interpretation of Remote Sensing Data. In: Proceedings of the 1st
Canadian Symposium on Remote Sensing. Ottawa, Ontario, p. 719-722.
One season's experience of remote sensing service with the Remote Sensing
Center in Alberta is described. This experience included several sources of
photography in a variety of locations throughout the auspices of the
interdepartmental Conservation and Utilization Committee. In addition,
several departments of the federal government, and the University of
Alberta, have participated in remote sensing photography within Alberta
during the same period. Prior to this time, the researcher had other
experiences with private contractors utilizing remote sensing in several
interdisciplinary studies. Based on these experiences, Thiessen is
convinced that the existing organization and use of remote sensing is not
meeting the challenge of the need for this capability, nor Is It
anticipating the full potential of the technology. The development of an
organization involving interdisciplinary analysis comprised by a variety of
disciplines and supported by several levels of government and research
ins iti tut ions is propsoed in order to realize the full potential of this
science with the greatest efficiency to our taxpayers.
Ulaky, Fawwaz T., and R. K. Moore. 1973. Radar Spectral Measurements of
Vegetation. In: Proceedings of the Am. Soc. of Photogrammetry Fall
Convention. Lake Buena Vista, Florida, p. 322-334.
During the 1972 growing season 4-8 GHz radar backscatter spectral data was
gathered at look angles between 0° and 70° for all four possible
polarization linear combinations. The data covers four crop types (corn,
milo, alfalfa, and soybeans) and a wide range of soil moisture content. To
insure statistical representation of the results, measurements were
conducted over 128 fields corresponding to a total of about 40,000 data
points. This paper investigates the use of spectral response signatures to
separate different crop types and to separate healthy corn from blighted
corn.
Underhi.il, M. A. 1972. Problems in Relating User Requirements to
Quantitative Parameters of Imagery Quality. In: Proceedings of the 1st
Canadian Symposium on Remote Sensing. Ottawa, Ontario, p. 533-546.
The techniques involved in the methodology of remote sensing are numerous
and are generally subject to reasonably precise analytical measurement. The
user (interpreter) of imagery from remote sensing systems is also trained in
a scientific field whether it be forestry, geology or civil engineering, for
example, and is used to dealing with precise analytical measurements in his
205
-------
own discipline. Unfortunately the quantitative terras which define image
quality do not easily or formally translate into "usefulness" in terms of
the discipline of the user requesting the imagery. Hence we are faced with
the problem of quantifing those attributes of imagery that make it useful to
the interpreter. More simply, what physical attributes constitute a "good"
picture? And is that degree of goodness the same for all users? This paper
will examine briefly some possible approaches in the problem of relating
quantitative measurements of image quality in the user's requirements in
different disciplines.
Wacker, A. G., and D. A. Landgrebe. 1972. Minimum Distance Classification
in Remote Sensing. In: Proceedings of the 1st Canadian Symposium on
Remote Sensing. Ottawa, Ontario, p. 577-599.
The utilization of minimum distance classification methods in remote sensing
problems, such as crop species identification, is considered. Minimum
distance classifiers belong to a family of classifiers referred to as sample
classifiers. In such classifiers the items that are classified are groups
of measurement vectors (e.g. all measurement vectors from an agricultural
field), rather than individual vectors as in more conventional vector
classifiers. Specifically in minimum distance classification a sample (i.e.
group of vectors) is classified into the class whose known or estimated
distribution of the sample to be classified. The measure of resemblance is
a distance measure in the space of distribution functions. The literature
concerning both minimum distance classification problems and distance
measures is reviewed. Minimum distance classification problems are then
categorized on the basis of the assumption made regarding the underlying
class distribution. Experimental results are presented for several
examples. The objective of these examples is to: (a) compare the sample
classification accuracy (% samples correct) of a minimum distance
classifier, with the vector classification accuracy (% vector correct) of a
maximum likelihood classifier; (b) compare the sample classification
accuracy of a parametric with a nonparametric minimum distance classifier.
For (a), the minimum distance classifier performance is typically 5% to 10%
better than the performance of the maximum likelihood classifier. For (b),
the performance of the nonparametric classifier is only slightly better than
the parametric version. The improvement is so slight that the additional
complexity and slower speed make the nonparametric classifier unattractive
in comparison with the parametric version. In fact disparity between
training and test results suggest that training methods are of much greater
importance than whether the implementation is parametric or nonparametric.
Watson, Robert D., W. R. Hemphill, and T. 0. Hess in. 1973. Quantification
of the Luminescence Intensity of Natural Materials. In: Proceedings of
the Am. Soc. of Photogrammetry Symposium on the Management and
Utilization of Remotely Sensed Data. Sioux Falls, South Dakota.
p. 364-376.
Rhodamlne WT is an artificial, water-soluble, organic dye used as a tracer
by hydrologists and oceanographers to monitor the dynamics of currents in
rivers and estuaries. In 1969, rhodamine dye was detected in sea water in
concentrations of less than 2 vig/liter (2 ppb) with the aid of a prototype
Fraunhofer line discriminator, an optical-mechanical remote-sensing device,
206
-------
which has been used in the detection of solar-stimulated luminescence from
aircraft. Continuing experiments with a laboratory fluorescence
spectrometer indicates that luminescence of some samples of crude and
refined petroleum exceeds the luminescence intensity of rhodamine dye in
concentrations of 10,000 g/liter (10,000 ppb). It has also been determined
that luminescence of some conifer needles and other plants is comparable to
dye concentrations of 50 g/liter (50 ppb). Conifer needles collected from
trees growing in high-copper-zinc soil west of Denver, Colorado, show
markedly greater luminescence intensity than needles from conifers growing
nearby in unmineralized areas. These luminescene intensities appear to be
within the sensitivity limits of a Fraunhofer line discriminator.
Wilson, C. L., and R. H. Rogers. 1975. Multispectral Data Systems for
Energy Related Problems. In: Remote Sensing Energy-Related Studies.
T. Nejat Veziroqu, ed. Hemisphere Pub. Co. Seattle, Washington.
p. 404-429.
A multispectral data system consists of data collection, data processing,
data analysis and interpretation, and information dissemination
(application) subsystems. The ERTS MSS and the data processing facility at
Goddard Space Flight Center are examples of the first two subsystems
respectively. The characteristics of airborne and satellite digital
multispectral scanner data are described, together with the digital data
processing and analysis techniques which have been developed to
automatically analyze and Interpret the data. Techniques for output product
generation and information dissemination are illustrated using strip mine
monitoring and environmental Impact assessment for power plant siting and
transmission line routine as examples. The problems associated with the
development of semi-automated performance of the cited applications are
discussed, as well as the current status of the development programs.
Expected trends in the development of operational application subsystems are
presented.
Worsfold, R. D. 1972. A Qualitative Study of Kodak Aerochrorae Infrared
Film, Type 2443, and the Effect Produced by Kodak Colour Compensating
Filters, at High Altitudes. In: Proceedings of the 1st Canadian
Symposium on Remote Sensing. Ottawa, Canada, p. 417-427.
AETE Uplands, Photo Development (now Canadian Forces Airborne Sensing Unit),
on 18 August 1970 carried out film/filter tests using Aerochrome Infrared
Film, Type 2443, at altitudes of 20,000 ft and 40,000 ft with V in ten 492 and
Vinten 547 aerial reconnaissance cameras. The purpose of the test was to
contribute film/filter samples to a catalogue of examples of forested land,
land forms, geologic formations, agricultural land, and industrial land
being compiled for mission planning. Utilizing the various film/filter
combinations and evaluating each filter with respect to the film, selected
positive transparencies, with seventy millimeter format and in
stereo triplets, were collected and analyzed for use in the Canadian Forces
Airborne Sensing Unit (CFASU). From the examples, a study was undertaken to
determine which of the examples could be interpreted in two ways; common
photo interpretation techniques and infrared information content. The
results could be tabulated in relation to each other. The film/filter
207
-------
combinations are available at CFASU for the use of investigators in mission
planning.
Yost, Edward, and S. Wenderoth. 1971. Multispectral Color for Agriculture
and Forestry. Photo. Eng. 37:590:604.
The potential usefulness of multispectral color photography for the
identification of crop and tree species has been demonstrated in a series of
controlled experiments using broad-band camera filters which approximate the
spectral sensitivity of color and color-infrared films. ' Independent
adjustment of exposure in each camera band, control over the gamma and
density of the photographic image, along with the ability to adjust the hue,
brightness, and saturation in viewing, resulted in greater Image chromatic
separation than could be achieved using subtractive color reversal films.
The capability to compensate for variations in the solar illuminant and
atmospheric attenuation using additive color viewing of multispectral
photographs was demonstrated.
Zsilinszky, V. G. 1972. Camera Mounts for 35 ram Mono and Multi-Spectral
Photography. In: Proceedings of the 1st Canadian Symposium on Remote
Sensing. Ottawa, Ontario, p. 441-450.
Supplementary aerial photography with miniature cameras is a commonly used
tool in various resource survey activities of the Ontario Department of
Lands and Forests. Developmental work, specific and routine applications
require both conventional and unconventional treatments. Sometimes a single
camera is capable of providing complete information, while at other times
combination photography is prescribed. Consequently, mounts of various
capacity have been engineered for single, two, three and four cameras. Each
working model is described and illustrated, principles involved are
discussed and it is suggested that the designs offered could .apply to most
miniature cameras and light aircraft with minor alterations or with no
modifications at all.
Additions to REMOTE SENSING—GENERAL INFORMATION
American Society of Photogrammetry. 1975. Manual of Remote Sensing. Falls
Church, Virginia. 2144 p.
The Manual of Remote Sensing was written (1) to replace and update material
contained in the Manual of Photographic Interpretation and (2) to provide
information on the new techniques of Remote Sensing and their uses.
Volume I includes extensive treatment of the electromagnetic theory behind
Remote Sensing; interaction of electromagnetic radiation with solid, liquid
and gaseous matter; and the instruments and their platforms used to obtain
remote sensor data. The techniques of processing these data to allow the
interpreter to obtain the maximum Information from them is also included.
Volume II treats the fundamentals of imagery interpretation, and illustrates
in detail how Remote Sensing may be applied to a wide variety of scientific
and natural resources fields.
208
-------
Lillesand, T. M., and R. W. Kiefer. 1979. Remote Sensing and Image
Interpretation. John Wiley and Sons. New York. 612 p.
This is a definitive remote sensing and image interpretation text. Its ten
chapters cover concepts and foundations of remote sensing; elements of
photographic systems; basics of airphoto interpretation; airphoto
interpretation for terrain evaluation; photogrammetry; radiometric
characteristics of aerial photographs; aerial therraography; multispectral
scanning and spectral pattern recognition; microwave sensing; and remote
sensing from space.
Swain, P. H., and S. M. David, eds. Remote Sensing: The Quantitative
Approach. McGraw-Hill, New York, New York. 396 p.
This technical reference book covers radiation and instrumentation in remote
sensing; fundamentals of pattern recognition in remote sensing; data-
processing methods and systems; biological and physical considerations in
applying computer-aided analysis techniques to remote sensor data; applying
the quantitative approach, and useful information for multispectral image
data: another look.
REMOTE SENSING OF WETLANDS
The following references are concerned with identifying wetland
vegetation, high water lines, and mapping wetlands using a variety of
photographic and multispectral scanner platforms. The references deal with
both coastal and freshwater wetland systems.
Anderson, Richard R., Virginia Carter, and John McGinness. 1973.
Applications of ERTS Data to Coastal Wetland Ecology with Special
Reference to Plant Community Mapping and Typing and Impact of Man. In:
Proceedings of the Third Earth Resources Technology Satellite-1
Symposium, Vol. 1, Technical presentations, Section B.
Complete seasonal ERTS-1 coverage of Atlantic coastal wetlands from Delaware
Bay to Georgia provides a basis for assessments of temporal data for wetland
mapping, evaluation, and monitoring. Both MSS imagery and digital data have
proved useful for gross wetland species delineation and determination of the
upper wetland boundary. Tidal effects and (band to band or seasonal)
spectral reflectance differences make it possible to type vegetatively
coastal wetlands in salinity-related categories. Management areas, spoil
disposal sites, drainage ditches, lagoon-type developments, and highway
construction can be detected indicating a monitoring potential for the
future. A northern test site (Maryland-Virginia) and a southern test site
(Georgia-South Carolina), representing a range of coastal marshes from
saline to fresh, were chosen for intensive study. Wetland maps were
produced at various scales using both ERTS imagery—band 5 and 7—and
digital data—bands 4, 5, and 7. A Bausch and Lorab Zoom Transfer Scope and
various overlay techniques were used with either 9 1/2" black and white
transparencies or enlarged black and white prints. Diazo color composites,
color enhancement techniques, and multi-spectral manipulation were used to
supplement this Information. Data will be useful for coastal wetland
209
-------
inventories, updating acreage estimates, mapping boundaries, detecting
seasonal changes, detecting and monitoring man's impact. Resolution
limitations allow for mapping to a 1/125,000 scale. Results are being
applied directly to a Dismal Swamp study with U.S. Geological Survey. There
is potential application to on-going programs in Georgia and South Carolina
and to the Coastal Zone Management Act and a National Wetlands law.
Anderson, Richard R., Virginia Carter, and John McGinness. 1973. Mapping
Atlantic Coastal Marshlands, Maryland, Georgia, Using ERTS Imagery.
In: Proceedings of the 39th Annual Meetng of the Am. Soc. of
Photogrammetry. Vol. 1. p. 615-625.
Eastern coastal marshes are the most extensive and productive in the United
States. A relatively low cost, moderately accurate method is needed to map
these areas for management and protection. Ground based and low-altitude
aircraft methods for mapping are time-consuming and quite expensive. The
launch of NASA's Earth Resources Technology Satellite has provided an
opportunity to test the feasibility of mapping wetlands using small scale
imagery. The test sites selected were in Chesapeake Bay, Maryland and
Ossabaw Island, Georgia. Results of the investigation indicate that the
following may be ascertained from ERTS imagery, enlarged to 1:250,000; (1)
upper wetland boundary; (2) drainage pattern in the wetland; (3) plant
communities such as "nef Juncue
roemerianu.8; (4) ditching activities associated with agriculture; (5)
lagooning for water-side home development. Conclusions are that ERTS will
be an excellent tool for many types of coastal wetland mapping.
Anderson, R. R., and F. J. Wobber. 1973. Wetlands Mapping in New Jersey.
Photo. Eng. 39(4):353-358.
The New Jersey Wetlands Act of 1970 required that mapping and inventory of
wetland along the marine coastal zone and tide-influenced estuaries of the
State be properly managed. A prime requirement was that map products have
validity which could withstand the challenge of litigation. Natural-color
and color-infrared aerial photographs at a scale of 1:12,000 were obtained
over two sites designated by the State. Final map products were prepared
containing (a) the upper wetlands boundary; (b) the line of biological mean
high water to establish state riparian lands; and (c) delineation of major
plant species associations of five acres or larger in size. The state-wide
wetlands mapping effort will be one of the largest operational remote
sensing projects ever conducted. The authors believe that the methods
developed, ecological data collected, and products prepared will have
far-reaching effects on future coastal zone programs.
Bartlett, D. S., V. Klemas, R. H. Rogers, and N. J. Shag. 1977,
Variability of Wetland Reflectance and its Effect on Automatic
Categorization of Satellite Imagery. In: Proceedings of the 43rd
Annual Meeting of the Am. Soc. of Photogrammetry. Washington, D.C.
p. 70-80.
A technique for training automated analysis of satellite (LANDSAT)
multispectal data based on in situ measurements of target reflectance was
tested and applied in delineating cover typed in Delaware's tidal
210
-------
wetlands. This technique evaluated used in situ measurements of target
radiance and an atmospheric correction procedure to derive reflectance
signatures for land-cover categories in preference to the relative radiance
signatures traditionally derived from training samples within the satellite
data itself. Land cover categorization of data from the same overpass in
four test wetland areas was carried out using a four-category classification
system. The tests indicate that training data based on in situ reflectance
measurements and atmospheric correction of LANDSAT data can produce
comparable accuracy of categorization to that achieved using more
conventional relative radiance training. The analysis of the four wetlands
cover categories (Salt March Cordgrass, Salt Hay, Unvegetated, and Water
Tidal Flat) produced overall classification accuracies of 85% by
conventional relative radiance training and 81% by use of in aitu
measurements. Overall mapping accuracies were 76% and 72% respectively.
Further refinement of the atmospheric correction and ground measurement
procedures should produce better accuracies in a more operational mode. In
addition, field measurements showed that variability in spectral reflectance
was, as expected, symptomatic of significant physical characteristics of the
test cover types such as time elapsed since tidal inundation of mud, plant
height, and growth form. Significant correlations were found between single
band reflectances and tidal Inundation and plant morphologic
characteristics. Optimazation of seasonal sampling procedures for detection
of plant morphologic parameters is suggested.
Boissonneau, A. N., and J. K. Jeglura. 1975. A Regional Level of Wetlands
Mapping for the Northern Clay Section of Ontario. In: The Third
Canadian Symposium on Remote Sensing. Edmonton, Alberta, p. 349-357.
A study in the Hicks Township, 72 kilometers northwest of Timmins, in the
Northern Clay Section of Ontario, was selected to test the feasibility of
using remote sensing data to map the wetland types of a proposed wetlands
classification framework. It was found that although broad patterns of
wetlands types could be mapping using densitometric analyses of LANDSAT
imagery, this section could be most efficiently classified using large-scale
aerial photography. It was also found that large-scale photography could be
used to identify the most detailed level of the proposed wetlands
classification. In the regions to the north of the Northern Clay Section
which are vast and almost exclusively wetlands, and for which LANDSAT
imagery provides the best or the only remote sensing data, it is anticipated
that densitometric analysis of LANDSAT imagery will be an efficient means of
extrapolating ground truth data to the lands of these regions.
Bright, C. R., and H. R. Pywell. 1978. Data Base Development for the
National Wetlands Inventory. In: Proceedings of the 44th Annual
Meeting of the Am. Soc. of Photogrammetry. Washington, D.C.
p. 329-334.
Autometric, Inc. has developed a data base system for the storage and
retrieval of digital geographic data. This information is stored on a
geounit basis which may be queried through a FORTRAN program written to
interrogate the data base. Reports on wetland area and types may be
generated along with 7 1/2 and 15 minute quadrangle plots utilizing an on-
line, high speed digital plotter. The data base structure is also
211
-------
supportive to aerotriangulation and digitizing stations for the generation of
wetland information.
Brocon, W. W. 1978. Wetland Mapping in New Jersey and New York. Photo. Eng.
44(3):303-314.
The wetlands of New Jersey and New York were mapped recently using 1:12,000
scale color and/or color infrared aerial photographs. In support of tidal
wetlands legislation, the Mean High Water (approximate position) and Upper
Wetland Boundary lines were delineated using a biological entity—plant
species. In New Jersey, dominant plant species were identified on each map
(Anderson and Wobber 1973). In New York, a broader classification system was
used based on plant species categories such as Coastal Fresh Marsh, High
Marsh, etc. These projects represent two distinct approaches by which wetland
surveys may be conducted. For successful Implementation of projects of this
type, it is critical that mapping conventions and procedures be developed at
the onset of the program. These mapping criteria, howeer, may be modified as
the program proceeds and the need arises. Also, a thorough understanding of
aerial photographic interpretations of plant species signatures under varying
conditions is essential.
Carter, Virginia. 1974. Remote Sensing Applications to the Dismal Swamp.
Paper Presented at the Great Dismal Swamp. Symposium, Old Dominion
University, Norfolk, Virginia. 34 p.
The recent Dismal Swamp Study (Public Law 92-478) is an example of the use of
remotely sensed data in a multidisclplinary study to assess the multiple-use
potential of a large Inland wetland. Remotely sensed data available for the
Dismal Swamp include aircraft photography, Earth Resources Technology
Satellite (ERTS) data, and thematic extractions. These data were applied to
(1) overall study area selection, (2) location of intensive study areas, (3)
hydrologic studies, (4) vegetation mapping, and (5) field studies Including
identification of special interest areas. The large size of the Dismal Swamp
and the inaccessibility of many interior parts makes remote sensing an
important tool to meet the needs of the current study as well as future
research and management.
Carter, V., A. Voss, D. Malone, and W. Godsey. 1976. Wetland Classification
and Mapping in Western Tennesee. In: Proceedings of the Second Annual
William T. Pecora Memorial Symposium. Sponsored by The American Society
of Photogrammetry and the U.S. Geological Survey, p. 206-220.
The U.S. Geological Survey and the Tennessee Valley Authority are presently
conducting a cooperative wetland mapping project in western Tennessee.
Existing wetland classification systems were considered too general to supply
needed management information, so a new system has been developed based
primarily on vegetation, and frequency and duration of inundation. There are
five forested wetland subclasses and seven non-forested wetland subclasses in
the new system. High altitude color infrared photography was acquired by the
National Aeronautics and Space Administration during several seasons. This
photography supplied the information on hydrologic boundaries and vegetation
that is needed for classification and mapping. Seasonal information was
212
-------
required to map the maximum number of categories. The methodology for
separating and delineating classes was carefully documented. The state (water
level) was determined for the time of overflights for sites where gage
stations are in operation. Under the new classification system, wetlands at
four sites were mapped at 1:24,000 scale as overlays on the existing
Geological Survey 7.5-rainute topographic map series. Adjacent land use was
also mapped, but in less detail than wetlands. Overlays for separate dates
were combined to make the final camera ready copy. A lithographed map of
wetlands and land use was made for one of the five quadrangles covering the
Reelfoot Lake site. The stage at time of photography was referenced to a
stage-duration curve, placed on the map collar, to show that boundaries are
representative of average water levels rather than extreme highs or lows.
Carter, Virginia, and Jane Schubert. 1974. Coastal Wetlands Analysis from
ERTS MSS Digital Data and Field Spectral Measurements. In: Ninth
International Symposium on Remote Sensing of Environment, Ann Arbor,
Michigan, p. 1241-1260.
Classification, delineation and evaluation of coastal wetlands can be made on
the basis of major vegetative associations. To produce wetland maps, two
vegetation-analysis look-up tables were developed for use in the ERTS ANALYSIS
System. These look-up tables are based on Earth Resources Technology
Satellite (ERTS) digital values in Multispectral Scanner (MSS) bands 4, 5, and
7 and were developed using seasonal spectral reflectance measurements from
field observations. Computer-generated maps at an approximate scale of
1:120,000 were produced for the primary test site, Chincoteague Marsh,
Virginia. There is a high degree of accuracy in the identification of wetland
features and plant associations. The classification was also tested on other
Atlantic coast salt marshes and a brackish marsh in the Chesapeake Bay.
Carter, Virginia, and Richard R. Anderson. 1974. Multispectral Analysis for
Wetland Studies. Paper Presented at the Winter Meeting of the Am. Soc. of
Agri. Eng., St. Joseph, Michigan. 16 p.
Multispectral data have been used successfully for wetland studies. The
applications discussed include the use of the Earth Resources Technology
Satellite (ERTS) Multispectral Scanner (MSS), the SKYLAB S190A Multispectral
Photographic Facility, and the Bendix 24-channel scanner. ERTS imagery and
digital data have been used to classify wetlands, map coastal wetland
features, and estimate primary productivity (Virginia and South Carolina).
ERTS data have also been used to study the vegetation and hydrology of the
Dismal Swamp (Virginia/North Carolina). SKYLAB data are being used to
classify and map wetlands in the Chesapeak Bay area. Multispectral data from
low-altitude aircraft are presently being used to determine species
composition in fresh, brackish, and saline marshes in the Chesapeake Bay area.
Civco, D. L., W. L. Kennard, and M. W. Lefor. 1978. A Technique for
Evaluating Inland Wetland Photolnterpretation: the Cell Analytical Method
(CAM). Photo. Eng. Vol. 44(8):1045-1052.
A procedure for objectively analyzing the inland wetland photointerpretation
of several investigators is discussed. Comparisons are made between wetland
photointerpretations, soils mappings, and ground truth. The technique, which
213
-------
permits mathematical treatments of cell-encoded wetland delineations, was
developed to test inland wetland mapping methods. Results indicate that
interpretation of false-color infrared (FCIR) aerial photographs produces more
accurate wetland delineations than can be obtained from soils maps, especially
at a mapping unit level useful in wetlands management. Desirable qualities of
photo interpreters who are to perform wetland delineations are discussed as are
the relative merits, in terms of wetlands characterization, of aerial
photography from different seasons.
Cowardin, L. M., and V. I. Myers. 1971. Remote Sensing for Indentification
and Classification of Wetland Vegetation. J. Wild. Manage. 38(2):308-314.
Multispectral photography and ground truth were obtained on an area 12 miles
(19.3 km) east of Bemidji, Minnesota, to identify and map wetlands less than 2
acres (0.8 ha) in size, to map emergent vegetation in lakes, and to explore
the feasibility of classifying vegetation from aerial photographs. Wetlands
less than 2 acres in size were identified on photography taken in May 1971,
and emergent vegetation was recorded on purposely overexposed infrared black
and white photography from a flight in September 1971. Several vegetation
types and species groups were recognizable with the aid of color, color
infrared, and black and white infrared photography. Proper timing of flights,
use of multispectral photography, and knowledge of the ecology of the area are
considered essential for wetland mapping by remote sensing.
Egan, W. G., and M. E. Hair. 1969. Automated Delineation of Wetlands in
Photographic Remote Sensing. In: Proceedings of the 7th International
Symposium on Remote Sensing of the Environment. Ann Arbor, Michigan.
Vol. 3. p. 2231-2252.
Precision automated photometric mapping of wetlands in Calvert County,
Maryland has been achieved in an operational system as the result of a program
including aerial color film (both true color and false color infrared)
calibration and control. Although the system was operated over this area, it
may be adapted to other areas. The recognition appears to be most accurately
achieved by microdensitometric analysis of the true color transparency in a
narrow band centered in the red (0.633 urn), on 3000-ft altitude Imagery. A
computer generated map is obtained.
Enslin, W. R., and M. C. Sullivan. 1974. The Use of Color Infrared
Photography for Wetlands Assessment. In: Remote Sensing of Earth
Resources Conferences. The University of Tennessee Space Institute,
Tullahoma, Tennessee. Vol. III. p. 697-720.
A study was undertaken of Pointe Mouille Marsh, located on Lake Erie, to
assess shoreline erosion and to inventory and evaluate adjacent land as
potential replacement for areas lost to erosion, and to provide better data
sources for management decisions. The results of the study were (1)
evaluation of low altitude oblique photography was useful in determining
specifications of operational mission requirements; (2) Accurate base map
revisions reflecting shoreline erosion were made using aerial photography and
a Zoom Transfer Scope; (3) An aerial land cover inventory provided data
necessary for selection of adjacent lands suitable for marshland development;
(4) A detailed inventory of vegetative communities (mapped from CIR), was made
214
-------
for management decisions; and (5) A carefully selected and well laid-out
transect was a key asset to photo interpretation and analysis of vegetation.
Fornes, Ann 0., and Robert J. Reimold. 1973. The Estuarine Environment:
Location of Mean High Water—Its Engineering, Economic and Ecological
Potential. In: Proceedings of the Am. Soc. of Photogrammetry Fall
Convention, Lake Buena Vista, Florida, p. 938-978.
The demarcation of mean high water is Important in terms of utilization and
protection of the coastal zone. The paper considers various methods by which
mean high water was located in a salt marsh. The work was conducted at two
study sites located on the Duplin Estuary, Sapelo Island, Georgia. Each site
had a number of photo-identifiable targets established prior to the time the
photographic mission was flown. Bench marks on a first order level were
available nearby in order to tie in elevations of the marsh to a U.S.G.S.
bench mark. Also, four recording tide gauges, monitored for over a year,
permitted determination of local mean high water.
Ground control for each site was established by surveying elevations of
the large white disks and obtaining distances between them. This information
was used to rectify color infrared positive transparencies for use in the Kern
PG-2 plotter. The elevation of the calculated mean high water was then
photogrammetrically located.
Delineation of plant species using color infrared photographs represented
the method of biologically locating mean high water. In this case, the
vegetation was considered to be an accurate indicator of environmental
conditions.
Finally, thermal Imagery was considered, attempting to locate an isotherm
coincident with mean high water. This was based on the premise that the most
frequently flooded zones would be wetter, thus cooler, whereas drier areas,
more often escaping flooding, would be warmer. Each method was then viewed in
terms of the cost, accuracy, speed of delineation and overall validity.
Topographic mapping was found to be slow and inaccurate, despite the fact
that it is the standard procedure used for upland surveys. Biological and
thermal mapping techniques were faster techniques by which to delineate mean
high water. Both were also more accurate, as such boundaries are easily
recognizable. Variation in area covered was greatest, however, in the thermal
imagery method. The biological method, therefore, was not only more accurate
and faster to delineate than topographic techniques, but also afforded a high
degree of reproducible results, not obtainable by the other two methods. With
the present knowledge of remote sensing techniques, biological delineation
appears to be the most effective method to locate mean high water.
Gallagher, J. L., D. E. Thompson, and R. J. Reimold. 1972. Remote Sensing
and Salt Marsh Productivity. In: Proceedings of the Symposium on Coastal
Mapping. Am. Soc. of Photogrammetry. Falls Church, Virginia, p. 18-31.
The feasibility of using photography or thermal imagery from fixed wing
aircraft for assessing salt marsh productivity is being investigated. Kodak
Aerochrome Infrared 2443 and Aerocolor negative 2445 films and a Bendix
Thermal Mapper are being used. Remote sensing flights (from 1,250 to
20,000 ft) are made in conjunction with acquisition of ground truth data
consisting of chlorophyll per unit area, the density of living and dead plants
by number, and bioraass for each species and each one-half meter height class.
215
-------
A number of windows in the 2 to 13 micrometer wavelength range are being
tested with the thermal mapper to ascertain which gives the best
discrimination of spatial productivity patterns. Color enhancement of the
imagery has proven useful in defining the extent of these areas and predicting
tidal hydrography in the intertidal zone.
Color patterns in the color-infrared photography, due in part to
differences in plant species, density, growth form and pigmentation, are being
quantified by: (1) planimetry of hand-drawn visual interpretations of
projected transparencies; (2) planimetry of visual interpretation using a Kern
plotter; and (3) microdensitometry of each zone defined by a Joyce Loebl
Microdens 1tometer coupled with a four color isodensitracer.
Photographs are used to predict productivity at untested locations within
the study areas and ground truth is collected at these sites during subsequent
sampling periods.
Gammon, P. T., and V. Carter. 1979. Vegetation Mapping with Seasonal Color
Infrared Photographs. Photo. Eng. 45(l):87-97.
The Great Dismal Swamp of Virginia-North Carolina is a forested wetland which
has been extensively altered by fire, timbering and ditching. Seasonal high
and low-altitude color infrared photographs of the swamp have been used to
Identify and map specific swamp vegetative communities. These photographs
provided the capability to distinguish among deciduous species, to evaluate
understory, to separate broad-leaved evergreen and deciduous species, and to
locate several special community types. Comparisons made of data from
different seasons frequently helped to distinguish between other wise obscure
classes. Vegetative cover classes for the Great Dismal Swamp were defined to
provide maximum habitat information for management of the swamp by the U.S.
Fish and Wildlife Service. These classes were based on dominant canopy
designations and 243 specific vegetative communities were distinguished.
Class combinations in the map units were ranked by relative dominance as
observed on the color infrared photographs. Evaluation of class accuracy was
accomplished by helicopter overflight using sample sites selected by two
methods. A canopy or understory map unit was considered correct if at least
one of the classes was Identified in the field sample. Using this criterion,
canopy accuracy was 93.8 percent and understory accuracy was 90.5 percent. A
vegetation map was prepared at a scale of 1:100,000 using 'a U.S. Geological
Survey 7.5 minute orthophotomosaic as the base map.
Garvin, Lester E., and Richard H. Wheeler. 1972. Coastal Wetlands Inventory
in Maryland. In: Proceedings of the Symposium on Coastal Mapping. Am.
Soc. of Photogrammetry. Washington, D.C. p. 18-31.
In 1970 the State of Maryland passed legislation to protect its coastal
wetlands from further despoliation through uncontrolled dredging, ditching and
filling. In October 1971, Raytheon/Autometric began working with the State
Agency responsible for implementing the law, the Department of Natural
Resources, to undertake a statewide inventory of coastal wetlands. The
legislation and inventory methodology are described. The methodology of
implementation is cost effective and represents a sound approach to the
problem of protecting finite natural resources.
216
-------
Guth, Jack E. 1974. Will the Real Mean High Water Line Please Stand Up.
In: Proceedings of the Am. Soc. of Photogranmetry Fall Convention,
Washington, D.C.p. 33-44.
The accurate location of the Mean High Water (MHW) line is of primary
importance for determination of coastal property boundaries. A clear
understanding of tidal characteristics is required to accomplish this. Some
historical survey methods are no longer acceptable. A proposed system is
identified by the initials TAG to represent its three basic elements: T for
tide data, A for aerial photographic mapping, and G for ground surveys. There
are areas where it is not practical to establish the MHW line by methods
normally used. In such cases political determination may be the most
reasonable solution; however, this alternative should be used with caution.
Hubbard, J. C. E., and B. H. Grimes. 1972. Coastal Vegetation Surveys.
In: Environmental Remote Sensing: Applications and Achievements, Eric C.
Barrett and Leonard F. Curtis, eds. Crane & Russak, New York. p. 129-
141.
Aerial photography has proved to be a fruitful means of documenting and
analyzing vegetation in coastal zones. This paper reports some of the recent
successes achieved through the investigation of coastal patterns recorded in
both the visible and non-visible portions of the electro-magnetic spectrum.
Drawing examples from the coasts of southern Britain, France and Spain, it
deals in turn with the vegetation of mudflats, salt-marshes, shingle beaches,
sand-dunes, and cliffs. Few such habitats lend themselves to easy mapping on
the ground, owing to such factors as inaccessibility, the tidal cycle, and the
instability of certain surfaces.
Kleraas, V., D. Bartlett, and F. Daiber. 1973. Mapping Delaware's Coastal
Vegetation and Land Use from Aircraft and Satellites. In: Proceedings of
the Am. Soc. of Photogrammetry Fall Convention, Lake Buena Vista,
Florida, p. 926-937.
Coastal vegetation species appearing in ERTS-1 images taken of Delaware Bay
have been correlated with ground truth vegetation maps and imagery obtained
from RB-57 and U-2 overflights. Multlspectral analysis of the high altitude
R3-57 and U-2 photographs indicates that four major vegetation communities can
be clearly discriminated from 60,000 ft altitude including (1) salt marsh cord
grass (Spartina alterniflora) (2) salt marsh hay and spike grass (Spartina
patens and Distichlis apioata), (3) reed grass (Phragmites conmmis), and (4)
hight tide bush and sea myrtle (Iva fruteaoena and Baccharia halimifolia). In
addition, human impact can be detected in the form of fresh water impoundments
built to attract water fowl, dredge-fill operations and other alterations of
the coastal environment. Overlay maps matching the USGS topographic map size
of 1:24,000 have been prepared showing the four wetland vegetation
communities, fresh water impoundments and alteration of the wetlands by
mosquito control ditching and dredge-fill operations. Using these maps for
basic ground truth, ERTS-1 images were examined by human interpreters and
automated multispectral analyzers. Major plant communities of (1) Spartina
alterniflora, (2) Spartina patens and Distichlia spicata, and (3) Iva
fruteacene and Baaaharie halimifolia can be distinguished from each other and
from surrounding uplands in ERT-1 scanner bands #6 and #7. Phragmites
217
-------
aommunis, which naturally occurs in small dispersed patches, can be identified
only in the heavily disturbed marshes of northern Delaware where it has
propagated over large areas. Fresh water impoundments built to attract water
fowl, major dredge-fill construction and other vestiges of human land-use can
also be identified in ERTS-1 scanner bands #5, #6, and #7. The potential for
monitoring such activity from space appears considerable. The maps showing
coastal vegetation species and land use, including changes in land use, are
being further developed into relative value maps for the wetlands. It is this
kind of information that public officials and planners need to assess socio-
economic benefits of coastal development with full recognition of potential
environmental impacts.
Klemas, V., F. Daiber, D. Bartlett, 0. Crichton, and A. Fornes. 1973.
Application of Automated Multispectral Analysis to Delaware's Coastal
Vegetation Mapping. In: Proceedings of the 39th Annual Meeting of the Am.
Soc. of Photogrammetry. Washington, D.C. p. 512-519.
Overlay maps of Delaware's wetlands have been prepared, showing the dominant
species or group of species of vegetation present. Five such categories of
vegetation were used indicating marshes dominated by (1) salt marsh cord grass
(Spartina altemiflova.), (2) salt marsh hay and spike grass (Spartina patens
and Distichlis spioata), (3) reed grass (Phrvigmitee eommunia), (4) high tide
bush and sea myrtle (Ivo. species and Boocharia halimi folia) t and (5) a group
of fresh water species found in Impounded areas built to attract water fowl.
In addition, major secondary species were indicated where appropriate. Small,
representative areas of each of the major marsh regions were analyzed and
enhanced to show detailed growth patterns not shown on the larger scale maps.
The mapping technique employed utilizes the General Electric
Multispectral Data Processing System (GEMDPS) to analyze NASA RB-57 color-
infrared Imagery. The GEMSDPS is a hybrid analogue-digital system designed as
an analysis tool to be used by an operator whose own judgment and knowledge of
ground truth can be incorporated at any time into the analyzing process. The
operator can combine his knowledge of the scene gained in the field with
electronic analysis and (1) measure the spectral characterises of any chosen
region of any size in the scene, (2) search the scene for regions with similar
characteristics and once they are identified, enhance and store them, (3)
modify the stored image if necessary to make it compatible with his knowledge
of the area, and (4) read out the percentage of the total scene occupied by
regions with the specified spectral signature. By repeating Che procedure for
other regions in the scene, the operator can quickly produce a composite of
regions of interest. The result is a high speed cost effective method for
producing enhance maps of a number of spectral classes—each enhanced spectral
class representative of a vegetative species or group of species.
Klemas, V., D. Bartlett, W. Philpot, R. Rogers, and L. Reed. 1974. Coastal
and Estuarine Studies with ERT-1 and Skylab. Remote Sensing of
Environment. 3:153-174.
Coastal vegetation, land use, current circulation, water turbidity, and ocean
waste dispersion were studied by interpreting ERT-1 and Skylab Imagery with
the help of ground truth collected during overpasses. Based on high-contrast
targets such as piers and roads, the ERTS-1 multispectral scanner was found to
have a resolution of 70-100 m, Skylab's S190A cameras about 20-40 m, and its
218
-------
S190B camera about 10-20 m. Important coastal land-use details can be readily
mapped using Skylab's imagery. On the other hand, the regular 18-day cycle of
ERT-1 allow observation of Important man-made and natural changes and
facilitates collection of ground truth.
Latham, James P. 1973. A Comparison of Interpretation and Photogrammetric
Methods for Delimiting the Mean High Water Position on Tropical Beach.
In: Proceedings of the Am. Soc. of Photogrammetry Fall Convention, Lake
Buena Vista, Florida, p. 803-818.
A determination of the position of mean high water on tropically-influenced
beaches has recently been accomplished both by skilled interpretation of
aerial photography and by the use of photogrammetric methods and tidal gauge
measurements. When a time-series of aerial images are available, physical
evidences of changing water lines can be interpreted to delimit the boundaries
of public and private land. The fact that such a line can be drawn within a
short time has advantages for public officials. More complex and time
consuming methods based on photogrammetric measurement and a year of tidal
gauge readings delay delivering a mapped line, and the data is applied to
beach slope measurements from a photograph taken at only one instant of time
to record a changing beach topography. This study compares the potentials of
these methods illustrated by their applications of the beach in Florida.
Linde, A. F., and T. P. Janisch. 1977. Cover Mapping Wetland Areas with the
Aid of 35 mm Low Altitude Color Photography. In: Wetlands, Ecology,
Values and Impact. Proceedings of the Waubesa Conference of Wetlands held
in Madison, Wisconsin, p. 388.
In 1959 research workers in the Wisconsin Department of Natural Resources
began experimenting with oblique 35 am aerial color photography to record
habitat change and map aquatic vegetation on state owned wetlands. Techniques
and equipment improved with time and experience. A motorized 35 mm Pentax
camera utilizing 250 exposure rolls of film is now available which provides
capabilities for low cost vertical photography in color. The best time to
record species associations and monotypes as delineated by color patterns is
after wetland vegetation has changed color following the first frosts. It was
found that the best altitudes for obliques color photography was between 800'
and 1500'. Vertical photography was accomplished between 1000' and 3000'.
Altitudes above 3000' are not satisfactory since intensity and contrast in
vegetation color patterns decreases with increasing altitude. Cost per acre
of ground coverage for vertical photography varied between 0.1<£ per acre at
3000* altitude to 4.4<^ per acre at 1000' altitude.
Lukens, John E. 1968. Color Aerial Photography for Aquatic Vegetation
Surveys. In: Proceedings of the 5th Symposium on Remote Sensing of
Environment. Ann Arbor, Michigan, p. 441-446.
This paper discusses two applications of color aerial photography that are of
interest to workers in water resources, especially those seeking a rapid and
economical aid for the definition and modification of the ecology of large
bodies of water.
Gross areas of species associations were mapped in water up to 18 feet
deep in the Finger Lakes of New York, using techniques similar to those used
219
-------
in photo-soils mapping. The various factors that must be considered for
photographing, interpreting, and napping submerged and emersed vegetation are
briefly discussed.
A second application of color aerial photography deals with the
assessment of weed control measures for floating aquatics in Chesapeake Bay.
McEwen, R. W., W. J. Kosco, and V. Carter. 1973. Coastal Wetland Mapping.
In: Proceedings of the Am. Soc. of Photogrammetry Fall Meeting. Lake
Buena Vista, Florida, p. 926-937.
The U.S. Geological Survey is conducting a research project in the vicinity of
Sapelo Island, Georgia, to Investigate procedures for interpreting,
delineating, and mapping coastal wetlands using remote sensing and
photogrammetrlc techniques. The study area contains a variety of coastal
marsh conditions, from saline to brackish, and extends from a mainland river
through sea island marshes to the Atlantic Ocean. Orthophotoquads are
prepared at 1:10,000 scale with a format of 2.5 minutes of latitude and 3.75
minutes on longitude. Coastal wetland boundaries and plant species
associations are interpreted and delineated on the orthophoto base. In
addition, the boundaries will be digitized for computer analysis. The primary
objective is to evaluate the accuracy, time and cost for mapping coastal
wetlands. The results of the Investigation should be of value to Federal and
State agencies with responsibilities for mapping or regulating the coastal
zone.
Niedzwiadek, H. A., C. W. Greve, and H. Ross Tywell. 1978. The Wetlands
Analytical Mapping System. In: Proceedings of the 44th Annual Meeting of
the Am. Soc. of Photogrammetry. Washington, D.C. p. 320-328.
The Wetlands Analytical Mapping System III is a computer-based system
developed for the National Wetlands Inventory Project, U.S. Fish and Wildlife
Service, for the purpose of producing a digital record with the classification
and geographic location of wetlands in the United States. The system consists
of a multi-photo analytical block adjustment program with interactive input
and edit capabilities. This program is used for producing the exterior
orientation parameters and other information required as input to the the
digitizing process. The programs for digitizing produce the desired digital
record of the wetlands on a geounit ("square" geographic parcel) by geounit
bases. These programs are interactive and include the editing functions
needed to create topologically valid data files to be incorporated into the
National Wetlands Data Base. The data base software consists of programs to
build and maintain the National Wetlands Data Base as well as provide map
plots and answers to users via a query language package.
Pestrong, Raymond. 1969. Multiband Photos for a Tidal Marsh. Photo. Eng.
35(4)-.463-470.
A variety of multiband imagery, including nine-lens multiband imagery in the
400-900 millimicron range, panchromatic, Ektachrome and Ektachrome-Infrared
photography, has been obtained for a tidelands area in San Francisco Bay. A
technique for comparing their relative utility for specific geomorphic
interpretations has been developed, whereby a subjective form of tracing
analysis may be correlated with a more objective (and quantitative) scheme of
220
-------
selected microdensitometer traverses across the various negative and positive
transparencies. The results suggest that the nine-lens multiband imagery is
excessive, and, that for a similar use as that of the photos studies, could be
reduced to four-lens imagery. The most useful frames are the 550-630
millimicrons bandwidth, the near-infrared, the Ektachrome color transparency,
and the Ektachrome-Infrared transparency. Various utilities are suggested for
each type of imagery, and increased experimentation by geologists with the
microdensitometer is urged.
Pheiffer, W. J., R. A. Linthurst, and J. L. Gallager. 1973. Photographic
Imagery and Spectral Properties of Salt Marsh Vegetation as Indicators of
Canopy Characteristics. In: Proceedings of the Am. Soc. of
Photogrammetry Fall Convention. Orlando, Florida. Part II.
p. 1004-1016.
Primary production is a driving force in the functioning of the salt marsh
ecosystem. An important factor in determining community photosynthesis is the
light adsorptance, reflectance and transmittance by the plant canopy. The net
effect of species difference, canopy architecture, leaf anatomy, pigment
concentrations and edaphic factors were recorded in eitu with a ISCO Model SR
spectroradiometer. Biological and environmental factors were measured and
their relationship to the spectral properties of the stands analyzed.
Spartina alterniflora Loisel., Juncue roemerianua Scheele, Salicornia Virginia
L., Sporbolus virginicus (L) DC stands were studied.
Reese, Frances. 1976. Remote Sensing of Wetlands. A paper written in Civil
and Environmental Engineering 552, University of Wisconsin-Madison,
Madison, Wisconsin. 25 p.
A review of the literature concerning the use of black, and white, color and
color infrared photography, thermal infrared imagery and radar in mapping
wetlands boundaries and species composition.
Reimold, Robert J., and Richard A. Linthurst. 1974. Remote Sensing-
Wetlands. Meeting Preprint 2143, American Society of Civil Engineers,
Nat. Mtg. of Water Resource Engineers. 20 p.
Coastal wetlands present an extremely harsh physical environment in which a
variety of organisms survive. This is, despite their subjection to periodical
wet and dry conditions as a result of tidal inundation and to alternating warm
and cold cycles daily, the coastal wetlands provide one of the most
biologically and ecologically valuable habitats presently known (Reimold and
Linthurst 1973). Estuaries, for example, serve as a nursery ground for marine
organisms by providing food and protection from larger predators. The
wetlands also serve as a physical barrier to protect the coast from severe
erosion during coastal storms and hurricanes.
There exists a variety of scientific methodologies to examine and study
the Importance and complexity of these wetland systems. Fornes and Reimold
(1973), Reimold et al. (1972), and Thompson et al. (1973) have considered
remote sensing technology as applicable to several specific wetland
problems. It will be the purpose of this paper to summarize and examine
multiple uses of remote sensing of wetlands and their potential applications
to similar systems.
221
-------
Robbing, J. Michael, and Marc J. Hershman. 1974. Boundaries of the Coastal
Zone: A Survey of State Laws. Coastal Zone Mgt. J. 1(3):304-331. p.
392-401.
A survey of coastal state legislation reveals several types of statutes
affecting activities occurring in coastal regions: coastal management
statutes, wetland statutes, and shoreline statutes. Each coastal state has
adopted methods to delineate coastal areas or features, whether an entire
coastal zone, a limited feature such as wetlands, or shorelands. Boundary
delineation is done according to linear measurements, political boundaries,
roads and highways, vegetation, elevation, tidal flow, and other factors. An
appendix is provided containing state statutory provisions relating to
boundary delineation techniques.
Scarpace, F. L., R. W. Kiefer, S. L. Wynn, B. Quirk, and G. Friederichs.
1975. Quantitative Photo-Interpretation for Wetland Mapping. In:
Proceedings of the 41st Annual Meeting of the Am. Soc. of
Photogrammetry. Washington, D.C.p. 750-771.
Analytical techniques and interactive computer programs for using color and
color-infrared aerial photographs as a data source for mapping of a wetland in
Wisconsin have been developed.
A portion of a color-infrared transparency (NASA RB-57 photograph) which
contains part of the Sheboygan Marsh wetland system and the Kettle Moraine
interlobate moraine system in Wisconsin has been digitized using a scanning
microdensitometer. This color-infrared transparency was scanned separately
through blue, green and red filters to extract density values from each of the
three film layers, and, through an appropriate transformation, the analytical
dye density of the film was determined for each point in the scanned area.
These analytical dye density values were transformed through sensitoraetrie
calibrations into "equivalent exposure" values which are related to the scene
reflectance. Using conventional photo-interpretation based on ground truth
observations in selected portions of the test site, the rangs of "equivalent
exposure" for each film layer for each resource type were determined using an
interactive computer program. Once the user/interpreter has determined these
values for each vegetative type, the vegetative types can then be
automatically mapped for the entire test site.
A comparison is made between a computer-drawn vegetation map semi-
automatic ally interpreted from a high-altitude, color-infrared photograph and
hand-drawn vegetation map Interpreted from the same photograph (and its stero-
pair) by conventional photo-interpretation techniques using a zoom stereoscope
and light table. The results seem to indicate that digital processing of film
imagery is a cost-effective method of mapping large wetlands.
Scher, J. Scott, and Paul T. Tueller. 1973. Color Aerial Photos for
Marshland. Photo. Eng. 39:489-499.
Color and color-infrared aerial photographs of water fowl habitats were
studied to determine their usefulness for marsh vegetation evaluation.
Attempts were made to determine the optimum film type, time of day, and time
of year for best results. Both color and color-infrared films proved to be
valuable for marsh evaluation. The larger scales (1:1,000) showed
interpretation results with more accuracy than did smaller scales (1:10,000);
222
-------
however, coverage was limited with large-scale photographs. Early morning
photographs were found to be the most interpretable as sun-spot and wave
effects were not prominent. The best time of year to photograph marsh
vegetation was found to be late summer (August-September) when the submerged
and floating plants were at a stage of maximum vegetative development.
Shima, Lurie Jessie. 1973. Wetland Vegetation Mapping Using Aerial, Color
Infrared Photography. M.S. Thesis, The American University. Washington,
D.C. 34 p.
Vegetation maps of dominant plant communities in a freshwater marsh on the
Patuxent River, Maryland were prepared during the spring and fall of 1971-1972
and correlated with low altitude, color infrared aerial photography. Plant
communities present were determined by field surveys, then compared to areas
of homogeneous color on the spring and fall photography.
A tonal signature was determined for several plant communities because of
their unique colors, saturations, and textures. Comparison of photography
made in nearby marshes demonstrated that three of the twelve spring and five
of the fourteen fall vegetation units mapped can be reliably identified.
Color fluctuations which produce a mottled effect on the photography
constitute the color range of a tonal signature. These fluctuations are
primarily caused by a quantitative variation of plant species within the unit
but are also related to the growth cycle and habit, vigor of the plant
species, and environmental conditions which affect the vegetation and turn Che
color of the recorded image.
Changes in the color, saturation, and texture of the spring and fall
photographs indicated plant succession, growth habit, weathering, aging and
vegetative decline.
Shima, L. J., R. R. Anderson, and V. P. Carter. 1976. The Use of Aerial
Color Infrared Photography in Mapping the Vegetation of a Freshwater
Marsh. Chesepeake Science 17(2):72-85.
Spring and fall vegetation maps were prepared from a freshwater marsh on the
Patuxent River, Maryland. Low altitude, color infrared (IR) aerial photos
were correlated with data obtained from field surveys. The vegetation units
mapped refer to areas of homogeneous color on the photos. These areas of
homogeneous color represent species associations or monospecific stands which
produce a distinctive tonal signature.
Color fluctuations within an area having a distinctive tonal signature
are primarily caused by a quantitative variation of plant species but are also
related to the growth habit, vigor of the the plant species, and environmental
conditions which affect the vegetation and in turn the color of the recorded
image. Changes in the color over the growing season reflect plant succession,
and vegetative decline. Tonal signatures of several plant associations were
due to their unique colors, saturations, and textures. Comparison of
photographs made in nearby marshes demonstrated that three of the twelve
spring and five of the fourteen fall vegetation units that were mapped can be
reliably identified.
223
-------
Thompson, Donald E. 1972. Airborne Remote Sensing of Georgia Tidal
Marshes. In: Operational remote sensing: an interactive seminar to
evaluate current capabilities. Am. Soc. of Photogrammetry Fall Meeting.
Falls Church, Virginia, p. 126-130.
Estuarine marshes of Georgia and all other marshes as well are essentially
finite in area. They represent undeveloped land and are subject to pressures
for manmade development of many sorts, especially as population Increases.
Scientists consider the marshes as a primary food production center for the
esturine and continental-shelf marine ecosystem. Marsh grasses are the basis
of the food production system—dead grass falls into the water, is broken up,
washed out with the ebbing tides, and used by living organisms in the water.
To provide decision makers with a factual basis for assessing relative
tidal marsh values, the University of Georgia Marine Institute at Sapelo
Island is engaged in a series of research projects to measure marsh
productivity quantitatively.
Airborne remote sensing of estuarine marshes, coupled with reliable
ground truth data, provides the only feasible, economic means of assessing
actual primary production. Work performed to date represents primary
production measurements on a scale never before attemped. Integration of
ground truth with photograhic and nonphotographic Images represents a first in
tidal marsh ecology.
Whitman, Ruth I., and Kenneth L. Marcallus. 1973. Textural Signatures for
Wetland Vegetation. In: Proceedings of the Am. Soc. of Photogrammetry
Fall Convention. Orlando, Florida. Part II. p. 979-992.
This investigation indicates that unique textural signatures do exist for
specific wetland communities at certain times in the growing season. When
photographs with the proper resolution are obtained, the textural features can
identify the spectral features of the vegetation community seen with lower
resolution mapping data. The development of a matrix of optimum textural
signatures is the goal of this research. Seasonal variations of spectral and
textural features are particularly important when performing a vegetation
analysis of fresh water marshes. This matrix will aid in flight planning,
since expected seasonal variations and resolution requirements can be
established prior to a given flight mission.
Wynn, Sarah L., and Ralph W. Kiefer. 1978. Color and Color Infrared 70 mm
Aerial Photography as a Monitoring Tool for Assessing Vegetation Changes
in a Large Freshwater Wetland. In: Remote Sensing of Earth Resources
Conference. Vol. VII, The University of Tennessee Space Institute.
Tullahoma, Tennessee.
The Environmental Protection Agency is presently funding a three year study of
the impact of siting a two-unit, 1000 megawatt coal-fired generating station
in a 1600 ha wetland. Aerial photography has been obtained on a nearly
monthly basis since 1971 covering the period of construction and a three year
period after the initiation of operation. The 1971-1974 aerial photographs
were 35 mm color and color infrared. The 1975-1978 photographs are 70 mm
color and color infrared at scale ranging from 1:10,000 to 1:40,000. Detailed
vegetation maps have been prepared from an extensive (four year) field survey
effort and from human photo interpretation. The principal impact on the
224
-------
wetland vegetation is due to the extensive leaking of the 200 ha cooling lake
built on a portion of the original wetland. Ground-water influx to the
remaining wetland has increased by a factor of six, floating up and eroding
much of the peat mat. Surface water levels are now consistent throughout the
year. In addition ground-water temperatures are now out-of-synch with the
normal temperature fluctuations by four to eight months. The result of the
above is widespread destruction of the peat mat and accompanying change of
wetland species from shallow water perrenials to more hydrophytic species in
some areas and weedy annuals in others. Both human photo interpretation and
extensive ground sampling data have been used to trace vegetation changes over
time.
WETLANDS—GENERAL INFORMATION
These references contain information on such subjects as wetland plant
succession, wetlands classification, location of the mean high water mark,
thermal alteration of aquatic ecosystems, wetlands as water purifiers,
peatland evolution, the development of wetland soils, and marsh productivity.
Auclair, A. N., A. Bouchard, and J. Pajaczkowski. 1973. Plant Composition
and Species Relations on the Untingdon Marsh, Quebec. Can. J. Bot.
51:1231-2147.
The purpose of this study was to identify significant species relationships
and underlying ecological gradients characteristic of the Huntingdon Marsh,
Quebec. In 1970 one hundred and seven 1-m samples of plant biomass were
obtained from the marsh in conjunction with environmental measurements.
These data were later analyzed using principal-components analysis.
The marsh complex divided unambiguously into emergent aquatic and sedge
meadow communities on the basis of distinct environmental and compositional
differences. Equisetwn fluviatile, Scirpus fluviatilis, Eleoaharie palustris,
and Scirpus validue were major species in the emergent aquatic community.
Respectively, these species dominated 29, 25, 16 and 40% of 51 quadrats on a
dry weight basis. Water depth accounted for almost one-third of the species
variation in this community. Interaction between submerged and floating forms
and competitive exclusion between dominant species explained much of the
remaining species variance.
On a dry weight basis, Carex aquatiHs, Car-ex lacuBtris, Cdlamagroetie
eanaderiBie, and Typha anguetifolia dominated 36, 16, 16 and 11% of the 56
quadrats on the sedge meadow. As a group, Carex spp. dominated 63% of the
quadrats. Disturbance related to chance perturbations, water depth, and the
incidence of fire accounted for much of the variation in this community.
The organization of emergent and sedge meadow communities was discussed
in relation to continuum and community concepts with particular reference to
relative changes in discontinuity of species relationships across the
environmental gradient.
Bay, R. R. 1968. The Hydrology of Several Peat Deposits in Northern
Minnesota, U.S.A. In: Proceedings of the Third International Peat
Congress. Quebec, Canada, p. 215-237.
225
-------
A comprehensive peatland hydrology study has provided data on the climate,
hydrogeology, water table levels, and run-off from forested peat deposits in
northern Minnesota. Ground-water studies identified two types of
hydrogeologic situations—perched bogs, independent of the underground flow
system, and ground-water bogs, which were influenced by storage changes in the
surrounding ground-water basin. Because the water tables are near the surface
in undisturbed bogs, they are Important in peatland hydrology. Bog water
table levels indicated storage opportunity, and their reaction to
precipitation was influenced in part by the type of peat material in the zone
of active fluctuation. Run-off was not evenly distributed. Most of the
annual water yield occurred in spring before June I, while summer and fall
water yields were generally low. Run-off was directly related to water level
in the peat deposits.
Bay, Roger R. 1967. Ground Water and Vegetation in Two Peat Bogs in Northern
Minnesota. Ecology 48(2):308-310.
Plant cover and water quality of bog waters are related to the surrounding
ground-water flow systems in two bogs—one perched above and isolated from the
regional ground-water system, the other nonperched and continuous with the
regional system. The nonperched bog has higher pH, higher specific
conductivity, and greater variety in plant cover than a perched bog.
Beard, Thomas D. 1969. Impact of an Overwinter Drawdown on the Aquatic
Vegetation in Murphy Flowage, Wisconsin. Department of Natural Resources,
Research Report #43. Madison, Wisconsin. 16 p.
A lowering of the water level on Murphy Flowage during the winter of 1967-68
resulted in a significant reduction in the distribution, relative abundance
and acreage of aquatic vegetation.
The five species in greatest dominance before the drawdown were most
affected, and collectively showed a reduction of 181.7 acres in the season
after the drawdown.
Bedford, B. L. 1977. Changes in Wetland Vegetation Associated with Leakage
from the Cooling Lake of a Coal-Fired Power Plant. M.S. Thesis. Univ. of
Wisconsin-Madison, Madison, Wisconsin. 39 p.
This study was undertaken to investigate the possible environmental effects of
a new coal-fired electric generating station on wetland plant communities
adjacent to the facility's cooling lake. The wetland vegetation has changed
quickly and dramatically due to changes in water temperature, water levels,
and water flow, directly or indirectly caused by the presence of the cooling
lake. The previously dominant sedge meadow communities have been replaced by
emergent aquatic species or by annuals. An equilibrium state has not been
reached.
Bedford, B. L., J. H. Zimmerman, and E. H. Zimmerman. 1974. The Wetlands of
Dane County, Wisconsin. Dane County Regional Planning Commission in
cooperation with the Wisconsin Department of Natural Resources. 581 p.
In this survey the emphasis was not on identifying and delineating wetland
areas in the manner of a general inventory but rather on the wetlands, the
surrounding upland, and watershed relationships as a unit, the unit necessary
226
-------
for management planning, this report is first a primer on the wetland
ecosystem. It then applies this information to evaluating the wetlands of
Dane County.
Bernard, John M. 1975. The Life History of Shoots of Carex lacuatris. Can.
J. Bot. 53:256-260.
Most shoots of Carex lacustrne live for about 12-14 months, emerging in
autumn, overwintering as shoots of up to 50 cm in length, and maturing during
the next summer. Others emerge in early spring but both groups die in late
autumn. A third class emerges in late July or August, grow to be over 50 cm
in length, and die in late autumn, living only 2-3 months. Flower initials in
this species begin growth in the September-October period and over winter
while about 1.0 cm in length. The shoots that develop inflorescences are in
general longer, heavier, and have greater basal diameter than those shoots
which do not flower, (tore shoots flower if the water level in the marsh was
high the previous year.
Bradley, W. G. 1972. Standing Crop and Productivity of Marsh Vegetation at
Saratoga Springs, Ca. Research Memorandum, Desert Biome, U.S. I.B.P. RM
72-44. 14 p.
Estimates on standing crop were made by summing the products of average height
of each species times its average percent cover. Biomass of marsh vegetation
was determined by harvest methods.
Data from 1966-67 was included for comparison. Plant communities were
described and arranged according to water availability as follows: Xeric
Shrub with dominant species of Larrea divariaata (creosote bush), Atriplex
hymenelytra (Desert Holly), and 4. parryi (Parry's Saltbush); Phreatophye
vegetation with dominant of Salt Cedar, inkweek, Dietichlie epicata (salt
grass) and Honey Mesquite; Salt Marsh vegetation dominated by Phfagmites (Reed
Grass), salt grass complex and Juncus.
The Xeric shrub communities consist of a single dominant species with two
or three species making up 90% of the total cover and with average cover
ranging from 2-4%. In Phreatophyte communities, found closer to the source of
water and on salt flats seasonally covered by shallow water, 2-4 species make
up 90% of the total cover which averages from 3-12%. The Salt marsh
vegetation overlaps in some species with the preceding communities but shows
an increase in diversity with 5-7 species having 5% or higher frequency, and
average total cover increasing to 60 or 80%. Within the hydric area of the
Spring, Ceratophyllum demerswn and Ruppis marituna are present.
Soil salinity was measured and found to be highest in the Phreatophyte
communities. It is suggested that soil surface salinity limits germination.
Salt grass reproducing by vegetative means does well on these highly saline
soils.
Standing crop estimates indices range from 0.89 for the salt flat
communities to 95 for the bulrushes. The % area utllizied by the various
communities ranges from 25% by open water and 23% by salt grass to 1% by
tamarisk. Production of annuals was low, not surpassing 0.2% cover during
study period. Perennials showed very slow annual growth increments.
Distichlis, Cresea, Phmgmites, Sairpue and Juncue made up most of the
standing crop biomass in the marsh. A peak of 2833 kg/ha for the green-living
parts of species occurred June through September.
227
-------
Bruns, V. P. 1973. Studies on the Control of Reed Canary Grass Along
Irrigation Systems. Agricultural Research Service. U.S. Dept. of
Agriculture. ARS-W-3. 17 p.
Eleven separate experiments on the control of reed canary grass along
irrigation channels were conducted near Toppenish, Washington, on the Wapato
Irrigation Project from 1961 through 1964.
When applied repeatedly amitrole-T at 4 pounds/acre was as effective as
amitrole at 8 or 12 pounds per acre and somewhat superior to dalapon at 15 to
25 pounds per acre in eliminating reed canary grass. However, even after
eight treatments in 3 years, a few shoots still survivied at the water line.
Such survival is a source of rapid reencroachment and spread. Initial fall
applications of the amitroles or dalapon were superior to initial spring
applications. Repeated applications of NH at 4, 6, or 8 pounds per acre did
not suppress the growth satisfactorily in the zone Immediately above the
waterline. None of the herbicides appeared to translocate readily in reed
canary grass.
Repeated applications of dalapon at 5, 10, or 15 pounds per acre did not
suppress reed canary grass effectively along an irrigation channel. Dalapon
at 20 pounds was about equally effective in controlling reed canary grass
whether the herbicide was applied in 40, 80, 160, or 320 gallons of water per
acre.
The effect of dalapon on reed canary grass when the herbicide was applied
at 20 pounds in 100 gallons on water per acre on April 12 was not enhanced by
the addition of kerosene at a rate of 0.5 percent (volume by volume).
Further, the effect of dalapon was enhanced by the addition of 10 different
commercially formulated surfactants at concentrations of 0.06 and 0.125
percent.
The optimum time for initially applying amitrole-T at 4 pounds per acre
in the spring was about May 1 when the reed canary grass was approximately
3 feet high at the water line and still in the preheading stage. To maintain
effective control during the season, two retreatments were necessary when
initial treatments were made before mid-April, whereas only one retreatment
was needed when initial treatments were made between mid-April and late-May.
Under favorable moisture conditions, fall applications of TCS at 100, 130
or 160 pounds per acre nearly eliminated reed canary grass. Seedlings of
desirable grasses the following spring resulted in dense stands. However, the
few surviving plants at the water line were growing profusely, were headed,
and were spreading and lodging in the water by early summer of the 2nd year
after treatment. Under dry weather conditions, the TCA treatments were less
effective in eliminating reed canary grass and no seedlings of desirable
grasses emerged the following spring. The ester of TCA at 12.5, 25 and 50
pounds per acre did not effectively control reed canary grass when applied in
either a water or an oil dilutent. Treatments with paraquat at 2 pounds per
acre controlled reed canary grass more effectively than treatments with diquat
at 2, 4, or 6 pounds per acre. However, none of the treatments was practical
for controlling sizeable infestations of reed canary grass along irrigation
systems. Of the eight persistent herbicides applied to reed canary grass in
late fall, dicamba produced toxicity symptoms most rapidly. However, dicamba
did not completely eliminate the reed canary-grass at the waterline, even at
40 pounds per acre. Only atrazine at 40 to 80 pounds per acre, or isocil at
228
-------
40 pounds per acre, eliminated all reed canary grass within 2 years after
application.
Foliar applications of diuron at 4.8 or 9.6 pounds or bromacil at 2.4
pounds in 300 gallons of water per acre, plus surfactant at the rate of 0.5
percent by volume of water, did not control reed canary grass effectively.
Buma, P. G., and J. C. Day. 1975. Reservoir Induced Plant Community
Changes: A Methodological Explanation. J. of Environ. Mngt. 3:210-250.
Reservoirs induce changes in established patterns of flooding, sedimentation,
and ground-water fluctuations. These changes affect vegetation structure.
This article explores a methodology to describe and map a flood plain
vegetation complex downstream from a reservoir. A classification technique
groups sampling units based on species compositions into plant communities.
Application of the technique to river systems before and after
impoundment would permit monitoring of induced vegetation changes.
Differences in plant community structure could then be related to changes in
physical parameters caused by dams. Simultaneous monitoring of undisturbed
systems in similar environmental conditions is necessary to form a basis of
comparison for systems disturbed by river impoundments. Prediction of the
major modifications attributed to river impoundment and their social, economic
and ecological implications are necessary in comprehensive, integrated river
basin planning.
Cain, Stanley A. 1928. Plant Succession and Ecological History in a Central
Indiana Swamp. 3ot. Gaz. 86:394-401.
This article describes conditions found in a half-drained swamp. All evidence
points to the fact that the vegetation found in this area in the past was that
found in undrained swamps. Lowering the water table caused Sphagnum to
disappear, buttonbush to be restricted to a moat area, and Calamagrosti-s to
spread. Two types of succession progressing from submerged aquatics to Carex
and Calamagrostie and a Typha succession progressing from Typha to upland
climax forest are cited.
Cowardin, Lewis M., and Douglas H. Johnson. 1973. A Preliminary
Classification of Wetland Plant Communities in North Central Minnesota.
Special scientific report—Wildlife No. 168, Washington, D.C., U.S. Dept.
of the Interior. Fish and Wildlife Service, Bureau of Sport Fisheries and
Wildlife, p. 1-33.
A classification of wetland plant communities was developed for a study area
in north-central Minnesota in order to analyze data on waterfowl use of
habitat that were gathered by radio telemetry. The classification employs
features of several earlier classifications in addition to new classes for
bogs and lakeshore communities. Brief descriptions are given for each
community, and the important plant species are listed. Discriminant function
analysis was used for 40 plant species. Seventy-five percent of the stands
studied were classified correctly by this technique. Average probabilities of
assignment to communities were calculated and helped to identify distinct and
poorly defined communities as well as the relationship among communities.
229
-------
Curtis, J. T. The vegetation of Wisconsin: An Ordination of Plant
Communities. The University of Wisconsin Press, Madison, Wisconsin.
657 p.
The Vegetation of Wisconsin is a comprehensive study of the vegetation of that
state. It is based on the study and analysis of data from 1400 stands and
synthesizes these data into an exposition of the relationships of the
vegetation of a large area. Southern forests, northern forests, grasslands,
savanna and shrub communities, and lesser communities are each the subject of
a chapter of this book..
Dix, R. L., and F. E. Smiens. 1967. The Prairie, Meadow and Marsh Vegetation
of Nelson County, North Dakota. Can. J. Botany 45:21-58.
The objectives of the study were to determine the general phytosociological
structure of the native prairies, meadows, and marshes of Nelson County, North
Dakota; to establish relationships between this structure and factors of the
physical environment; and to evaluate the relationships between the vegetation
of Nelson County and True and Mixed prairies. Frequency values and
environmental measurements on soil texture, salinity, pH, and an estimate of
the drainage regime were obtained in 100 stands selected to represent the
vegetational diversity within the county. The drainage regime proved to be
the most important single environmental factor in determining the
vegetation. A phytosociological drainage regime gradient was then established
by assigning indicator values to selected species, and the behaviors of all
species and environmental factors were displayed along this gradient. The
vegetational display was then divided into six units: the uplands into high
prairie, mid-prairie, and low prairie and the lowlands into meadow, marsh, and
cultivated depressions. Each vegetational unit is described. Comparisons are
made between each unit and the related vegetation in surrounding areas. It is
concluded that Nelson County should be considered to be within the
geographical area of the True Prairie, although the frequent occurrence of
western grassland types and western species suggests that the county is within
the tension zone between the True and Mixed prairies.
Frazier, B. E., and G. B. Lee. 1971. Characteristics and Classification of
Three Wisconsin Histosols. In: Proceedings of the Amer. Soil Sci. Soc.
Vol. 35:776-780.
On the basis of morphology, pH, and solubility in Naj^Oi, three Wisconsin
histosols were classified as (1) Fibrist, (2) Memist, and (3) Saprist.
Fiber content was found to be the single most useful characteristic in
the classification of these histosols, and in quantifying various stages of
decomposition and soil formation in histlc materials. Fibric material
consisted of 70% or more fiber as determined on a gravimetric basis, using a
140 mesh sieve to separate fibrous (> 0.1 mm) from non-fibrous « 0.1 mm)
material. The fiber content of hemic material ranged from 35 to 60%; saprlc
material contained < 15% fiber. Sodium pyrophosphate extract color (SPEC) is
a useful parameter in the characterization and classification of his tic
materials.
Results of other analysis showed that the carbon content of the organic
fraction was highest in the Saprist and lowest in the Fibrist. Oxygen and
hydrogen contents were, in general, inversely related to fiber content. Total
230
-------
nitrogen appeared to be dependent on botanical composition of parent plants
and microbial activity of the soil; the relatively high nitrogen content of
certain subsurface layers may be related to illuvial deposition of mobile,
nitrogen-containing substances. Mineral content was likely influenced by
additional colluvial or colian sediments.
Frolick, A. L. 1941. Vegetation on the Peat Lands of Dane County,
Wisconsin. Ecol. tono. 11:118-140.
Ecological studies of the vegetation were conducted on the peat lands of Dane
County, Wisconsin, particularly with respect to the nature of the plant
successions. The secondary plant successions, brought about by a number of
introduced, bio tically controlled factors, namely, artificial drainage,
cutting and grubbing of trees and shrubs, moving, grazing, burning, and
concomitant soil disturbances are emphasized.
Peat lands comprise 52,288 acres or 6.8 percent of the total county area,
nearly all located within the glaciated part of the county. The peat, chiefly
of the water-deposited type, occurs in numerous beds of various sizes.
The natural drainage systems of the peat lands are in an immature stage
of development. Between 1900 and 1926, approximately 40 percent of the peat
land was artificially drained by 18 major projects.
The existing evidence is that the vegetation of much of the peat land at
the time of settlement consisted of two principal types, the Larix consocies
and the Calamagroetis-Carex associes. The Larix consocies is quasi-stable and
accordingly recognized as preclimax.
The primary plant successions which have been of importance on the peat
lands are the bog sere and the hydrosere. The bog sere was the most common
type in the past, but has largely disappeared. Only relicts were found of a
number of the developmental stages. The Larix consocies has been the most
tolerant of the changing environmental conditions and it still occupies a
small acreage of peat land. The primary plant succession is, at present,
almost entirely of the hydrosere type because of more effective drainage. The
Populus-Salix associes is recognized as subclimax to the Queraus-Carya
association although conclusive evidence is lacking as to the true climax.
Introduced biotically controlled disturbance has been so general that
much of the vegetation on the peat lands is now undergoing secondary
succession. These successions are discussed and traced from the Larix
consocies and the Calamagroetis-Carex associes stage of vegetation in relation
to the factors of artificial drainage, cutting of trees, mowing, and
grazing. Specific suggestions are made as to the most economically desirable
ecological treatment of the various vegetational stages. The burning of the
vegetation on wet or frozen peat has little effect on herbaceous vegetation
but will usually prevent the ecesis of shrub and tree species. If the peat is
dry, burning frequently results in various degrees of destruction of both peat
and vegetation. Three degrees of burning are recognized, namely, superficial,
medium and deep. The subseres following burning are discussed.
Other peat disturbances, the most important of which are the digging of
drainage ditches and the consequent formation of ditch banks, play a role in
modifying the vegetation.
231
-------
Gibbons, J. W., and R. R. Sharitz. 1974. Thermal Alteration of Aquatic
Ecosystems. American Scientist 62:660-670.
Heated effluents may function to enrich or to stress an ecosystem, depending
upon the biological feature examined. However, the potential for negative
impact on aquatic environments must not be underestimated. The ultimate
consequences of the sometimes drastic alteration of behavior patterns and
life-history phenomena in the surviving inhabitants of thermal areas have yet
to be assessed. The relatively short time span of thermal field studies has
not allowed thorough understanding of the biological chain reactions that may
take place as physiological and genetic adjustments are made.
Grant, Robert R. Jr., and Ruth Patrick. 1970. Tinicum Marsh as a Water
Purifier. In: Two Studies of Tinicum Marsh. The Conservation
Foundation. Washington, D.C. p. 105-123.
A series of studies were done on Tinicum Marsh to evauate its role in reducing
the pollution of Danby Creek. These studies were designed to determine: (1)
the degree of degradation of Tinicum Marsh by pollution from Danby Creek; (2)
the role of Tinicum Marsh wetlands in the reduction of nitrates and
phosphates in Danby Creek water; (3) the role of Tinicum Marsh wetlands in the
production of oxygen, and the reoxygenation of Danby Creek water; and (4) the
productivity of the wetlands as measured by oxygen produced by a known area of
swampland. Results showed that life in the marsh had been severly injured by
organic pollution entering the marsh. The authors also conclude that the
marsh plays a significant role in improving water quality.
Hanson, Herbert C. 1951. Characteristics of Some Grassland, Marsh and Other
Communities in Western Alaska. Ecol. Mono. 21(4):317-339.
A study of vegetation changes associated with marsh drawdowns at Agassiz
National Wildlife Refuge, Minnesota, revealed that the development of five
types of vegetation on mud flats during the first year was influenced by seed
availability, soil type and moisture, season and duration of drawdown, and the
amount of stranded algal debris. The more an area combined early season
drawdown, rich soil types, slow rates of mud flat drainage, and small amounts
of stranded algae, the greater was the development of emergent aquatics.
In the second year of drawdown, most areas developed greater amounts of
upland and shoreline weeds and fewer emergents. Areas originally exposed
before August of the first year lost emergent cover during the second year,
while the reverse was true of areas exposed later in the first year. Specific
changes were influenced by density and composition of residual vegetation,
soil types, and soil moisture. During longer drawdowns, the soil dried more
completely, and over a 5-yr period nearly solid stands of willow developed.
Upon reflooding, mud flat and shoreline annuals were eliminated and
marshes of cattails, soft-stem bulrush, sedges, spike-rush, willows and
aquatic annuals developed in the first year. Specific development in
subsequent years was determined by the nature of the residual vegetation and
the depth of the restored water. Spike-rush and soft-stem bulrush were
destroyed by flooding with over 15 inches of water in 3 yr and in any
continuously flooded area in 4-5 yr. These species persisted only in
shoreline evaporation zones. Common cattail and sedges were gone from
continuously flooded areas in 4-5 yr and also persisted only in shoreline
232
-------
evaporation zones. "Hybrid" cattail remained unchanged in 25 in. of water
throughout 5 yr of flooding. Two-year-old willows died in 2-4 yr in all
depth, but willows on 4- to 5-yr-drawdown areas were killed only where flooded
with 24 in. for 3-4 yr or 18 in. for 5 yr.
Depending on water depths and cover types, 1- or 2-yr drawdowns at 5- to
10-yr intervals are required to maintain emergent marshes at this refuge.
Stands of hybrid cattail may be an exception.
Sago made outstanding growth and seed production in the first year of
reflooding. Significant changes in soil chemistry and nutrient availability
which probably occurred during drawdown, are suspected to be a contributing
factor to this growth.
Until present limited knowledge of the consequences of drawdown is
enlarged, the technique should be used only for specific purposes with
proper control and study.
Helfgott, T., M. W. Lefor, and W. C. Kennard. 1973. Proceedings of First
Wetlands Conference. The Institute of Water Resources. The University
of Connecticut, Storrs. 95 p.
Papers were presented on wetland soils and geology, inland wetlands and
ground water in eastern Connecticut, and wetland hydrology. They offer
sound information on the types of soils and geologic conditions in which
wetlands are found in a heavily glaciated area and the part wetlands play in
the hydrologic cycle.
Heinselman, M. L. 1970. Landscape Evolution, Peatland Types and the
Environment in the Lake Agassiz Peatlands Natural Area, Minnesota.
Ecol. Mono. 40(2):235-261.
The vegetation and peatland types of the Lake Agassiz Peatlands Natural Area
are related to topography, waterflow patterns, water chemistry, and the
evolution of the landscape as recorded by peat stratigraphy. Eight peatland
types are distinguished: (1) microtrophic swamp, (2) weakly minerotrophic
swamp, (3) string bog and patterned fen, (4) forest island and fen complex;
(5) transitional forested bog, (6) semi-ombrotrophic bog, (7) ombrotrophic
bog (raised bog), and (8) raised bog drain. Consistent differences in pH,
Ca, and Mg were found between waters of contrasting peatland types. These
differences agree with the division of peatland types by degree of mineral
soil water influence (minerotrophy). A general topographic alignment of
vegetation and peatland types agrees with the hypothesis of chemical
controls. Vegetation types often have sharp boundaries related to changes
in water properties, peat surface configuration and paths of waterflow.
Landscape evolution included five phases: (1) recession of Lake
Agassiz about 11,700 yr ago, (2) organic sedimentation of local basins
beginning 11,000 yr ago. Aquatic peats eventually covered 6% of the
substratus; (3) development of fens, marshes and carr during the post-
glacial warm-dry interval, beginning about 8,000 yr ago. These peatlands
built the sedge peats that now cover 46% of the substratum. Paludification
caused water tables to rise, and most water basins were overgrown; (4)
invasion of minerotrophic swamp forests around about 5,000 yr ago as that
climate cooled and precipitation increased. These forests built the basal
forest peats that now cover 48% of the substratum. (5) Capture of part of
the watershed by Northeast Brook about 3,100 yr ago, which caused a water
233
-------
table divide and mineral depletion, initiated sphagnum invasion, and led to
development of the present ombrotrophoic raised bogs. As convexity grew, a
sharp vegetation and chemical gradient developed along the limit of mineral
soil water. Myrtle Lake rose steadily with paludification and now stands
11.8 ft (3.6 m) above the ridge 1 mile (1.6 km) north of the lake. Water
tables rose 10-20 ft (3.0-6.1 m) over much of a 70-square-mile (181 knr)
area. This history does not agree with early concepts of succession which
postulate a trend toward mesophytism with peat accumulation. The only
"direction" here is a possible trend toward landscape diversity.
Heinselraan, M. L. 1965. String Bogs and Other Patterned Organic Terrain
Near Seney, Upper Michigan. Ecology 46(1):185-188.
Treeless string bogs and topographically oriented strips of bog forest have
been discovered near Seney, Michigan, lat 46° 15 N, perhaps the southern
limit of patterned bogs on the North American continent. Patterned ground
has developed through paludif ication of a sandplain dotted with extinct
dunes and sloping about 8 ft/mile. Many peatlands in Michigan, Minnesota,
and Wisconsin have similar slopes and exhibit patterning in various
degrees. Thus the principles that can explain the patterns and bog-forming
processes at Seney may apply to large areas of forested and treeless
peatland. Studies should be directed toward the interrelations between
vegetation, water chemistry, local geology, peatland topography, peat
hydrology, peat accumulation, and physical geomorphic processes.
Heinselman, M. L. 1963. Forest Sites, Bog Processes and Peatland Types in
the Glacial Lake Agassiz Region, Minnesota. Ecol. Mono. 33(4):327-374.
This study was concerned with forest sites and bog processes on the Lake
Agassiz peatlands in northern Minnesota.
The identity of the patterned bogs and fens of this region was
established. Features that clearly mark the Lake Agassiz peatlands as
members of this circumboreal group include string bogs (Strangmoor),
topographically oriented forest islands, and fields of regularly spaced
islands.
The decisive influence of water movement patterns on floris tics and
forest sites was underscored. The key seems to be the degree of isolation
from mineral-influenced ground water. The course of such waters through
bogs is often marked by water-track vegetation types. A tentative
classification of peatland types is proposed.
Theoretical implications are discussed. Neither the processes of bog
expansion nor the patterned bogs and fens of the Lake Agassiz region fit the
classical picture of succession in the Lake States. Conclusions are that:
(1) Few bogs in this region are the result of single successional sequence.
(2) The bog types cannot be regarded as stages in an orderly development
toward mesophytism. (3) Raising of bog surfaces by peat accumulation does
not necessarily mean progression toward mesophytism. Such rises often cause
concurrent rises of the water table and promote site deterioration. (4) The
climax concept does not contribute to understanding bog history in this
region.
234
-------
Isirimah, N. 0., and D. R. Keeney. 1973. Contribution of Developed and
Natural Marshland Soils to Surface and Subsurface Water Quality. Water
Resources Center, University of Wisconsin-Madison, Madison, Wisconsin.
30 p.
Preliminary qualitative estimates of the role of natural and developed
marshland soils as a nitrogen or phosphorus source or sink were obtained by
a limited ground and surface water survey of a marsh adjacent to Lake
Wingra, near Madison, Wisconsin and by laboratory investigations of nitrogen
and phosphorus transformations in soil samples from this marsh and from an
acid bog in northern Wisconsin. Results are presented on the rates and
pathways of these transformations. The results obtained by this
investigtaion indicate that the marsh studied does not act as a significant
nutrient sink. Thus while removal of the marsh by draining or filling might
result in more N and P entering the lake, its presence does not appear to be
a factor in lowering lake productivity. Nutrient input into the lake
probably could be lowered by discharging storm sewers on mineral soil rather
than on the marsh.
Isirimah, N. 0., and D. R. Keeney. 1973. Nitrogen Transformation in
Aerobic and Water Logged Histosols. Soil Science 115(3):123-129.
Wetlands (bogs, marshes, etc.) are a valuable natural refuge for birds,
animals and fish. However, little is known of their value as a nutrient
(particularly N and P) source or sink to the associated surface and ground-
water supplies (Lee L966). In recent years, many of these wetlands have
been drained to make them suitable for agricultural or residential use. The
aerobic conditions and associated higher temperatures accelerate
decomposition of the organic soil matrix, and scattered evidence (Avnimelech
1971, Bentley 1969, Lee 1966) indicates that marshland drainage results in
significant increases in the formation and subsequent leaching of NCU-N to
surface and subsurface waters.
In this work, samples of organic soils (Histosols) representing a wide
range of botanical and chemical composition were subjected to incubation in
the laboratory for 1, 3 and 6 months under water logged or aerobic
conditions to estimate the N balance and N pollution potential of marshland
soils under natural and drained conditions.
Jeglum, John K. 1971. Plant Indicators of pH and Water Levels in Peatlands
at Candle Lake, Saskatchewan. Can. J. Bot. 49:1661-1676.
Quantitative data on vegetation, depth to water level, and pH of both moist
peat and water from 113 stands of peatland near Candle Lake, Saskatchewan,
are used to demonstrate relationships of peatland species to classes of pH
and depth to water level, and to recognize plant indicators for the various
classes. Weighted average and similarity coefficient techniques are used to
estimate pH and depth to water level from total species lists and restricted
lists of important species. Total species lists, combined with either
weighted average or similarity coefficient techniques, yield indices with
the highest correlations with the true values and the lowest standard errors
of estimate. Depth to water level and pH are recognized as two important
environmental correlates with f lor is tic and vegetational variation in
peatlands.
235
-------
Jervis, Robert A. 1963. The Vascular Plants and Plant Communities of Troy
Meadows—a Fresh Water Marsh in Northern New Jersey. The Bulletin—New
Jersey Academy of Science 8(2):1-21.
1. Troy Meadows, a 1,800 acre inland fresh-water marsh, is located in the
old basin of glacial Lake Passaic in the southeastern portion of Morris
County, New Jersey.
2. Its history of use by white man began with the clearing of the land,
formerly swamp forest, and subsequent drainage. Agriculture in the form of
hay production followed, and failed, as drainage became less feasible,
largely because of the subsidence of decaying peat. A retrogression to open
water then must have occurred in places, reinitiating an aquatic succession.
3. The flora was sampled with regard to composition and distribution, using
frequency determination based on 100 evenly spaced grid points.
4. The broad patterns of plant community composition and distribution were
determined by extensive reconnaissance and the use of a cover-abundance
rating for each species in a number of quadrats in stands of each community
type.
5. Water depth, and depth and nature of the substratum were the major
environmental variables investigated.
6. Troy Meadows contains at least 236 species of vascular plants,
representing 67 families; among the best represented of which are the
Cyperaceae, Gramineae, Compositae, Polygonaceae, and Labiatae. A total of
41 species were present at 25% or more of the frequency sampling points,
among which the following displayed the widest distribution: Impatiens
capensie, Boehmeria eyiindrica, Carex stricta, PeltandrvL wirginica, and
Lewvi minor.
1. Ten distinct plant communities were found to repeatedly occur in a
complex mosaic throughout the marsh. These were mapped and the areal cover
of each determined. The Cattail Community, covering almost 50% of the
marsh, was the most Important of these.
8. The following communities appear to be arranged along a decreasing
moisture gradient: Open Aquatic, Cattail, Sedge Swale, Sedge-Shrub and Wet
Meadow and Littoral. Other communities, influenced more in their
distribution by physiographic factors, are the Floodplain and Riverbank
communities which flank the Whippany River, and the Lotic and Streambank
communities found in the water courses which feed the marsh.
9. A complex interaction of influences^-including physiographic, edaphic,
climatic, biotic, and, not the least, anthropogenic—is thought to be
responsible for the present patterns of distribution of the flora and
vegetation. Chief among these interacting influences in Troy Meadows appear
to be: water level and its fluctuations; the texture, organic content and
depth of the soil; alluvial transport and deposition of suspended materials;
the mobility of disseminules of various elements of the flora; the
availability of colonizable substrate and the range of tolerance of each
species to the environmental spectrum encountered in its establishment in
the community; animal utilization of vegetation; the influence of man's
activities; ditching, fire and stream pollution; and chance.
236
-------
Kadlec, John A. 1962. Effects of a Drawdown on a Waterfowl Impoundment.
Ecology 43(2):267-281.
This report covers an evaluation of pilot drawdown on the Backus Lake
flooding project in north-central lower Michigan and its effect on
vegetation, waterfowl, soil, water and bottom fauna. The investigation
included two growing seasons before and one after the temporary drainage
during the summer of 1958.
Keefe, Carolyn W. 1972. Marsh Production: a Summary of the Literature.
Marine Science 16:163-181.
A review has been made of studies of freshwater and saltwater marsh
production. Reasons are discussed for the unusually high production of
these communities when compared with terrestrial communities and
phytoplankton. The role of marsh plants as food for consumers is also
discussed.
Kraper, G. L., and H. F. Dulkbert. 1974. A Biological Survey of Kraft
Slough. The Prairie Naturalist 6(3):35-55.
Kraft Slough is a moderately brackish, semipermanent marsh of approximately
950 acres in western Sargent County, eight miles east of Oakes, North
Dakota. The marsh has been identified as one of 111 tracts in the states
that contain representative bio tic communities in a relatively undisturbed
condition (Kantrud 1973). The purpose of this research was to provide
detailed documentation of the wildlife and wildlife habitats occurring at
Kraft Slough. This study was prompted by a proposed plan to create a
reservoir in the Kraft Slough basin. Under this plan, 20 or more feet of
water would be stored in the basin during the summer months to provide a
water supply for irrigation of the Oakes Area-East Side; a segment of the
Garrison Diversion Unit (Figure 1). The water surface area of the reservoir
would be approximately 1440 acres at full capacity. Several large marshes
occur a few miles east of Kraft Slough in an area that once formed the
eastern arm of glacial Lake Dakota. Approximately 8,000 acres of wetlands
including Burns Slough, Big Slough and Meszaros Slough are located in the
glacial lake basin. Under the proposed plan, these wetlands would be
drained and the basins would form part of the acreage to be irrigated from
the reservoir developed at Kraft Slough (to be named Taayer Reservoir).
At present, ownership of the marsh is divided among state, federal and
private concerns. The North Dakota Game and Fish Department owns 80 acres
(Lake Taayer Game Management Area) and 480 acres are in Federal ownership.
Public lands lie principally in the central marsh while five private
landowners have peripheral holdings.
Lee, G. Fred, Eugene Bentley, and Robel Amundson. 1975. Effects of Marshes
on Water Quality. In: Coupling of Land and Water Systems, A. D.
Hasler, ed. Springer-Verlag, New York. 298 p.
Marshes and other wetlands in which there is a profuse growth of aquatic
plants are common in many parts of the world. Wisconsin contains many
thousands of hectares of marsh vegetation which typically stands from a
half meter to several meters above the normal water elevation during the
237
-------
growing season. Water depths range from a few centimeters to several
meters.
There is a certain flux of nutrients to the marsh from ground waters,
surface flow and direct precipitation and gas exchange. The outflow is
manifest in deposition of materials in the sediments, gas exchange and the
transport in streams draining the marsh. The marsh is a complex hydrologic,
chemical and biochemical system which can transform various elements into
compounds that may improve water quality or have a deleterious effect.
Wetlands are often considered low-value land since in their normal
condition they cannot be used for most agricultural activities or urban
development. Also, some farmers have capitalized on the large amounts of
nutrients stored within the marsh to develop muck farming after draining the
marsh. The drainage of a marsh changes the release of aquatic plant
nutrients. This chapter discusses results from the University of Wisconsin
Water Chemistry Program on effects of marshes on water quality. Data on the
chemical composition of waters discharged from several Wisconsin marshes as
well as studies on the leaching of aquatic plant nutrients from drained
marshes are primarily the results of studies by Bentley (1969) and Amundson
(1970). Both of these students' theses should be consulted for additional
details.
Madson, Carl R., et al. 1975. Wetland Losses Associated with Highway
Construction in Western Minnesota. Paper presented at the 37th Midwest
Fish and Wildlife Conference, Toronto, Ontario. 14 p.
Highway construction in the north-central United States threatens vast
numbers of wetlands. As a part of highway construction, roadside ditches
are often shaped so that runoff water reaching the right-of-way will move to
the nearest creek or river. These ditches provide drainage outlets for
fields and wetlands adjacent to the roadway, resulting in a loss of valuable
wetlands. It is estimated that 99,292 acres of wetlands were drained in 19
counties of western Minnesota as a result of road construction.
Recommendations are made to reduce these losses.
Millar, J. B. 1973. Vegetation Changes in Shallow Marsh Wetlands Under
Improving Moisture Regime. Can. J. Bot. 51:1443-1457.
Changes in species composition and plant cover were studied in relation to
moisture regime over a 10-yr period in 71 shallow marsh wetlands in the
grassland and parkland regions of Saskatchewan. Decreases in density of the
shallow marsh emergents Polygcmum coocineum, carex atherodes, Sooluchloa
feetuaaaea, and ffleocha-ria paluetris occurred with greater-than-normal water
depth at the start of the growing season but 2 or more years of continuous
flooding were required to eliminate emergent cover completely and convert
the wetland to open water. Repeated autumn reflooding also resulted in
complete elimination of emergent species. Changes in species composition
occurred when basins were grazed and as vegetation reestablished after
cultivation but no changes followed mowing or burning. .Alepocurus aequalis,
Beakmannia eysigachne, Glyceria grvcndia, and G. pulchella are designated as
"disturbance" species on the basis of their response to soil-exposing
events. Presence of small amounts of deep marsh emergents in shallow marsh
wetlands is not considered a reliable indicator of wetter moisture regime.
Species composition of rooted submergents in a wetland can be used as an
238
-------
indicator of its moisture regime. Shallow marsh wetlands in basins of
1 acre (0.1 ha) or less experienced little year-long flooding and converted
to open water only under atypical conditions. Larger wetlands required
basin depths in excess of 36 in. (96.4 cm) to have any amount of year-long
flooding and to convert to open water. These basin size and depth criteria
have applications in habitat evaluation by waterfowl managers.
Phillips, John. 1970. Wisconsin Wetland Soils. Wisconsin Department of
Natural Resources. Research Report # 57. Madison, Wisconsin. 24 p.
This report describes wetland soils found in Wisconsin. It addresses soil
morphology, soil relationships in wetland communities, and wetland soil
properties in relation to land use.
Sears, Paul B., and Elsie Janson. 1933. The Rate of Peat Growth in the
Erie Basin. Ecology 14:348-355.
There appears to be a period of maximum compression during the first 15 to
20 years, and after that much slower compression. Comparing the
measurements, such as they are, for the first 70 years with those from peat
certainly between 6,000 and 8,000 years old, it does not appear far from
wrong to speak of a mean rate of peat accumulation of between 20 and 30
years to the inch for the past several thousand years, in the Great Lakes
area. It is of interest to know that the application of this conventional
rate to correlated deposits in Ohio has given a chronology for postglacial
climates which is essentially that of the European periods of Blytt and
Sernander as computer from clay varves by deGeer.
Seddon, B. 1972. Aquatic Macrophytes as Limnological Indicators.
Freshwater Siol. 2:107-130.
Species of submerged and floating-leaved aquatic raacrophytes have been
placed in a series based on their patterns of occurrence in an ordination of
floristic lists. Two chemical parameters from lake water analyses are
correlated with the species assemblage in individual lakes. Trophic
categories are defined on the quantitative chemical characteristics of lake
waters. The range and limiting tolerance of solute content for many aquatic
species are described and related to these trophic categories. Restriction
towards eutrophic conditions is considered as an obligate relationship
reflecting physiological demands. Some dystrophic and oligitrophic species
are shown to have wide tolerance and are thought to be exclueded frm sites
of higher trophic status by competition rather than physiological
limitation.
Sheldon, R. B., and C. W. Boylen. 1975. Maximum Depth Inhabited by Aquatic
Vascular Plants. The American Midland Naturalist 97(1):248-254.
In situ observations of submerged, rooted aquatic plants by a diver equipped
with SCUBA have shown that the maximum depth distribution of a number of
submerged species in a clear freshwater lake (Lake George, New York) is
greater than previously reported. Maximum depth for any species was 12 m
for Elodea canadeneie. Water clarity is sufficient to allow 10% of the
light intensity hitting the surface during midsummer to penetrate to this
depth. The number of submergent species drops linearly from 38 to 1 m to
239
-------
one at 12 m. Data are presented for the maximum depth of occurrence for 28
vascular macrophyte species, and population densitlties of these species at
their preferred and maximum growth depths compared. The effect of several
environmental parameters on depth inhabited by rooted aquatics is discussed.
Sherff, Earl E. 1972. The Vegetation of Skokie Marsh, with Special
Reference to Subterranean Organs and Their Interrelationships. Bot.
Gaz. 53:415-435.
This paper discusses general features of the vegetation at Skokie Marsh.
Ecological factors studies were daily evaporation rate, depth of the water
table, pH of marsh water and soil type. Lastly Scherff discusses
subterranean organs of wetland species and their interrelationships. Root
systems of different species can function in a complementary or competitive
fashion.
Spence, D. H. N., and J. Chrystal. 1970. Photosynthesis and Zonation of
Freshwater Macrophytes. 1. Depth distribution and shade tolerance. New
Phytol. 69:205-215.
The mean, range and standard deviation are given of the depths of water
above the soil surface in which a number of Potomogeton species occur in
Scottish lochs. Sun leaves of these species were produced in unscreened
containers in a glasshouse and their rates of net 02 production were mea-
sured at irradiances of from 1.34 to 7.08 cal/cm /hr in a Warburg apparatus,
using Warburg buffer no. II as bathing solution. Using only leaves of
species for which the rates appeared to be unaffected by buffer solution
during the short experiments, it was shown that the shade tolerance of these
leaves is correlated with the natural depth distribution of the species.
This valid contrast in inherent photosynthethetic response between some deep
water species (e.g., P. praelongus, P. obtusifoliue) and some of shallow
water (e.g., P. polygonifolis) indicates that light may be as important as
substrate or competition in controlling the zonation of freshwater
macrophytea.
Vitt, D. H., and N. G. Slack. 1975. An Analysis of the Vegetation of
Sphagnim-VottLinat&d Kettle Hole Bogs in Relation to Environmental
Gradients. Can. J. Bot. 53:332-359.
Eight Sphagnum-dominated kettle bogs in northern Michigan were analyzed to
elucidate vegetation patterns of both vascular plants and Sphagnum species
in relation to measured bog gradients. Methods of both direct and indirect
gradient analysis, including ordination and cluster analysis, were used.
Community types as delineated in the ordination are discussed, including
species distribution for Sphagnum and vascular plants. Segregation of
community types followed gradients of pH, light, and calcium and magnesium
ion concentrations. Two types of kettle-hole bogs were distinguished
surrounding acid and alkaline bog lakes respectively, each with its own
continuum of community types.
240
-------
Vogel, Richard J. 1973. Effects of Fire on the Plants and Animals of a
Florida Wetland. The American Midland Naturalist 89(2):334-347.
A total of 754 birds were recorded on a portion of a pond shoreline during
63 visits for 4 months following a controlled burn, while 236 birds were
observed on an adjacent and comparable, but unburned, shore line. Only 5 of
the 35 bird species encountered were seen more often on the unburned site.
Fire-induced bird and mammal injury or mortality was unobserved even though
the burn resembled a wildfire. Birds showed no fear of the fire and some
were attracted to the smoking landscape. While some cold-blooded vertebrate
mortality occurred, most herptiles survived, and alligators used the burned
shoreline almost exclusively. Mammal populations of burned and unburned
areas appeared similar 4 months after the fire.
Animal responses are considered related to the fire removal of the
heavy grass mat that otherwise covered the water and soils and the foods
contained therein, and physically impaired new plant growth. Burning also
produced an earlier, more rapid and far more productive growth of wet-
prairie plants.
Vogl, Richard J. 1969. One Hundred and Thirty Years of Plant Succession in
a Southeastern Wisconsin Lowland. Ecology 50(2):248-255.
The post-glacial history of a marl and peat marsh contained evidence that
early hydrarch succession may have been relatively rapid due to higher
plant, as well as invertebrate animal, productivity. Pristine open marsh,
sedge meadow, and wet prairie were held in quasi-equilibirum by alterations
of floods during wet periods and fires during drought. Fires either checked
terrestrial advancement or turned it back to earlier aquatic stages by
organic substrate removal. Recent fire control and continued lowering of
water levels hastened intermediate hydrarch succession by quickly and
directly converting aquatic to terrestrial sites. A peat burn increased
soil pH and soil nutrients, particularly the phosphates, and eliminated
plant competition so that open marsh was immediately invaded by aspen
forest, which will be converted to lowland hardwood forest. Recurring fires
would perpetuate the aspen, but burning decadent aspen forest might
originate true prairie. Although fire is usually catastrophic and
retrogressive, it produced successional stability and even acted as a
successional accelerator in this lowland.
Vogl, Richard J. 1961. The Effects of Fire on a Muskeg in Northern
Wisconsin. J. of Wildlife Mngt. 28(2):317-329.
The effects of prescribed burning on the vegtation of Powell-Flambeau Marsh
were studied during summers of the years 1959 through 1962. This area,
located in north-central Wisconsin, is a hybrid community of open sphagnum
bog or treeless muskeg and sedge meadows. The marsh is being managed to
increase its productivity for wildlife, particularly for geese, ducks,
sharptailed grouse (Pediceetes phasianellue), and white-tailed deer
(Odocoileus virginianus). The burning was analyzed quantitatively using 14
paired stands, one member of each pair being an unburned control adjacent to
the burned area. The vegetation within each stand was sampled using quadrat
frequency studies. To evaluate the effects of fire, all plant species were
divided into groups called Increasers, decreasers, neutrals, invaders, or
241
-------
retreaters, depending on their responses to fire as reflected in average
percent frequency changes. Results indicated that prescribed burning
produces a conversion or retrogression from conifer swamp dominated by trees
to open sphagnum bog or muskeg dominated by sedges and ericaceous shrubs.
The muskeg may be changed further to northern sedge meadow, dominated by
sedges and supporting a minimum of woody vegetation. This sedge meadow
successional stage is considered more desirable than the other types because
it allows the greatest movement, feeding, and nesting of game birds. Fire
also improves game habitat by reducing the "rough" of woody and nonwoody
plants, stimulating new and palatable growth, and increasing fruit and seed
production.
Walker, B. H., and R. T. Coupland. 1968. An Analysis of Vegetation-
Cnvironmnt Relationships in Saskatchewan Sloughs. Can. J. Bot.
46:509-522.
This study examined the relationships between the distribution of herbaceous
species and some of the major environmental factors in sloughs. Frequency
distribution of species was studied in 64 stands. Environmental data,
collected in 40 of these, included weekly readings of water level,
fortnightly readings of pH and total dissolved solids in water, and texture
and organic matter content of topsoil and subsoil. An association table of
24 leading dominants, arranged so that strongly associated species .were
close together, corresponded closely to their observed order along a
moisture gradient. Environmental scalars were constructed to combine data
on pH and total dissolved solids, as well as initial water depth and rate of
water loss. The soil data showed very little association with species
distribution. Synthetic scalars for water regime and water chemistry were
plotted against one another to obtain an arrangement of stands. Plotting
the frequency distribution of the leading dominant species over the
environmental arrangement of stands showed most of the species to be
strongly affected by the water regime and somewhat less affected by
salinity. A few were restricted to a very narrow range of one of these
factors, while others flourished in all segments of the environment. The
relationships suggested by this analysis are largely in agreement with those
suggested by the vegetation analysis alone.
Walker, B. H., and C. F. Wehrhahn. Relationships Between Derived Vegetation
Gradients and Measured Environmental Variables in Saskatchewan Wetlands.
Ecology 52(l):89-95.
Thirty-four relatively undisturbed stands of vegetation in shallow marsh,
non- to slightly saline wetlands in south-central Saskatchewan were examined
with respect to environmental influence on species distribution. Four
environmental gradients account for the bulk of variation in the
vegetation. They are, in decreasing order of Importance, disturbance
(dispite the fact that all stands chosen are relatively undisturbed),
available nutrients, water regime and salinity. The greatest variation in
the data from these stands as a whole is in their salinity, but this is not
reflected in the vegetation. The correlation between water regime and
available nutrients is negative. A number of other factors show minor
correlations with the vegetation and with each other. The methods of
prinicpal components analysis used in this study was a valuable aid in the
242
-------
interpretation of the data. It provides estimates of the proportions of (1)
the variance associated with each principal component (2) the total
variation in the vegetation data that can be assigned to variation in the
environmental measurements.
Weller, Milton, W., and Cecil S. Spatcher. 1965. Role of Habitat in the
Distribution and Abundance of Marsh Birds. Special report No. 43, Dept.
of Zoology and Entomology, Ag. and Home Econ. Expt. Station, Iowa State
Univ. of Science and Technology. Ames, Iowa. 31 p.
Severe drought during the 1950s produced dramatic changes in the vegetation
of midwestern glacial marshes and in the abundance and distribution of marsh
birds. Change in marsh habitat quality and quantity were studied in
relation to bird populations in two small central Iowa marshes, Little Wall
and Goose lakes near Jewell. General observations also were made on several
larger marshes in northwest Iowa near Ruthven. These marshes were nearly
dry in 1956 and became densely vegetated. With gradually rising water
levels, plants flourished, and bird populations increased. Gross cover maps
demonstrated the change in cover-water ratio and interspersion. Population
estimates showed the changes in distribution and density of various species
of marsh birds. During dry periods, only adaptable species such as
redwinged blackbirds, were present. As water levels Increased, densely
vegetated areas were opened up by muskrat cuttings and yellow-headed
blackbirds, coots, piedbilled grebes and least bitterns became established
and increased in numbers. Maximum bird numbers and diversity were reached
when a well-interspersed cover-water ratio of 50:50 occurred. By 1962,
muskrats and high water had eliminated virtually all emergent vegetation
with the result that all species except redwings were eliminated. A similar
pattern occurred on marshes throughout Iowa, and similar changes have been
noted throughout the glacial marsh region during this and previous post-
drought periods.
Habitat changes permitted a measure of habitat preference and
adaptability in several species. Populations shifted from area to area
around the marsh as conditions changed because of muskrat cuttings.
Redwings used shoreward vegetation and were the most tolerant of changing
conditions. They utilized a higher percentage of brush and tree nest sites
over land as emergent vegetation disappeared. Yellow-headed blackbirds were
restricted to robust emergent vegetation standing in water but used only
those areas adjacent to open water. Coots and pied-billed grebes both
nested over water in cover of medium density with sizable adjacent water
openings. Both were quite tolerant of open-marsh stages, and nest losses in
coots at that time were often due to wind damage.
Black terns selected low, natural nest sites or built nests low to the
water in sparse emergent vegetation where they were protected from wave
action. Forster's terns nested in higher sites, such as active muskrat
houses often in open-water areas, or built nests higher above the water than
those of black terns.
The only competition noted was among shoreward nesting redwings and
over water nesting yellowheads. Some Interspecific chases were observed;
yellowhead dominated redwings in the ideal yellowhead habitat, but redwings
occasionally nested in yellow-head territories in small patches of
vegetation not used by yellowheads.
243
-------
Evolution of nest-site selection seems to have been influenced by
general habitat of the ancestral stocks (terrestrial versus aquatic), by
mode of locomotion (perchers, walkers, swimmers, and flyers) and by use of
the major emergents (shoreward or water's edge). The vertical height and
resulting "layers" of vegetation, their robustness and their relationship to
water, influence species use and, thereby, species diversity.
Short-term fluctuations in marsh habitat conditions seem common in
marshes as a result of change in rainfall and subsequent water level
changes. The dry and wet, open stages are the least productive of birds,
while the semi-marsh is ideal. Marsh birds have adapted to these
conditions, and marsh bird populations are characterized by pioneering
ability and mobility. A variety of marsh types and sizes of marshes in a
given area are essential to the preservation of marsh bird diversity.
Marshes are highly productive ecosystems characterized by dramatic short-
term fluctuations. There are periodic invasions of terrestrial flora and
fauna during dry years, while wet years produce a pond or lake-type
community. The viewpoint of marshes as transient serai stages is challenged
because of their duration of life and because of the equally dramatic
changes that may occur in surrounding terrestrial biomes. It is suggested
that a biome-type classification be applied to lakes, marshes, swamps, and
bogs.
Westlake, D. F. 1967. Some Effects of Low-Velocity Currents on the
Metabolism of Aquatic Macrophytes. J. of Expt. Bot. 18:187-205.
A prototype apparatus for making determinations of oxygen exchanges under
controlled conditions of water flow is briefly described and some problems
of technique are discussed in detail. The results include determinations of
the photosynthesis and respirations of Ranunculus pseudofluitane and
Potomogeton peatinatus in natural waters at velocities between 0.02 and
0.05 cm/sec, and some examination of effects of changes in irradiance and
oxygen concentration. Flow was laminar at all velocities. At low velocities
photosynthesis increased rapidly with velocity, but the rate of increase
became less at higher velocities. The size of the effects varied with the
metabolic capacity of the plant. For healthy shoots of R. pseudoflu-itans
the maximum rate of photosynthesis was six times the probable static rate.
These velocities are less than those in open water in streams, or even in
the littoral of lakes, but may be comparable with the velocities within
weed-beds.
Wharton, Charles H. 1970. The Southern River Swamp—a Multiple Use
Environment. Bureau of Business and Economic Research, School of
Businesss Administration, Georgia State University. 42 p.
Water quality data from .both federal and state sources indicate that the
Flint and Alcovy Rivers, with their adjacent swamps, apparently have the
ability to clean pollutants from water. Swamp streams appear to eliminate
human wastes and may remove toxic pesticides which, by way of complex food
chains, might be dangerous to man, perhaps for miles below the source.
Swamps induce deposits of silt and organic materials which become useful to
the biotic community. The swamps have been called "giant kidneys". The
swamp and its stream-channel seem intimately associated functionally, and
appear to form a natural hydrogeobiological water treatment system. The
244
-------
value of the cleansing action of 6 miles and 620 acres of the Flint River
Swamp Is equivalent to sewerage treatment of a city of 50,000. Potentially,
the Alcovy has over three times this ability, worth $990,000 per year. De-
armoring of the Alcovy1s banks and the impairment of normal silt movement by
channelization would cost an estimated $23,232 annually.
White, Keith L. 1962. Shrub-Carrs of Southeastern Wisconsin. Ecology
46(3):286-304.
Shrub-carr vegetation over its full range of variation in southeastern
Wisconsin was quantitatively analyzed and the influence of origin,
environmental factors, and disturbance on the vegetation was Investigated.
Plant composition was sampled in 76 stands distributed over 13 counties. A
list of common shrub-carr species was derived by combining presence with
frequency or intercept data. Salix petiolarie and Cormus stolonifera were
the most common of 38 shrub species. The vegetation had three distinct
layers, an upper dominant shrub layer, an intermediate tall herb, grass and
sedge layer, and a low diminutive herb layer, but there was no
stratification within the shrub layer. The pattern of plant distribution
was very heterogeneous, due to irregularities in the soil surface and to
disturbance. The response of common species to disturbance was indicated by
arranging stands along a disturbance gradient. Most shrub-carr3 la
southeastern Wisconsin originated in the 1930s when shrubs colonized
abandoned mowing meadows. The relative stability of the shrub-carr as a
community appeared to be due to shrub resp rout ing after disturbance.
Lowland forest undoubtedly invades the shrub-carr, although fire may kill
seedlings and saplings and thus retard tree invasion.
ENVIRONMENTAL MONITORING, ORGANIZING, AND POLICYMAKING
These references discuss environmental impact assessment, environmental
decisionmaking, standardization of ecological surveys, quantitative ecology
and impact assessment, regional environmental management, and wetlands
management.
Amir, Shaul. 1976. Land Resources Assessment Framework: a Tool for
Environmental Policy-Making. J. Environ. Mngt. 4:1-13.
Among the many difficulties that public agencies face in their attempt to
develop and implement environmental policy, three are most important:
(a) lack of a simple process for natural resource analysis and evaluation,
(b) lack of a defined land use allocation criteria and (c) lack of data in a
form that enables speedy environmental impact evaluation.
The purpose of this work, which is part of a large research project, is
to suggest an analysis process and a planning framework that can be used in
the development of conservation policy on a national or regional scale. The
evaluation process suggested is composed of four main steps: (a) division
of the area for analysis into land units, (b) assessment of the unit's
conditions and reclamation potentials, (c) identificiation of natural values
and (d) identification of impact areas. In addition to a detailed
explanation of the framework, the paper includes the findings from the
application of the framework to a 300 km case study area in northern
245
-------
Israel. From the initial application of the framework it seems a valuable
tool for land resources planning and management for the development of
conservation policy and its environment of proposed development plans.
Betters, D. R., and J. L. Rubingh. 1978. Suitability Analysis and Wildlife
Classification. J. Environ. Mngt. 7:59-72.
The determination of wildland suitability for various uses and their proper
classification for use is becoming increasingly important to planning. This
article discusses an approach to constructing suitability indices for a
number of various uses while considering many different criteria important
to evaluating use possibilities. A multivariate statistical technique is
then utilized to develop a hierarchical suitability classification. This
classification offers a tool for analyzing the sensitivity of use
suitability to level of classification.
Bisset, R. 1978. Quantification, Decisionmaking and Environmental Impact
Assessment in the United Kingdom. J. Environ. Mngt. 7:43-58.
Increasing concern for the environment in the United Kingdom (U.K.) has led
to demands that major policies and large-scale developments be subject to
detailed impact assessment. A number of different methods have been devised
for this purpose. Some of these involve the quantification and aggregation
of impacts. A method developed in the United States (U.S.) which exhibits
these characteristics is described and discussed. Also, a method which has
been used in the U.K. is considered. It is shown that both these methods
have serious disadvantages. In particular, they mask contentious items in
an assessment thereby avoiding conflict between those in favour of a
proposal and those against. Sections of the community in favour of certain
policies or developments may therefore find these methods useful as a means
of controlling public debate on the merit of proposals. It is concluded
that open discussion of impacts would not be aided by these methods and that
their use in the U.K. should be opposed.
Boyer, Donald E. 1973. A Case History of Remote Sensing Techniques in a
Resource Inventory Process. In: Proceedings of the Am. Soc. of
PhoCogrammetry Fall Convention. Orlando, Florida, p. 473-480.
Using remote sensing techniques and on-site investigations, a multi-
discipline team conducted a physical, biological and visual resource
Inventory of the Oregon Dunes National Recreation Area. Inventory and
interpretative data were developed on the common base of the geomorphic
feature and its processes, further stratified by the plant communities.
Their relationships to seasonal ground water fluctuation, wildlife habitat,
aesthetic value and stages of dunal stabilization or degradation were
established. An integrated report with interpretations was provided to a
planning team as the foundation of the Recreation Development Master Plan.
Bradley, M. D. 1973. Decision Making for Environmental RRsources
Management. J. Environ. Mngt. 1:289-302.
In a world of accelerating scientific and technological advance, of rapid
social and economic appraisal and reappraisal of resources, increasingly
complex choices have to be promptly made. The cost of faulty decisions
246
-------
weighs heavily, however, as society becomes increasingly concerned with
broad environmental side effects, especially where these effects may be
irreversible. Focus upon the forces and resistances acting upon the
decisionmakers when questions of resource management are judged is needed to
understand more completely the factors that contribute to decisions which
ultimately reflect changes in the landscape, and which affect future public
use and enjoyment of the environment.
Bunce, R. G. H., and M. W. Shaw. 1973. A Standardized Procedure for
Ecological Survey. J. Environ. Mngt. 1:239:258.
A large-scale survey of semi-natural woodlands in Britain was carried out in
the summer of 1971. The main aim of this project was to produce an
objective, user-oriented classification of woodland ecosystems for use by
practical conservationists. The shortcomings of traditional methods of
ecological surveys are discussed briefly, and the paper then continues to
examine the underlying principles which should govern the design of a valid
and useful method of survey. The use of multlvariate statistics provides
the ecologist with an effective means of understanding complex variation,
but, unless the data to be analyzed are standardized and obtained by a valid
sampling procedure, the results will be of limited use. The final section
of the paper provides an outline description of the method developed to meet
the above requirements and used in the woodland survey.
Bush, P. W., and W. G. Collins. 1972. The Application of Aerial
Photography to Surveys of Derelict Land in the United Kingdom. In:
Environmental Remote Sensing: Application and Achievements, Eric C.
Barrett and Leonard F. Curtis, eds. p. 169-181.
Local authorities are required to submit annually statistical information
relating to the amount of derelict land which exists in their areas. This
paper outlines one method of acquiring this information in cases where
aerial photography is available, which could be utilized by planners having
a modest knowledge of air-photo interpretation techniques. It briefly
considers both the definition and classification of derelict land, and shows
that air survey has substantial advantages over the traditional field
survey: speed, accuracy, economy and the amount of information that can be
collected in detailed surveys of derelict land.
Carter, Virginia and Doyle G. Smith. 1973. Utilization of Remotely-Sensed
Data in the Management of Inland Wetlands. In: Proceedings of the Am.
Soc. of Photogrammetry Symposium on the Management and Utilization of
Remote Sensing Data. p. 144-158.
Remote sensing provides a powerful tool to meet critical management needs
for inventory and classification of inland wetlands as well as for
evaluation of the wetland role in the hydrologic cycle, identification of
significant wetlands for wildlife preservation, and monitoring of wetland
change. Remotely-sensed data are being presently utilized for wetland
management in the Dismal Swamp (Virginia-North Carolina) and in wetlands of
central and southern Florida.
Congress recently authorized the Department of the Interior to conduct
a comprehensive study to establish the feasibility of preserving and
247
-------
protecting the Great Dismal Swamp. The Dismal Swamp is partly owned by the
Department of the Interior and is of importance to the U.S. Army Corps of
Engineers, the U.S. Department of Agriculture, and numerous state and local
organizations as well. High altitude photography flown by U-2 aircraft can
be used for gross vegetation mapping, boundary determination, and selection
of sites for intensive study. Low-altitude photography is useful for more
detailed mapping. Black and white orthophoto quadrangels currently under
preliminary stages of preparation in the U.S. Geological Survey will provide
up-to-date maps of the Swamp at 1:24,000 scale. ERTS provides the big
picture—the entire Swamp is visible on one ERTS frame—and permits
observation of seasonal change and monitoring of significant ecological
shifts. In southern Florida, ERTS is providing information for water
management in the wetlands north and south of Lake Okeechobee where droughts
place significant demands on water that is also needed for maintenance of
the Everglades National Park. Water level and precipitation data are
collected in near real time by the DCS (Data Collection System). These data
are correlated with ERTS imagery that portrays the area! extent of standing
water for prediction and management of water flow.
Curtis, L. F. 1972. Remote Sensing for Environmental Planning Surveys.
In: Environmental Remote Sensing: Application and Achievements, Eric
C. Barrett and Leonard F. Curtis, eds. Crane and Russak, New York.
p. 89-109.
The remote sensing techniques available for environmental monitoring are
discussed with special reference to remote sensing platforms and sensing
systems. Examples of remote sensing studies using infra-red line-scan and
multiband photography in Britain are described. Particular applications of
infra-red line-scan to shelter-belt studies in rural areas are outlined.
Multiband photograhy is examined in respect of its potential application to
land-use, soil and vegetation studies. Illustrations of the use of image
enhancement by colour additive methods are included, together with examples
of densitometer measurements from multiband photography.
DeGlorla, S. D., S. J. Daus, and R. W. Thomas. 1975. The utilization of
remote sensing data for a multidisciplinary resource inventory and
analysis within a rangeland environment. In: Proceedings of the Am.
Soc. of Photogrammetry Fall Convention. Phoenix, Arizona, p. 640-659.
The Bureau of Land Management (BLH) is charged with the multiple-use
management of National Resource lands which encompass over 130 million
hectares in the eleven western states and Alaska. Due to the vastness of
these lands, the BLM realizes the need to integrate remote sensing
applications technology into their planning system. Various remote sensing
techniques were utilized to produce map products for assessing the
applicability of these techniques to the BLM system. Techniques include
manual analysis of LANDSAT-1 and high-altitude, color-infrared photography,
and the application of discriminant analysis and multi-stage sampling
techniques in a human-machine interactive analysis of single-date LANDSAT-1
digital tape data. Manual analysis of single-date LANDSAT-1 imagery
provided landscape vegetation resource maps. The high-altitude photography
was utilized to produce vegetation-type and "sensitive area" maps for two
BLM Planning Units. Acreage and productivity estimates by major vegetation
248
-------
type were generated using multistage estimates from sampling units on
LANDSAT-1 digital data, high-altitude photography, very large-scale aerial
photography, and systematically collected ground data. The information
generated will provide the BLM with timely and cost-effective information
regarding the vegetative resource.
Dirschl, H. J., and D. L. Dabbs. 1972. The Role of Remote Sensing in
Wildland Ecology and Environmental Impact Studies. In: Proceedings of
the First Canadian Symposium. Ottawa, Ontario, p. 339-344.
Public and industrial interest are increasingly focused on northern
environments which until recently have remained relatively unaffected by the
technological era. Formidable problems of environmental protection result
from major resource developments involving massive environmental
alterations. The magnitude of these problems makes it imperative for
ecologists to critically examine the investigational methods and procedures
which they have traditionally used. In searching for a "common denominator"
in all wildland ecological and environmental impact studies, we are led to
the realization that landform exerts a fundamental control over local
environment. Aerial photographic Imagery provides a bird's-eye view of the
landscape, facilitating the identification and delineation of landform units
with which biological phenomena can be correlated.
Two examples are dicsussed where air photo interpretation has proved to
be the most efficient means of study. Sequential aerial photography in the
Peace-Athabasca Delta has facilitated an understanding of the complex
deltaic processes and the study of plant succession and wildlife habitat
changes resulting from a recently modified river regime. Air photo
interpretation is the basis for extensive renewable resource inventories and
terrain sensitivity studies now underway in the Mackenzie River Valley. The
proposed construction of a natural gas pipeline, and possibly an oil
pipeline, across northwestern Canada and Alaska has prompted ecological
studies of vast Arctic and Subarctic areas. The manner in which air photos
are being used as a working tool in these studies is discussed.
Doiron, Linda N., and Robert T. Wilson. Remote Sensing Techniques for
Wildlife Inventories in the Coastal Marsh. In: Remote Sensing of Earth
Resources Conference. The University of Tennessee Space Institute,
Tullahoma, Tennessee. Vol. III. p. 685-696.
The coastal marshes of Louisiana are recognized as a rich and important
resource which must be managed and used wisely. Management of the muskrat
(Oncatra zibethicue), a major inhabitant of the marsh is an important key in
wise marsh management. In order to manage the muskrat, population and
distribution information is of vital importance. Color infrared aerial
photography can possibly play an important role in providing this necessary
information.
Everhardt, L. L. 1976. Quantitative Ecology and Impact' Assessment. J.
Environ. Mngt. 4:27-70.
Some of the issues of environmental Impact assessment are reviewed from the
point of view of quantitative ecology, and on the assumption that
evaluations are done on a site by site basis. Two approaches are examined
249
-------
in detail, one being the traditional experimental approach and the other one
attempting to predict impacts from data and models. The experimental
approach suffers from the fact that there is no true replication. A
pseudodesign is proposed, employing pre-operational data on a site and a
control area constrasted to post-operational data on both areas, and
substituting replication in time for true replicate areas. Even so, the
limitations of animal census methods and substantial variability make it
doubtful that any but major changes can be detected experimentally.
Predictive techniques, employing methods developed primarily for fisheries
management, may be preferable to the baseline and monitoring concept.
However, these methods have not yet been adequately adapted to the present
purpose and some gaps can be foreseen. One is the lake of knowledge about
stock-recruitment, when the recruits are at very early life history
states. The population regulation problem is identified as a major issue in
impact evaluation. Questions are raised as to the utility of data or
productivity and species diversity, as presently used. It is concluded that
we must take stock of what has been done in impact evaluation, and attempt
to reach a concensus as to future methodology.
Fischer, D. W., and G. S. Davies. 1973. An Approach to Assessing
Environmental Impacts. J. of Environ. Mngt. 1:207-227.
*
The analysis proposed in this paper is designed to permit the assessment of
the likely impact of man's development and management activities on the
environment. The complete assessment consists of four sequential steps:
(1) identification of planned and induced activities, (2) identification of
relevant elements of the environment likely to be altered, (3) evaluation of
Initial and subsequent impacts, and (4) management of beneficial and adverse
environmental impacts that are generated by the planned and induced
activities over time. The emphasis in the paper is upon the identification
and evaluation of environmental impacts because this subject sets the stage
for subsequent management of the environment. Three steps are used to
identify and evaluate environmental feasibility. These are briefly
illustrated with examples from forestry and water management. The
discussion of identification and evaluation assumes that engineering and
economic evaluations are being done simultaneously along with the
environmental Impact analysis. The environmental analysis is to be
accomplished by a small multidisciplinary team which would guide, co-
ordinate and interpret environmental studies being done by various technical
specialists. The paper also includes a brief review of environmental impact
assessment methods developed primarily in the United States.
Fisher, A. C., and J. V. Krutilla. 1974. Valuing Long Run Ecological
Consequences and Irreversibilities. J. of Environ. Econ. and Mngt.
1:96-108.
In this paper we consider the special nature and implications, for economic
theory and policy, of resource uses that involve adverse effects on the
physical environment that are difficult or impossible to reverse.
Distinctions are drawn between reversible and irreversibly activities,
between replaceable and irreplaceable resources. The existence of an
"irreversibility premium" is demonstrated under certain plausible
conditions, including uncertainty and shifting time perspective.
250
-------
Garluskas, A. B. 1975. Conceptual Framework of Environmental Management.
J. Environ. Mngt. 3:185-203.
This paper explores the conceptual framework of environmental management.
Based on ecological principles, environmental management produces the least
environmentally disruptive decision-making path by interdisciplinary
integration of multi-disciplinary knowledge. Environmental management is
the intellectual force that can synthesize specialized views and objectives
and merge them to guide the human society toward a compatible existence with
nature. In its general is tic approach, environmental management visualizes
the whole as well as its parts. Through an environmental management
framework the relationships and Inter dependencies can be viewed and assessed
in total perspective. The recognition that man-made artificial systems
exert a spectrum of stresses on the environment is a prerequisite for the
use of the environmental management framework. In its philosophy
environmental management is holistic, stressing ecological complexity and
interdependency of man and nature. Functionally, it is a businesslike
approach to controlling environmental disruption.
Glenn-Bird, S. J. 1972. Remote Sensing Evaluation of Environmental Factors
Affecting the Developmental Capacity of Inland Lakes. In: Proceedings
of the First Canadian Symposium on Remote Sensing. Ottawa, Ontario.
p. 755-764.
The objectives of this paper are first, to point out the multidisciplinary
aspects of remote sensing, and second to apply them to a specific
investigation concerning the developmental capacity of the inland lakes of
Ontario. Consequently, the paper is presented in five sections, as follows:
1. Multidisciplinary aspects of remote sensing.
2. Purposes and phasing of lake capacity study.
3. Regional uses of remote sensing for lake capacity study.
4. Evaluation of parameters using remote sensors.
5. Specialized studies required in remote sensing.
6. Conclusions and recommendations.
Goldstein, Jon H. 1971. Competition for Wetlands in the Midwest: an
Economic Analysis. Resources for the Future, Inc. Washington, D.C.
110 p.
This study was written to provide a social and economic model to evaluate
the many values of wetlands. Goldstein is concerned with "...private and
social values in the wildlife sectors with the benefits of waterfowl
population maintenance being borne largely by farmers." Goldstein discusses
the costs incurred in draining wetlands as well as the government
agricultural incentives to drain wetlands; wetlands as breeding habitats;
and lastly the distribution of hunting land in Minnesota. This study
addresses the incentive structure for having a more efficient allocation of
wetlands in the central and Mississippi flowages.
Gupta, T. R., and J. H. Foster. 1973. Institutional Framework Affecting
the Use of Inland Wetlands in Massachusetts. The Cooperative Extension
Service, University of Massachusetts. U.S.D.A. and County Extension
Services. 39 p.
251
-------
This study discusses the relationship between wetlands and man as
conditioned primarily by institutional considerations. Emphasis is on the
description of forces affecting wetland usage and the social value of such
lands. Forces studied include the laws governing wetland usage; nature of
ownership, ownership costs such as taxes and goals and plans of the owners
of wetlands. Also considered are the activities of groups such as real
estate dealers; construction firms and the influence of public opinion. The
study is based on the survey of wetlands in fourteen Massachusetts towns and
cities. Communities were selected on the basis of a judgment sample so as
to represent variations in topographical, geological, social and economic
forces as well as the dispersion of different types of wetlands throughout
the state. Forty-five wetland owners located in these communities were
interviewed during the summer of 1971 using a structural questionnaire.
Discussions were arranged with the tax assessors, real estate agents,
contractors, members of conservation commissions in these plus a few
additional towns. Discussions were also held with the officials of the
Massachusetts Department of Natural Resources (DNR) to develop an
understanding of the laws governing wetland usage and problems in their
implementation. Some public hearings dealing with the question of
alteration and/or preservation of wetlands were attended to get a feeling
for public opinion on the matter.
Initial assumptions were:
(1) There is a general attitude of indifference towards wetlands amongst
people in general and wetland owners in particular.
(2) The higher the tax liability on wetland, the greater the chances of its
alteration.
The following discussion is divided into five sections dealing, respectively
with the (1) laws governing wetland usage; (2) pattern and length of wetland
ownership, and the owner's attitudes towards the same; (3) assessments and
taxation of wetlands; (4) owners' opinions on wetland taxes; and (5) some
suggestions for alternative institutional arrangements.
Haslam, S. M. 1973. The Management of British Wetlands. I. Economic and
Amenity Use. J. Environ. Mngt. 1:303-320.
Wetlands are an important aesthetic amenity much enjoyed by many visitors.
They also bear a wide variety of saleable products, including thatching reed
and sedge, marsh hay, litter, craft materials, peat and turf, and can be
profitably let for shooting, fishing, and grazing. Most lowland wetlands
have been managed for their exploitable products for many centuries and can
be maintained with these communities only by the continuation of this
management. The addition of paying visitors should provide sufficient funds
to pay for the management. Intensive recreation does disrupt a habitat, but
many uses can satisfactorily co-exist on a large, or a series of small,
wetlands.
Haslam, S. M. 1973. The Management of British Wetlands. II.
Conservation. J. Environ. Mngt. 1:345-361.
Conservation is management for biological quality. In general, if the
vegetation is managed to the desired end, the appropriate animal life will
follow without special treatment (though macrofauna may need greater freedom
from disturbance). The variables most usually controlled are water regime,
252
-------
cutting, burning, and grazing. Much diversity is obtainable by using
different combinations of these factors, which are the factors under which
most lowland wetland communities have evolved and to which they are
adapted. Examples are given of relations between communities or species and
management variables.
Heyland, J. D. 1972. Vertical Aerial Photography as an Aid in Wildlife
Population Studies. In: Proceedings of the First Canadian Symposium on
Remote Sensing. Ottawa, Ontario, p. 212-236.
The advantages and disadvantages of visual censuses of wildlife populations
are discussed. It is noted that oblique photographs have limited use in
census procedures and are most useful for panoramic, illustrative
purposes. It is suggested that vertical photography provides the best
method of census ing many animal and bird populations. Vertical photography
of the population of Greater Snow geese, during the spring and fall
migration periods along the St. Lawrence River, has made it possible to
accurately census the geese, distinguish young from adults, to separate
family units, to determine ranges of brood sizes and mean broods and to
obtain age ratiios. Experimental vertical photography has shown that
several species of waterfowl, some terrestrial ungulates, narwhal, and
beluga could probably be accurately censused using this technique.
Johnson, B. G. 1974. Developing Data for Environmental Impact Studies. J.
of Tech. Asso. of the Pulp and Paper Industry. 57(9):81-84.
Environmental impact studies leading to the preparation of the Environmental
Report have become a common undertaking associated with almost every private
or governmental activity that could potentially affect the environment.
Considerable effort is being devoted to assuring that the environmental
impact of a proposed development is kept at a minimum. The objective of the
impact study is to develop a sufficient data base to permit a meaningful
assessment of environmental conditions that exist on and in the immediate
vicinity of the proposed site. With these data, the best judgment can be
made relative to the expected impact from construction and operation of the
planned facility prior to undertaking the project. This predictive
information is reviewed by appropriate regulatory agencies and the public
and approval to proceed is given if there is assurance that the environment
will be protected. The impact study also produces the data base to which
subsequent preoperational and operation data can be compared to confirm that
the initial predictions of environmental impact are valid. Consequently,
defining the project scope, implementing the study plan, evaluating the
results, and preparing the Environmental Report are essential steps to sound
environmental planning.
Kirby, C. L., and P. I. Van Eck. 1977. A Basis for Multistage Forest
Inventory in the Boreal Forest Region. In: Fourth Canadian Symposium
on Remote Sensing. Quebec, Canada, p. 72-94.
Developments in the interpretation of LANDSAT imagery, ultra-small and
large-scale aerial photography and their application in a multi-stage
sampling design are presented. Merchantable softwood area determined by
means of computer-assis ted interpretation of a winter LANDSAT scene
253
-------
correlates highly with wood volume estimates obtained from aerial
photographs and ground samples. As a result the amount of sampling required
in succeeding stages is reduced. Ultra small-scale, infrared color aerial
photography is evaluated as a tool for estimating stand volumes and for the
preparation of forest cover and soils maps; it is found to be accurate and
efficient. It may not be used in primary and succeeding stages of the
design, however. Equations for predicting Individual tree diameter and
volume from measures on large-scale aerial photographs are developed. The
test of a multistage sampling design indicates that accurate volume
estimates for large areas may be obtained especially when measurements from
LANDSAT imagery and small and large-scale aerial photographs are used to
provide a dynamic information system.
Larson, Joseph S. 1975. Evaluation Models for Public Management of
Freshwater Wetlands. Proceedngs of the 40th North American Wildlife and
Natural Resources Conference, p. 221-227.
State statutes protecting the public values of freshwater wetlands have been
in effect for nearly 10 years in several northeastern states. Early
versions of statutes gave the responsibility of administering these laws to
a state natural resource agency. As experience has been acquired, the
tendency has been to shift this responsibility to local agencies, such as
Town Conservation Commissions. These are local boards like health and
planning boards, and they are responsible for regulating the use of
wetlands. In some cases they are empowered to acquire land for conservation
purposes.
In Massachusetts alone, 351 separate local commissions and one state
appeal agency are examining requests to alter or destroy wetlands. Many
commissions are buying wetlands to protect their natural values. In each
case value judgments are being made and priorities are being set. There
exist few guidelines for evaluation of freshwater wetlands and as
competition for land in highly urbanized states grows keen the critrla used
to justify protection of a wetland are being examined critically by
developers and natural resource agencies alike.
This report presents highlights of results of a team research effort at
the University of Massachusetts to develop a better basis for decision-
making in wetland preservation and to attach economic values to freshwater
wetlands. Early progress of our work was reported to the Thirty-Sixth North
American Wildlife and Natural Resources Conference (Larson 1971) and this is
a summary of our completion report (Larson 1975).
Lillesand, T. M., and W. P. Tully. 1975. Remote Sensing, Water Quality and
Land Use: From the Obvious to the Insidious. In: Proceedings of the
Am. Soc. of Photogrammetry Fall Convention. Phoenix, Arizona.
p. 582-615.
The influence that land use exerts on water quality ranges from the obvious
to the Insidious. Two case study examples are presented which demonstrate
the utility of remote sensing in monitoring land use and water quality in
"obvious" and "insidious" scenarios, respectively. The former is typified
by a photographic and thermal study of Ononodaga Lake in Syracuse, New
York. The shoreline of this lake is urbanized, industrialized; these
shoreline land uses dominate the water quality of Onondaga Lake, which is
254
-------
highly polluted. The later scenario is exemplified by Chaumont Bay, located
along the eastern Lake Ontario shoreline. Non-point sources of pollution
associated with upland land use characterize this region. In such cases,
improved land use planning and control, in the context of water quality
preservation and restoration, entails increased linkage and synthesis of
land use, water quality and hydrologic data. The role remote sensing can
play in providing this linkage and synthesis is presented conceptually.
Milfred, C. J., D. E. Parker, and G. B. Lee. Remote Sensing for Resource
Management and Flood Plain Delineation. Photo. Eng. 35(10):1059-1063.
Boundaries of rare floods, such as a 100 year recurrence interval flood
which is widely use for planning and regulatory purposes, are ordinarily
plotted by engineering procedures. In this study, flood plain boundaries
were interpreted on (1) panchromatic, and (2) color aerial photographs along
a stream in a glaciated area of southern Wisconsin. The accuracy of these
boundaries was determined by comparison at 29 cross sections with those of
an Intermediate Regional Flood plotted by the U.S. Army Corps of
Engineers. Boundaries on both types of photography agreed with the
engineering boundary at 28% of the cross sections, were within 100 feet at
67% of the cross sections, and within 300 feet at 95% of the cross sections.
Flood plain boundaries were most accurately delineated where physiographic
landforms were well defined. The results indicate that airphoto
interpretation can be a useful tool to delineate flood plain boundaries
where the lack of hydrologic data, time and funds prohibit plotting
boundaries by traditional engineering methods.
Mitnick, B. M., and C. Weiss, Jr. 1974. The Siting Impasse and a Rational
Choice Model of Regulatory Behavior: an Agency for Power Plant
Siting. J. of Environ, and Econ. Mngt. 1:150-171.
Reasons for the current siting impasse, including participatory activism,
regulatory failure, the multiplication of considered interests, and the
"investigative careat" are reviewed, and the existing logic of regulator
behavior, viewed as a rational choice model, is summarized. Three sets of
goals for agency participants are identifieid: personal goals of agency
decisionmakers, organizational goals of the agency as a whole, and goals of
agency clients; and the incentive system of a new administrative agency
structured so as to make satisfaction of these goals contribute toward
informed and Impartial decision-making. Major structural components of the
agency would be a Director, a Corps of Examiners, counsel(s) for special
interest(s) of special merit, and Public Counsel, and a Research and
Information Office. The model is applied to the case of a regional or
state-level agency to handle power plant siting.
Moore, W. C. 1972. Remote Airborne Sensing of Water Pollution: Rideau
River Drainage Basin. In: Proceedings of the First Canadian Symposium
on Remote Sensing. Ottawa, Ontario, p. 211-233.
The Rideau River drainage basin has been identified by the Economic Council
of Canada as one of four basins in eastern Ontario and Quebec significantly
affected by water pollution. In addition the basin's proximity to the
nation's capital helps make it an attractive region for recreation and
255
-------
tourism, but the water quality of the Rideau system is slowly deteriorating
and the aesthetic values of the system are being seriously degraded. To
monitor the situation, an extensive water-sample analysis program is
underway, but much more is required before meaningful and effective action
can, realistically be planned to correct the water pollution problem.
Multispectral remote airborne sensing can rapidly provide accurate, detailed
information for a better understanding of the complex environmental
interrelationships that are so important to water pollution control. It is
also important that airborne sensing Imagery serves as a focus, or a
synthesis, of raultidisciplinary teams on the basis of whole drainge
basins. In this paper, multispectral Imagery from three different reaches
of the Rideau system are examined to evaluate remote airborne sensing
techniques for investigating water pollution in the Rideau River drainage
basin.
Munn, L. C. 1975. Problems in Employing Remote Sensing in Environmental
Impact Studies. In: Proceedings of the Third Canadian Symposium of
Remote Sensing. Edmonton, Alberta, p. 367-370.
There is a growing concern among Canadians for the deterioration in the
quality of our natural environment. The impact of many activities on the
biophysical resources must be understood and considered in all phases of
project planning, development and operation. The planning process is
discussed and it is suggested that the late consideration of environmental
concerns is a major deterrent of the use of remote sensing in impact
assessment. Biophysical processes are interdependent one on the other and
require the skills from a number of disciplines, often requiring a
biophysical team. It is difficult to find qualified research scientists who
understand the construction and operational aspects of a project as well as
remote sensing techniques.
Mutch, W. E. S. 1974. Land Management—An Ecological View. J. Environ.
Mngt. 2:259-267.
The model of land use decision making as a simple sequence of ecologist:
economist:politician-administrator is rejected as both unrealistic and
theoretically unsound. The ecologist generally lacks precise knowledge of
the production functions and of the stability in the systems he confronts;
he is unable to determine what is feasible in management unless and until he
knows the capital and labour resources that might be allocated to the
system. It is characteristic of managed systems that they offer
progressively less multiple use, and that they lose diversity and inherent
stability, as intensity of management increases. Reference is made to the
difficulty and costs of applying artificial controls to replace the lost
checks provided by natural diversity.
Parker, H. Dennison. 1975. Remote Sensing for Western Coal and Oil Shale
Development Planning and Environmental Analysis. In: Remote Sensing,
Energy-Related Studies, T. Nejat Verziroglu, ed. Hemisphere Publishing
Company. Washington, D. C. p. 171-187.
There are two broad categories of application of remote sensing technology
in the development of fossil fuel resources in the western U.S. The first
256
-------
includes pre-constraction site evaluations, land use and usability mapping,
and environmental baseline data acquisition. The second category involves
long-term environmental monitoring. The geographic magnitude of these
developments, particularly when multiple mines or processing plants are
considered on a regional scale, precludes the use of conventional ground-
based analysis techniques. The time frame in which these resources must be
developed also limits the utility of conventional methods. This paper will
discuss both categories of remote sensing applications and the overall role
that remote sensing can play in furthering the national goal of major
dependency on internal sources of energy.
Sievering, Herman, and James Sinopoli. 1976. A Framework for Regional
Environmental Management. J. of Environ. Mngt. 4:141-147.
A framework for environmental decision-making is described in which both
qualitative and quantitative aspects of regional problems can be integrated
into a problem-solving context. The techniques employed in this framework
are computer simulation, games, and vote-trading. The paper concludes that
through this framework: (a) environmental analysts can assess public value
structure goal sets which can be used in the development of regional
simulations, and (b) in turn, the quantitative aspects of the problems will
be more easily communicated to the affected public. A brief description of
the application of the framework is also presented.
Smith, D. W., R. Suttling, D. Stevens, and T. S. Dai. 1975. Plant
Community Age as a Measure of Sensitivity of Ecosystems to
Disturbance. J. of Environ. Mngt. 3:271-28.
The method outlined, based on sound ecological principles, provides an
objective evaluation of the ecological cost of human activities in natural
ecosystems. The "cost" is judged in terms of the potential time needed to
replace destroyed terrestrial vegetation. The approach involves, firstly,
the development of a classification for all terrestrial vegetation types
within a region of study. However, an existing classification may be used,
when necessary, modified to suit the purposes desired. Secondly, after the
major types are identified, they are assigned an ordinal rating which is
proportional to their age of development. Finally, once a study region has
been evaluated the fractional area of each class is multiplied by its rating
and the products are summed for any given unit area, e.g. 2 kilometers.
The later information, which defines the ecological sensitivity of a region
on a unit area basis, can be mapped and thus provide a sound basis for land-
use planning decisions. While the method provides an objective evaluation
of the landscape it is recognized that other criteria must also be
considered in making any comprehensive planning proposals.
Sondhelm, M. W. 1978. A Comprehensive Methodology for Assessing
Environmental Impact. J. Environ. Mngt. 6:27-42.
A methodology for assessing environmental impact is developed and tested.
Advantage of this technique over other methods include: the ability to
evaluate simultaneously a large number of project alternatives; the
capability of incorporating directly a very broad definition of
"environment" in the assessment process; the segregation of the subjective
257
-------
components of the study; the possibility of including direct public
participation in the assessment process; the use of interval or ratio rating
schemes instead of ordinal ones; and the examination of specific potential
impacts in the way(s) deemed most suitable. The methodology was devised in
response to a problem involving whether a dam should be constructed at a
given site; however, it should be noted that the methodology is applicable
to a wide variety of situations.
Terborgh, John. 1975. Faunal Equilibrium and the Design of Wildlife
Preserves. In: Ecological Systems, F. B. Galley and E. Medina, eds.
Springer-Verlag, New York. p. 369-380.
A recurrent theme in science is that the utilitarian harvest from new
theoretical developments comes after some delay and often in unanticipated
quarters. So it can be said of modern island biogeography whose founders,
MacArther and Wilson (1963), perceived the simplicity and analytical
tractability of Isolated ecosystems. Now it is becoming increasingly
apparent that the methods and the way of thinking they developed are
extensible to a much larger range of situations, including the design of
f aunal pr es erv es.
The author's own experience has been largely with birds; while the
arguments presented in this book pertain especially to them, there is no
evident reason why the principles should not apply equally well, with
appropriate modification, to other groups of animals. This presentation is
organized into three sections. First, evidence is considered that tropical
forest bird species, with few exceptions, have very limited dispersal and
colonization abilities in relation to their temperate counterparts. Second,
an examination is made of the kinetics of species loss on a forested Island
cut off from the normal interchange with adjacent forested regions. Last,
the conclusions to be drawn from these results are applied to the problem of
how to optimize the design of fauna! preserves.
Tueller, Paul T., and D. Terry Booth. 1975. Large Scale Photograph for
Erosion Evaluations on Rangeland Water Sheds in the Great Basin. In:
Proceedings of the Am. Soc. of Photogrammetry Fall Convention. Phoenix,
Arizona, p. 708-753.
The practicability of using vertical, aerial photography to inventory
erosion conditions on arid and semiarid range water sheds in the Great Basin
has been determined. Established erosion movement transects resulted in the
assurance that large scale (1:600) 70 mm sequential color photographs in
stereo pairs can be used to detect and inventory soil movement. Soil
surface factors which lend themselves to evaluation of erosion were flow
patterns, wind erosion, litter movement, vesicular horizons, bare ground,
rills and gullies. Ground observations were compared with photographic data
to develop descriptions, keys and guidelines for the interpretation of each
erosion condition. Specific examples of each soil surface factor have been
developed. Photo evaluations on these large scale photographs were found to
be as accurate and less costly than ground techniques. Coats involved in
flight time and interpretation averaged less than $0.025/h ($0.01/acre).
258
-------
Warner, M. L., and D. W. Bromley. 1974. Environmental Impact Analysis: a
Review of Three Methodologies. Technical Report. Institute for
Environmental Studies, Departments of Forestry and Agricultural
Economics, and Water Resources Center. Univ. of Wisconsin-Madison,
Madison, Wisconsin, p. 65.
The National Environmental Policy Act of 1969 (NEPA) required the filing of
environmental impact statements by Federal agencies proposing major actions
significantly affecting the quality of the human environment. This research
suggests bases for the evaluation and further development of methodologies
used to prepare impact statements. Three methodologies are critically
analyzed. There are: the "Leopold approach," suggested by Luna B. Leopold,
et al. (1971), the "Battelle approach," developed at Battelle's Columbus
laboratories for the U.S. Bureau of Reclamation (Dee et al. 1972), and the
"WRC approach," contained in the "Principles and Standards for Planning
Water and Related Land Resources" of the U.S. Water Resources Council
(1973). Specific criteria for methodology evaluation are developed within
the areas of: technical ecological content, practical applicability, and
political utility. These criteria are designed to emphasize a "full-
disclosure law" interpretation of NEPA. The methodologies are examined
using each set of criteria in turn. To provide a more concrete setting for
this analyst, a test case Involving a proposed U.S. Bureau of Reclamation
water resources development project in Southwest Idaho was used. Data
collection consisted of a point-by-point comparison and related desirable
characteristics to each methodology. These data are analyzed for overall
methodological conformance to the criteria to yield conclusions on the
strengths and weaknesses of the methodologies.
Wetland Use in Wisconsin. Present Policies and Regulations. Wisconsin
Department of Natural Resources. Madison, Wisconsin, p. 24.
This report is one of a series of reports written for the Statewide Water
Resources Plan, Visions of Tomorrow. Phase one of the plan was published in
November 1973. The purposes of the Plan are to: (1) describe water resource
management alternatives; (2) examine the tradeoffs associated with each; (3)
solicit people's preferences with respect to the alternatives; (4) present
this information to decision makers; and (5) publish the decision makers'
choices, which in effect become the plan. A previous publication, Wetland
Use in Wisconsin: Historical Perspective and Present Picture, provided a
historical analysis of wetlands and wetland use in Wisconsin. This report
presents an overview of the mosaic of policies, regulations and laws which
apply to these wetlands. Both reports were written to provide background
information needed in consideration of alternative futures, policies and
regulations for the use of the state's wetlands.
Wright, Colin. 1974. Some Political Aspects of Pollution Control J. of
Environ. Econ. and Mngt. 1:173-186.
In this paper some principles of optimal control theory are applied to an
examination of the posible differences that political and economic decisions
making may have in the area of pollution control. The main points are that
(1) Pollution Control Boards (PCB's) may behave as though they place weights
on control and benefit functions that differ from market determined weights,
259
-------
(2) divergencies between political and market weights impose welfare losses,
and (3) given that PCB's may be succeeded by another board the current PCB
may adapt its behavior to counteract or enforce the expected future behavior
of the new PCB.
Wynn, S. L., and 0. L. Loucks. 1975. A Social and Environmental History of
Parfrey's Glen. Trans. Wisconsin Academy of Sciences, Arts and Letters.
53:26-53.
The social and environmental history of Parfrey's Glen, and the evolution of
its management as a natural area, are traced over the past 120 years.
Studies of the Impact of visitors on the Glen, particularly on the
vegetation, were carried out to evaluate recreational carrying capacity. A
procedure using a Disturbance Index was developed to measure the degradation
in vegetation on upland habitats in the Glen, measured between June 1970 and
September 1971. Results indicate that the relatively undisturbed vegetation
near the trails is being degraded rapidly, while areas that are already so
damaged as to preclude further deterioration show no recovery. Results also
suggest that if the Disturbance Index were determined annually it would show
whether the vegetation is recovering under new management practices.
Guidelines for management policies in sensitive natural areas such as
Parfrey's Glen are offered to allow public access but allevaite the impact
of human use.
LAND USE CLASSIFICATION SYSTEMS
Anderson, James R., Ernest E. Hardy, and John T. Roach. 1972. A Land Use
Classification System for Use with Remote-Sensor Data. Geological
Survey Circular #671, update now called Geol. Survey Professional Paper
964.
The framework of a national land use and land cover classification system is
presented for use with remote sensor data. The classification system has
been developed to meet the needs of Federal and State agencies for an up-to-
date overview of land use and land cover throughout the country on a basis
that is uniform in categorization at the more generalized first and second
levels and that will be receptive to data from satellite and aircraft remote
sensors. The proposed system uses the features of existing widely used
classification systems that are amenable to data derived from remote sensing
sources. It is intentionally left open-ended so that Federal, regional,
state, and local agencies can have flexibility in developing more detailed
land use classifications at the third and fourth levels in order to meet
their particular needs while at the same time remaining compatible with each
other and the national system. Revision of the land use classification
system as presented in U.S. Geological Survey Circular 671 was undertaken in
order to incorporate the results of extensive testing and review of the
categorization and definitions.
Doverspike, George E., Frank M. Flynn, and Robert C. Heller. 1965.
Microdensitometer Applied to Land Use Classification. Photo. Eng.
41:294-306.
260
-------
The fundamental acquisition of land use data from aerial color photographs
would be expedited if the process could be automated. However, color
density alone does not seem to offer a solution to differentiate land use on
the photographs. Although aperture size affected density readings, no
improvement in land use discrimination could be ascribed to the aperture
area. Moreover, the geometric shape of the microdensitometer aperture
(circular, slit or square) was of little or no significance. Density
differences in the blue region of the spectrum offered more possibilities in
separating ten land use classes than did the red or green.
Frazier, Bruce E., and Gerhard B. Lee. 1975. Effectiveness of a Computer
Land Use Planning System Utilizing Generalized Data. In: Proceedings
of the Am. Soc. of Photogrammetry Fall Convention. Phoenix, Arizona.
p. 754-777.
Computerized land use planning systems, developed by governmental and
private entities, have proliferated in recent years. This investigation of
the capabilities of one such system for locating linear tracts of land in an
agricultural region (e.g., highway corridors) also evaluates the usefulness
of generalized soil survey information in conjunction with other remotely
sensed data.
Spatial computer models were applied to an area in eastern Wisconsin.
Generalized data included soil associations, land cover, slope, and land use
stored by 1/9 km cells. Models were constructed to represent land use ideas
supported by various sectors of society. Routes were automatically selected
by each model and plotted on a medium intensity soil map to test their
effectiveness in avoiding prime agricultural land.
In an area that was mainly classified in USDA capability classes I and
II (85%), routes selected by the various models included from 46% to 86%
classes I and II land. Other tests indicated improvements in model
performance with increasing length. Subtle differences in models provided
significantly different route locations within 17 km. Major difference in
models showed significantly differnt locations within 6 km.
Kiefer, R. W., and M. L. Robbins. 1972. Computer-Based Land Use
Suitability Maps. Paper presented at the 1972 Am. Soc. of Civil Eng.
Annual and National Environmental Engineering Meeting. Houston,
Texas. 38 p.
A computer-based method for generating land use suitability evaluations for
urbanizing areas is presented. These evaluations are based on an analysis
of the physical characteristics of land, such as topography, soil class,
soil drainage, flood hazard, and depth to bedrock. This method allows a
great deal of flexibility in analysis and provides the ability to look at
land use suitability from a variety of viewpoints or development policies.
The use of this computer-based system to prepare residential land use
suitability from a variety of viewpoints or development policies. The use
of this computer-based system to prepare residential land use suitability
maps for a 59.5 km area containing 5,950 one-hectare data cells is
illustrated. Such a method could be a powerful planning tool when used in
conjunction with an evaluation of the other social, economic, political, and
environmental factors that shape the patterns of urban development.
261
-------
DATA ANALYSIS INFORMATION
References discussing diversity indices, association analysis, crosstab
capability, minitab capability, geographic information systems, data available
from high platform remote sensors, sampling and summarizing data for plant
community classification.
Allen, T. F. H., and J. F. Koonce. 1973. Multivariate Approaches to Algal
Stratagems and Tactics in Systems Analysis of Phytoplankton. Ecology
54(6):1234-1246.
Numerical classifications and principal components ordinations were performed
on species from 57 weekly samples of phytoplankton from Lake Wingra. The data
were considered in absolute and relative terms before and after transformation
to presence/absence and logarithmic quantities. The data were also analyzed,
taking into account growth rates in the samples, by means of a transformation
that replaced the scores of species present by the productivity of the sample
as determined by C uptake/biomass. It is shown that different transformations
can reveal different but biologically meaningful aspects of the data. These
different biological aspects are species similarities based on either short-
term survival expedients in particular environmental circumstances, species
tactics, or long-range growth patterns involving breadth of tolerance and
place in the community, that is, species strategems. Most phytoplankton
species in Lake Wingra adopt one of three stratagems: either ungrazed, slow-
growing and very persistent, or ungrazed, fast-growing and of intermediate
duration, or grazed fast-growing and ephemeral. Tactical information is
relevant to particular systems, while strategic information is needed in
ecosystem comparison and for models applicable to several systems.
Allen, T. F. H., and S. Skagens. 1973. Multivariate Geometry as an Approach
to Algal Community Analysis. Br. Phycol. J. 8:267-287.
Multivariate analyses are put in the context of more usual approaches to
phycological investigations. The intuitive common-sense involved in methods
of ordination, classification and discrimination are emphasized by simple
accounts which avoid Jargon and matrix algebra. Warnings are given that
artifacts result from technique abuses by the naive or over-enthusiastic. An
analysis of a simple periphyton data set is presented as an example of the
approach. Suggestions are made as to situations in phycological
investigations, where the techniques could be appropriate. The discipline is
reprimanded for its neglect of the multlvariate approach.
Bar tell, S. M., T. F. Allen, and J. F. Koonce. Ordination of Community
Structural Dynamics in Lake Wingra Phytoplankton. Memo report for
internal use in the U.S.-I.B.P. Eastern Deciduous Forest Biome Program.
35 p.
We use a multivariate analytic approach to describe compositional changes and
structural dynamics of the Lake Wingra phytoplankton. Principal component
analysis ordinates temporal changes in species composition in samples
collected regularly from March 1970 through August 1973. The analyses allow
us to track a "community particle" as it moves in a species dimensioned space
through time. Community trajectories based upon data averaged over three
262
-------
sample depths (surface, 1, and 2 meters) indicate two compositionally stable
periods (winter and summer) separated by spring and autumnal transition
periods when composition changes rapidly. While some positive correlation
exists, the spring transition is not merely the reversal of the autumn
compositional changes; each transition is distinct. We examine nine
physio chemical parameters for causal mechanisms underlying observed trajectory
behavior. Single parameters are correlated with loadings of samples on
individual axes. More importantly, we correlate community behavior to changes
in a complex environment.
Bryant, N. A., and A. L. Zobrist. 1979. IBIS: A Geographic Information
System Based on Digital Images Processing and Image Roster Data Type.
In: Symposium Proceedings Machine Processing of Remotely Sensed Data,
LARS. West Lafayette, Indiana.p.69-73.
There is a pressing need for geographic information systems which can manage
spatially-referenced data, that perform certain types of spatially-oriented
processing, and that are current and comprehensive. Polygon overlay and grid
cell information systems access data for selected areas, but their data files
are time consuming to generate and frequently costly to process. Updating of
land use changes for such systems may become prohibitively expensive. In
response to the present dilemma, a system is presented that makes use of
digital image processing techniques to interface existing geocoded data sets
and information management systems with thematic maps and remotely sensed
imagery. The basic premise is that geocoded data set can be referenced to a
roster scan that is equivalent to a grid cell data set.
Several technical problems have been overcome to achieve a workable
system. First, digital image file handling, Image manipulation, and image
processing capabilities must be provided. Second, image data must be
registered or indexed to spatially-referenced tabular data so that processing
steps which involve both types can operate. Third, a data interface must be
provided between the different data types so that the results of processing
can be represented. Finally, image processing analogs must be developed for
existing geo-base file computational steps (e.g., overlay, aggregation, cross
tabulation, etc.). The system is now in use on a test basis.
Grossman, John S., Roger L. Kaesler, and John Cairns, Jr. 1974. The Use of
Cluster Analysis in the Assessment of Spills of Hazardous Materials. The
Am. Midland Naturalist 92(1):94-114.
The macrobenthic community of the Clinch River, near Carbo, Virginia, has
twice been subjected to acute stress caused by major industrial spills from a
power plant. The first spill, which resulted in a high pH shock, was from a
fly-ash retaining pond in 1967. The second was an acid spill in 1970 with
consequent low pH stock. Stream surveys were made in 1969, 1970 and 1971.
This paper reports the results of Q-raode cluster analysis of presence-absence
data on the total aquatic insect fauna, several orders of insects considered
separately, and Gastropoda from those surveys.
Recovery from the effects of the fly-ash spill by all elements of the
fauna studied except the Gastropoda was well underway by the summer of 1969.
Nevertheless, the insect fauna in samples from the area affected by the spill
was still different from that in unpolluted reaches of the stream although it
was not possible to discriminate between remnant effects of the spills and
263
-------
chronic stress due to the day-to-day operation of the power plant. The spill
of acid in 1970 eliminated many elements of the fauna from about 30 km of the
river. Again, by the end of the summer recovery was well underway for all
groups, except Gastropoda. Cluster anaysis was particularly useful in
determining the effects of type of substrate, time of sampling, longitudinal
succession, and flooding on the composition of the macrobenthic community. It
is suggested that one effect of flooding may be to make the fauna more
monogeneous so there is a more nearly equal distribution of macrobenthic
organisms among the stations from which samples were collected.
Gauch, Hugh G. Jr., 1973. The Relationship Between Sample Similarity and
Ecological Distance. Ecology 54(3):618-622.
Similarity measures for samples from natural communities show a complex,
curvilinear decrease with increasing separation of samples along environmental
gradients. The form of this decrease for samples without sampling errors has
been analyzed and found to be a complement of a non-standardized error
function for percentage similarity, and similar functions for coefficient of
community and Euclidean distance. For samples affected by sampling error,
altered and somewhat flattened curves result. These relationships are
Important in many ordination techniques. In particular it is demonstrated
that Bray-Curtis ordination can be improved by application of the inverse
function transforms with moderate beta diversity (up to 5 half-changes). Such
transforms produce similarity measures which are linear with respect to
separation along the gradient, up to the point beyond which samples have
practically no species in common and similarity measures of any sort are
consequently meaningless.
Gilmer, David S., Steven E. Miller, and Lewis M. Cowandin. 1973. Analysis of
Radiotracking Data Using Digitized Habitat Maps. J. Wildlife Mngt.
37(3):404-409.
A method is described that provides a rapid and accurate analysis of habitat
used by radio equipped animals. The digitizer (basically an X-Y plotter in
reverse) converts maps into digital formats describing each habitat unit as a
polygon that closely approximates the actual shape of the unit. The
coordinates of each polygon are then stored on magnetic tape. Habitat
classification data and other information are coded and combined with the
proper polygon coordinates. This results in one file containing all habitat
data. A computer program with inputs of tracking data and habitat data
provided a listing of the habitat used by the animals studies. Analysis of
habitat used by radio-equipped ducks is demonstrated using this method.
Green, Roger H. 1974. Multivariate Niche Analysis with Temporally Varying
Environmental Factors. Ecology 55(l):73-83. ,
Data consisting of samples of species' presences in association with
measurements on a set of environmental variables can be used to determine
environmental factors separating the species. If the multiple discriminant
model is modified by a covariance extraction of time effects applied to the
within-species, and total deviation squares and cross products matrices prior
to the discriminant analysis, then temporally varying environmental parameters
can be Included. If the distribution of sampling in space is consistent over
264
-------
time, then factors separating the species in time, as well as in space, can be
determined; if it is not, then separation in space can still be determined
even if the samples were collected at different times. The multiple
discriminant model is analogous to the Hutchinsonian niche model: its use is
illustrated with artificial data, and it is then applied to data from a
benthic stream community to demonstrate heterogeneity of niche sizes, and
separation of species' niches in space and in time by different environmental
factors related to substrate type and water depth. Trophically similar
species are more environmentally separated than are trophically different
species; the separation is spatial for herbivore-detritivores, and temporal
for carnivores.
Green, Roger H. 1971. A Multivariate Statistical Approach to the Hutchison
Niche: Bivalve Molluscs of Central Canada. Ecology 52(4) :543-556.
The use of multiple discriminant analysis to identify the significant and
independent ecological factors separating species distributions is proposed
and discussed. Such an analysis was performed on 345 samples containing a
total of 10 bivalve molluscs species from 32 lakes in Manitoba, Ontario, and
Saskatchewan. Measurements of nine ecological parameters were associated with
each sample. Five discriminant functions account for 95% of the among-species
variance, and 40 of the 5 are ecologically interpretable. Three of these,
accounting for 80% of the among-species variance, are interpreted as bases of
trophic, rather than physical or chemical, separation. There is separation of
species on each discriminant function. The use of discriminant score to
classify lakes with maximum relevance to species distributions is demonstrated
and discussed. A generally applicable measure of environmental heterogeneity
based upon this type of analysis is proposed. The value of this type of
analysis in quantifying ecological concepts derived from the Hutchlnson niche
model is discussed. An example is given of a reduced available niche
resulting in the loss of two species, smaller realized niches for the
remaining species, and greater niche overlap.
Hughes, Roger N., D. L. Peer, and K. H. Mann. 1972. Use of Multivariate
Analysis to Identify Functional Components of the Benthos in St.
Margaret's Bay, Nova Scotia. Limnology and Oceanography 17(1):111-121.
Trends in the frequencies of occurrence of polychaetes and echinoderms in St.
Margaret's Bay, Nova Scotia, were isolated by principal components analysis
and cluster analysis. Five principal components accounted for 70% of the
total variance, three of them (46% of the variance) associated with sediment
type, and one (16% of the variance) with distance from the head of the bay,
probably with tidal water movements. The largest component of variance (23%)
was due to the difference between fauna from clay (characterized by Pectinaria
hyperborea and Synapta sp.) and fauna from more heterogeneous, coarser
sediments. It was concluded that the large, contiguous area of soft mud in
the deep part of the bay supports a reasonably well-defined, integrated
community.
Hughes, R. N., and M. L. H. Thomas. 1971. Classification and Ordination of
Benthic Samples from Bedeque Bay, an Estuary in Prince Edward Island,
Canada. J. on Life in Oceans and Coastal Water 10:227-235.
265
-------
An attempt was made to Identify the causes of the distribution of benthos
within Bedeque Bay using multivariate techniques programmed for the
computer. Both classification by a hierarchical cluster analysis, and
ordination by principal components analysis suggested that a large proportion
of the variance in the data was directly or indirectly correlated with a
salinity gradient. Classification divided the species into two main groups:
(a) in the upper half of the estuary where lower salinities and larger
salinity fluctuations occurred, and (b) in the lower half of the estuary with
a more stable salinity regime. The group b species were further subdivided
into those preferring soft mud and those preferring sandier sediments. The
group a species were divided into a well-developed oyster association and
various sub-groups less strongly associated with oysters. Five principal
components were required to account for 50% of the variance in the data. The
first axis accounted for 20% of the variance and was shown by a non-parametric
test to be correlated with the salinity gradient. Axes II to V could not be
interpreted, but possibly represented complex species interactions. By
providing hard substrates and altering the nature of the sediment, oysters and
mussels produced conditions suitable for many other species.
Nieraann, Bernard, J. Jr., X. A. Bonilla, S. R. Brun, and R. A. Rose. 1975.
Rural Landscape Assessment: A Comparative Evaluation of High Platform
Remote Sensors. Dept. of Interior, BOR and College of Agricultural and
Life Sciences, University of Wisconsin-Madison, Madison, Wisconsin.
243 p.
The investigation consisted of comparing and evaluating high altitude color
infrared photography (1:120,000) LANDSAT 1 Paper Products (1:250,000) and
LANDSAT 1 false color enhanced Images with conventional resource assessment
results. The comparison included the use of computer assisted geographical
sampling and cluster analysis techniques. The comparative results indicate
that high altitude color infrared photography is quite comparable with
conventional assessment methods both in time and replicability. LANDSAT
results compared with conventional methods are not as effective but much less
time is required to obtain results. Good results were obtained from LANDSAT
in the measurement of river character.
Ohman, Lewis F., and Robert R. Ream. 1971. Wilderness Ecology: a Method of
Sampling and Summarizing Data for Plant Community Classification. N.
Cent. Forest Exp. Sta. St. Paul, Minnesota. 14 p.
Presents a flexible sampling scheme that researchers and land managers may use
in surveying and classifying plant communities of forest lands. Includes
methods, data sheets, and computer summarization printouts.
Ryan, T. A., B. L. Joiner, and B. F. Ryan. 1976. MINITAB Student Handbook.
Duxbury Press, North Scituate, Massachusettes. 341 p.
\
MINITAB: Student Handbook is used with Minltab, a computing system developed
to relieve students of the computational drudgery usually associated with
statistics. Minitab includes descriptive statistics, simulation, binomial and
polynomial distributions, the normal distributions, one sample confidence
intervals and tests for population means, comparing two means, correlation and
266
-------
regression, analyses of variance, chi-square tests and contingency tables, and
non-parametric statistics.
Swan, J. M. A. 1970. An Examination of Some Ordination Problems by Use of
Simulated Vegetational Data. Ecology 51(1):89-102.
Hypothetical vegetation models were made to simulate numerical changes in
species populations along a single environmental gradient. A single
ordination procedure was evaluated by its ability to detect the ecological
information in the hypothetical models. The procedure was successful when the
data were drawn from a short length of the gradient but became progressively
less so as longer lengths of the environmental gradients were included in the
data. This parallels an increase in the number of stands from which each
species is absent in the total data set. Zero values appear to mask
ecological information, and an intuitive method of assigning "degree of
absence" values to the data is described. After this adjustment, ordination
patterns were easier to interpret because ecological information was
concentrated in fewer axes.
Schubert, J. S., and J. Thie, and D. Gierman. 1977. Computer Processing of
LANDSAT Data as a Means of Mapping Land Use for the Canada Land
Inventory. In: The Fourth Canadian Symposium. Quebec, Canada, p. 268-
281.
Monitoring of land use activities is essential for the design and modification
of land use policies, plans and regulations. To facilitate the measurement of
land use changes, the Lands Directorate, Environment, Canada, is supporting
and carrying out research to assess the usefulness of satellite remote sensing
for mapping and updating Canada Land Inventory land use information.
Methods for computer processing of LANDSAT data were compared for
identifying areas in northern Alberta where changes in land use have
occurred. The source data were land use maps generated from data stored in
the Canada Geographic Information Systems. Surface cover classes were
separated on autumn LANDSAT data. These classes related well to land uses as
defined by CLI for the stored source data. Recent changes in land use were
observed during field study and were successfully classified and mapped using
computer-processed LANDSAT data.
Visual classification of images generated by computer was far superior to
the three computer classification methods tested: supervised and non-
supervised interactive methods and a new automatic method implemented for this
study at Lands Directorate. While the computer methods were not significantly
different in classification accuracy, the new automatic method was least
expensive.
Seigal, Sidney. 1956. Non-Parametric Statistics for the Behavioral
Sciences. McGraw-Hill Book Company, New York. 312 p.
This is a reference book on the use of non-parametric statistics. It covers
the one sample case, the case of two related samples, the case of two
independent samples, the case of k related samples, the case of k independent
samples and measures of correlation and their significance.
267
-------
Wishart, D. 1970. Clustan IA User Manual. Computer. Laboratory, University
of St. Andrews, St. Andrews, Scotland. 118 p.
This is a manual which describes a set of Fortran IV programs designed for the
collective study of several methods of cluster analysis and other raultivariate
procedures. The emphasis throughout the development of this package has been
to simplify program specifications so that routines are easy to use but still
offer many non-standard options.
White, E. J., and D. K. Lindley. 1976. The Reduction of Climatological Data
for Ecological Purposes: a Preliminary Analysis. J. of Environ. Mngt.
4:161-182.
Principal component analyses were carried out on records from the Moor House
Climatological Station, as a preliminary to analysing data from many stations
distributed over the United Kingdom, and, subsequently, to relating climate to
plant response. The use of daily data, with the Penman estimate of
evapotranspiration added, was compared with the use of 5-daily, monthly and
quarterly means, for 1960 to 1969 inclusive. The first five components
generally expressed temperature or energy, dampness, windiness, snowfall and
windiness respectively, and acounted for 81 to 93% of variability. It was
considered that quarterly means were a suitable basis for future
calculations. The stability of the analyses resulting from different sets was
compared. Selection of the variables was discussed, and while it is possible
to select variables by the importance they have as sources of variation with
the meteorological data, their relative importance in explaining response by
any one plant or animal will not be known until such response has been
measured and relationships with meteorological variables have been assessed.
MISCELLANEOUS
Baldwin, Helene L., and C. L. McGuinness. 1963. A Primer on Ground Water.
U.S. Dept. of Interior, U.S. Geological Survey, N.O. G.S. 64-160. 25 p.
This primer defines ground water and where and how it is found, tells how
ground-water quality is defined, and describes wells and their Impact on
ground water. Lastly it discusses managing water resources. This primer is a
good place to start for persons who know nothing about hydrology.
Bell, David T. 1974. Tree Stratum Composition and Distribution in the
Strearns ide Forest. The Am. Midland Nat. 92(l):35-46.
The woody vegetation of the streamside forest in Robert Allerton Park, Piatt
Co., Illinois, is described in relation ot the distribution of river level
frequencies of the Sangamon River. The habitats most frequently flooded are
dominated by Acer sacahavimm. With decreasing flooding frequency, dominance
is transferred to Celtis ocaidentalis and Queraua imbrnearia. The areas
experiencing no flooding are dominated by Q. alba. Changes in the
vegetatlonal structure at elevational increments of 0.304 m (1 ft) are
discussed. The principle that communities change gradually along
environmental gradients is Illustrated in a vertical elevation of less than
4 m.
268
-------
Bormann, F. H., T. G. Siciama, G. E. Likens, and R. H. Whitaker. 1970. The
Hubbard Brook Ecosystem Study: Composition and Dynamics of the Tree
Stratum. Ecol. Monographs 40(4):373-380.
The synecology of tree species was studied in a mature second-growth forest in
the Hubbard Brook ecosystem. The forest, on a 13-ha undisturbed watershed
ecosystem covering a 245-m range of elevation, has a basal area of about 23
nrhs~ . Dominance is shared by Acer 8aaahartunt Fague grandifolia, and Betula
alleghanieneiB. Direct gradient analysis and regression analysis indicated a
strong response in both stand and species characteristics to an elevational
complex gradient. Sugar maple shows a decreasing trend; balsam fir, paper
birch, and mountain ash show increasing trends. Beech, red spruce, mountain
maple, and striped maple show intermediate patterns. Seedlings and saplings
respond to the elevational gradient as do larger trees; however, the behavior
of trees, seedlings and saplings of the same species is clearly different.
The Hubbard Brook ecosystem is located in relation to the vegetational
zonation systems of earlier authors. The only generally agreed upon
vegetational boundary, ca. 760 m (2,500 ft), is accounted for by a steepened
rate of environmental change in the vicinity of that elevation. Various lines
of evidence indicate that the present second-growth forest at Hubbard Brook
approximates old-age mature northern hardwood forest. Therefore, the
biogeocheraical, productivity, and ecological data obtained from this study are
representative of a mature ecological system in dynamic balance with regional
and local controlling factors, e.g., climate, geology, and topography.
Brown, Jerram L. 1964. The Evolution of Diversity in Avian Territorial
Systems. The Wilson Bui. 76(2):160-169.
What are the conditions which facilitate or hinder the evolution of
territoriality? No generally accepted solution to this problem has yet been
found—perhaps because too specific an answer has been sought for too general
a question. Instead, the diversity of systems of territorial and other
aggressive behavior has come to be well appreciated, as evidenced in recent
reviews of territoriality (e.g., Kuroda 1960, Carpenter 1958, Hinde 1956) and
the impossibility of providing a specific answer applicable to all types of
territoriality is now realized.
Arguments continue, however, over which selection pressures are the
primary factors influencing the development of certain types of
territoriality. This continuing dialogue can be observed in the recent
contributions bearing on the "function" of territoriality by Peters (1962),
Wynne-Edwards (1962), Kuroda (I960), Kalela (1958), Stenger (1958), and
others.
The present paper offers a new orientation to the problem by presenting a
general theory for the evolution of territorially with special reference to
its diversity among species. Since most of the previous theories have already
been shown to be untenable or severly limited (see especially Carpenter 1958,
Tingergen 1957, and Hinde 1956, for criticism of them), little attention will
be given to them here. '
Chatterton, W. A., J. L. Clapp, E. F. Epstein, and B. Niemann, Jr. 1977.
Incremental Implementation of a Modern Multi-Purpose Cadastral System.
In: Proceedings of the 43rd Annual Meeting of the Am. Soc. of
Photogrammetry. Washington, D.C. p. 694-709.
269
-------
An examination of the land information situation in the United States
indicates that efforts to generate an adequate system exists in all stages, at
all levels of government in a variety of agencies in both the public, semi-
public, and private sector. One dominant charateristic of their ongoing
efforts is that they tend to serve a single purpose and have not developed
adequate institutional and conceptual mechanisms to take advantage of the
common spatial base required for geographically related land information.
The development of multi-purpose cadastres which will meet the land
information needs must be based upon a geometric framework adequate to meet
requirements at the individual parcel level. This type of framework is
essential (1) because these requirements are the most demanding with regards
to scale, accuracy, and to the representation of proprietary interests; and
(2) because it is possible to aggregate information from the detailed level to
the policy level but not the reverse.
In order to Implement modern multi-purpose cadastral systems, incremental
approaches which are constrained by the broad object must be developed and
followed. These approaches must reflect that a multi-purpose cadastral system
is a dynamic mechanism which is characterized by incremental change in the
existing system. The goal is a sequence of incremental changes over an
extended period of time which will provide maximum service as it evolves and
yet preserves future options.
Curtis, John T. 1959. The Vegetation of Wisconsin. University of Wisconsin
Press. Madison, Wisconsin. 657 p.
This is a definitive study which establishes the geographical limits and
species compositions of the vegetation of Wisconsin as well as describing, as
much as possible, the environmental relations of the vegetation communities
existing within this state. Although the information presented on forest
communities is unsurpassed, the information about wetlands is limited.
Dobbin, James A. 1978. Interpretation of Satellite and Aircraft Imagery for
Planning/Design and Management of Marine Parks and Reserves. In:
Proceedings of the 44th Annual Meeting of the Am. Soc. of
Photogrammetry. Washington, D.C. p. 93-117.
The establishment of marine parks and reserves represents an important new
approach for the protection of critical marine ecosystems. Interpretation of
remotely sensed imagery could be an effective method for collection,
classification, and analysis of resource information for planning and managing
marine parks and reserves. This potential was examined in two case studies
using Landsat, high and low altitude aircraft Imagery, and the technique of
density slicing to supplement existing information obtained from ground
observations. In both case studies, interpretations revealed important new
information and established the value of these techniques for site specific
analyses. Landsat Imagery could also be a vital tool for a survey team
attempting the efficient acquisition of up-to-date data, especially in remote
areas, or for the planning of regional systems of marine parks and reserves.
Dolan, R. 1973. Coastal Processes. Photog. Eng. 49:255-271.
Along sandy coasts the configuration of the shoreline is seldom straight, but
rather crescentic in plain view. Crescentic forms can serve as indicators of
270
-------
beach and inshore bar-trough relationships, as well as places along the coast
where surge and overwash may focus during storms. The increased availability
of high-altitude aerial photographs offers, for the first time, a data source
for the investigation of crescentic features of sand coasts.
Dunn, Michael C. 1976. Landscape Evaluation with Photographs: Testing the
Preference Approach to Landscape Evaluation. J. Environ. Mngt. 4:15-26.
The development of techniques for landscape evaluation is traced and the
fundamental differences between measurement and preference methods stressed.
The paper then reports the results of a case study which sought to examine
public preferences for constrasting landscapes, and to investigate the
effectiveness of photographs in representing landscapes.
Gates, David M. 1970. Physical and Physiological Properties in Plants.
In: Remote Sensing with Special References to Agriculture and
Forestry. Natl. Research Council, Natl. Academy. Washington, D.C.
p. 224-252.
The appearance of plants and of vegetated surfaces to multlspectral sensors or
to the human eye depends on their interaction with radiation. A plant or
vegetated surface may be viewed actively by reflected sunlight and skylight or
passively by the emission of thermal radiation from the plants.
The precise spectral quality and intensity of plant reflectance and
emittance depends on leaf geometry, morphology, physiology, chemistry, soil
site, and climate. It is the purpose here to discuss those physical and
physiological properties of plants that are significant for multispectral
sensing of vegetation. Some description is given of the appearance of
vegetation and soils.
Gates, David M., and C. M. Benedict. 1963. Convection Phenomena from Plants
in Still Air. Am. J. of Bot. 50:563-573.
The free convection from leaves in still air was observed by means of
schlieren photographs of broad-leaved and coniferous trees. A quantitative
measure of the rate at which energy was convected away from the leaf was
obtained by photographing the size of convection plume, measuring its rate of
flow by means of movie photography, and measuring the temperature of the plume
with a fine thermo-couple. The heat load on a leaf and the surface
temperature of the leaf were obtained with a total hemispherical radiometer
and an infrared radiometer respectively. The observations of free convection
from broad-leaved plants confirmed the values predicted using heat transfer
theory for heated plates. The observations with the branches of coniferous
trees gave values which were not readily available from heat transfer
theory. The schlieren system can also be used to observe forced convection
from plants in wind.
Goff, Glenn F., and Zedler, Paul H. 1972. Derivation of Species Succession
Vectors. Am. Midland Naturalist 8:397-412.
By placing size classes of individual species in sequence, within a reference
ordination model, a series of succession vectors is produced. These vectors
show how the pattern of association of a given species changes in relation to
growth. Analysis of data from 879 forest plots in northeastern Wisconsin
271
-------
shows some tree species entering into successional sequences in accordance
with existing theory. Other species, particularly those of the zeric sand
plains and poorly drained sites, are apparently as self-perpetuating as the
traditional "climax" species of mesic sites.
Gurk, Herbert M. 1973. The Need for More Information and Less Data. In:
Proceedings of the Am. Soc. of Photogrammetry Symposium of Management and
Utilization of Remote Sensing Data. Sioux Falls, South Dakota, p. 513-
527.
Higher resolution, more spectral bands and greater frequency of coverge are
desires of most users of remote sensor data. Yet the outputs from current
observation satellites and aircraft are taxing available facilities and not
being fully utilized by potential users. It is acknowledged also that much
data being collected may be redundant. Techniques such as data compression,
statistical sampling, and mixed-highs multispectral sensing can reduce the
data greatly while retaining the Important information. Data volume
reduction, problems of utilization, and usefulness are discussed for each
technique. Experimental results and current programs are presented, and
recommendations are made for specific programs.
Hairs ton, Nelson G. 1959. Species Abundance and Community Organization.
Ecology 40(3):404-415.
The organization of natural communities was studied from the standpoint of the
relative abundance of the species of micro arthropods in the soil of 2 similar
communities on a long-abandoned field. An examination of the details of
distribution shows a continuous inverse relationship between abundance and
clumping of the more than 100 species studied. This observation provides a
basis for explaining why various empiricially determined "indices of
diversity" are not constant when increasingly large samples from the same
community are analyzed. The strong clumping of the rare species means that
with increased sampling new rare species are more likely to be added to the
total than are additional specimens or rare species already recorded.
In spite of the dependence of the index of diversity on sample size, it
can be used to show some interesting relationships, including a crude but
completely objective separation of samples from similar communities. It is
thus quite useful in situations such as plankton studies where there is no a
priori separation of samples on the basis of the appearance of the area from
which they came.
The most important use of data of this kind is in MacArthur's model based
upon the biological hypothesis that the niches represented by different
species abundances are contiguous but non-overlapping. This Implies that food
determines the abundance of all species, since it is the only factor that is
completely utilized and cannot be shared by different species. MacArthur's
assumption that the sizes of the niches conform to a random distribution is
not confirmed. The departure from randomness is found to be greater with
increased sample size as long as the samples come from the same community, but
decreases if heterogeneous material is added in the form of samples from
another community. These results indicate the degree of organization of the
community, since, from information theory, it follows that organization is
measured as departure from randomness. The organization of a community
results from the outcome of interspecific competition for the available
272
-------
resources, and is expressed both in relative abundance and the spatial
distribution of the constituent species.
Hett, Joan M., and Loucks, Orie L. 1968. Application of Life-Table Analyses
to Tree Seedling in Quetico Provincial Park, Ontario. Forestry
Chronicle, p. 1-4.
Data for numbers of living seedlings by two-year age-classes under old-growth
forests permit examination of life-table analysis as a tool for following
seedling population relationships. The population depletion curves for white
pine, balsam fir and red maple are compared. The negative exponential
depletion model is demonstrated as a potentially sensitive method for
quantitative comparison of silvical response characterisitcs.
Hough, A. F. 1965. A Twenty-Year Record of Understory Vegetational Change in
Virgin Pennsylvania Forest. Ecology 46(3):370-373.
The understory vegetation in a 4,080 acre tract of virgin hemlock-hardwood
forest on the Allegheny National Forest, located in northwestern Pennsylvania,
was studied over a 20-yr period by means of color and black-and-white
photographs taken at 5-yr intervals from 1942 to 1962. The declines which
took place in the understory were believed to result from browsing by the
resident white-tailed deer population. The deer herd, under very light
hunting pressure, has depleted the browse supply and damaged advance
reproduction of hemlock and hardwoods, preventing understory recovery during
the 1942-62 period. Unless relieved, this continued browsing of the
understory vegetation will eventually reduce and endanger the scientific and
educational value of the area.
Idso, S. B., and C. T. deWit. 1970. Light Relations in Plant Canopies.
Applied Optics 9:177-174.
A theory of light relations in plant canopies is presented which has potential
applications in remote sensing and photo-synthetic modeling of plant
canopies. Predictions of the model are compared with field measurements of
light reflection and transmission in a corn crop. Both reflection at the top
of the canopy and transmission at the bottom are predicted within 1% of the
measured values. Profiles connecting these upper and lower limits are equally
well approximated. Variations in the predictions with altitude angle of the
sun are confirmed by the observation of several investigators.
Knight, D. H., and 0. L. Loucks. 1969. A Quantitative Analysis of Wisconsin
Forest Vegetation of the Basis of Plant Functions and Gross Morphology.
Ecology 50(2):219-234.
The structural and functional features of the trees, shrubs, and herbs in the
upland forest communities of Wisconsin have been studied for their potential
in describing relationships between vegetation and environment. Structure is
defined here as the spatial arrangement of the plant biomass, e.g., height,
leaf size, and growth form. Functional features include those which are
apparent adjustments or responses to the environment, e.g., deciduous ness,
shade tolerance, method of seed dispersal, and fire resistance. Quantitative
estimates of the importance of plant features such as these, regardless of
species, were obtained for each of 149 forest stands distributed throughout
273
-------
Wisconsin in all types of upland forest vegetation. Using the index of
similarity, c » 2w/a + b, each of the 149 stands was compared to every other
stand on the basis of 29 tree structural-functional features. The spatial
relationships in the resulting ordination demonstrate that the upland forest
communities can be distinguished and studied using these features. Although
adjacent stands in the ordination are smilar on the basis of tree structure
and function, they can be different in species composition. The usefulness of
grouping species with similar structural-functional features is discussed and
related to the concept of ecological equivalence. Finally, the spatial
relationships in the ordination are used to derive indices that allow
calculation of stand values along structural-functional coordinates. Although
the indices are based solely on plant structure, the resulting coordinates can
be used to infer some functional features of the vegetation as well as
successional status and certain environmental relationships.
Kuchler, A. W. 1973. Problems in Classifying and Mapping Vegetation for
Ecological Reg ional ization. Ecology 54(3)-.512-523.
Important research carried on currently in ecological regionalization calls
for a close look at the role of classifying and mapping vegetation, as both
these activities can be of fundamental significance in regional ization. A
correlation of classifying and mapping vegetation with ecological regions
requires an analysis of vegetation, classifications, regions, and maps.
The analysis of vegetation revealed the character of biogeocenoses, plant
communities, and continua and, incidentally, made it clear that the correct
term for the science of vegetation studies is phytocenology. Problems of
vegetation boundaries can develop when continua are compared with
transitions. This is important in mapping where the nature and location of
boundaries is of major significance. Vegetation is best divided into natural
and cultural vegetation and further subdivided on the basis of (1) physiognomy
and structure, (2) floristics, (3) community dynamism, and (4) community
relations with their respective biotopes.
When these units were applied to an analysis of classifications, it
developed that a basic distinction must be made between highly flexible,
purely descriptive and essentially classless approaches on the one hand, and
clearly organized hierarchies on the other. Serious difficulties can arise
when a detailed description of vegetation is related with a classification,
and an important distinction emerging from these findings is between worldwide
and regional classifications. Multiple mapping at large scales evolved into a
particularly useful and enlightening method.
However, the often demonstrated correlation between phytocenoses and
environmental conditions must not lead a researcher to falsely optimistic
conslusions, as it may not be applicable in the humid tropics. Aubreville,
Poore, Wyatt-Smith, Koriba, Kuchler and Sawyer, etc. have illustrated the need
for caution in interpreting such correlations.
An analysis of some aspects of regions demonstrated that the
relationships between vegetation types and biotopes must be clarified before
meaningful ecological regions can be established. This need was illustrated
with the map and inset maps of the Hunter Valley region in New South Wales,
which proved most revealing.
274
-------
The chief problem of maps in ecological regionalization was found to be
the map scale. Scale problems can usually be solved without much difficulty,
but they must be clearly understood if the results are not to be misleading.
The very nature of the biogeocenose implies by definition that the
geographical distributions of biocenoses and of biotope are most intimately
related. The various analyses revealed that vegetation may be regarded as a
tangible, integrated expression of the biogeocenose. Maps showing the
geographical distribution of the natural (or the potential natural) vegetation
do therefore and thereby also reveal ecological regions.
Leopold, Luna B. 1968. The Hydrologic Effects of Urban Land Use. U.S.
Geological Survey Circular 554. 18 p.
Cities cause local but severe changes in the hydro logic cycle. The pavement
and roofs of urbanization greatly increase the percentage of the land's
surface which is impervious to water. Rather than infiltrate into the ground,
a high proportion of precipitation runs off into streams, causing greater
flooding than in the country. Urban land use promotes erosion and produces
large quantities of sediment.
Urbanization also decreases the quality of water in two ways. First
waste materials, including dissolved solids, pathogenic bacteria, and heat are
added to the water. Second, the high flood peaks and low rate of infiltration
lower the recharge of ground water, and this descreases the amount of water
normally flowing in streams. Hence, there is less water available for such
uses as municipal supply and safely diluting discharges of sewage.
Finally, urbanization commonly causes streams to lose their
attractiveness. Increased floods cause scoured or muddy stream channels.
Trash in the channels adds to the disfigurement. Reduced oxygen content and
reduced water flow alter aquatic life and contribute to turbid, slimy, smelly
s trearns.
In this selection Luna Leopold explains and elaborates on these
changes. It is interesting to note that despite the considerable amount of
research done on urban hydrology, there are many gaps in our understanding.
Leopold is one of the few modern scientists to evaluate and quantify the
esthetics of landscape (Leopold 1969). Streams flowing through cities,
especially, could enhance the quality of urban life if they were properly
understood and managed, although the variety and complexity of urban effects
on hydrology appear to work against this.
Leopold, Luna B., and Walter B. Langein. 1962. The Concept of Entropy in
Landscape Evolution. Geological Survey Professional Paper 500-A.. 20 p.
Entropy, a conceptual framework for describing the distrubution of energy in
natural systems, is expressed in terms of the probability of various states.
The principle of entrophy is based on the concept that when energy in a river
system is as uniformly distributed as may be permitted by physical
constraints. This represents the most probable conditions that can exist for
this natural system. From these general considerations, equations for the
longitudinal profiles of rivers are derived that are mathematically comparable
to those observed in the field. The most probable river profiles tend toward
the condition in which the downstream rate of production of entropy per unit
mass is constant. Hydraulic equations are insufficient to determine the
velocity, depths, and slopes of rivers that are themselves authors of their
275
-------
own hydraulic geometries. A solution becomes possible by introducing the
concept that the distribution of energy tends toward the most probable. This
solution leads to a theoretical definition of the hydraulic geometry of river
channels that agrees closely with field observations.
The most probable state for certain physical systems can also be
illustrated by random-walk models. Average longitudinal profiles and drainage
networks that were derived in this fashion have the properties Implied by the
theory. The drainage networks derived from random walks have some of the
principal properties demonstrated by the Horton analysis; specifically, the
logarithms of stream length and stream numbers are proportional to stream
order.
Lind, Christopher T., and Grant Cottarn. 1969. The Submerged Aquatics of
University Bay; a Study in Eutrophication. The Am. Midland Naturalist
81(20):353-369.
The submerged aquatic plants of University Bay, Lake Mendota, Dane Co.,
Wisconsin, were sampled using the line intercept method. Twenty-one lines
were set perpendicular to the shoreline so that they extended into the bay to
the depth at which submerged aquatic plants ceased to grow. All plants
intercepting the line were recorded within consecutive half-meter segments of
the line. The data were used to construct a contour map of the vegetated
portions of the bay and to delimit the plant communities. Six plant
communities were found. Data on plant height and standing crop were obtained
from quadrat samples taken at biweekly intervals from four regions with the
bay. The data were compared with studies made in 1922. Marked changes in
quantitative composition have occurred since 1922, with the most marked
difference .being the great increase in Myriophyllum exalbeeeene and the
complete disappearance of several species that were formerly major components
of the vegetation.
Lingenfelter, R. E., and Gerald Schubert. 1973. Remote Sensing of Stream
Flow Rates: Correlation of Meander and Discharge Spectra. In: Remote
Sensing and Water Resources Management Proceedings No. 17. p. 404-418.
This report describes a basic study of river meander patterns and discharges
in which the authors attempeted to correlate the discharge spectrum of a river
with the river meander power spectrum determined from aerial and satellite
imagery. Such a correlation could provide a technique for remote sensing of
the water resources of large geographical areas. Although a large enough
number of rivers were not yet studied to attempt a correlation between the
discharge and the meander spectra, some significant characteristics of both
spectra were discovered. Discharge frequency spectra based on long term
records of dally streamflow were found to have an inverse power law dependence
on discharge. This is shown to reflect the short term decay of Individual
floods which are found to have an inverse power law dependence on discharge.
This is shown to reflect the short term decay of individual floods which are
found to have an Inverse power law dependence on time. Meander power spectra
for a number of river reaches, digitized from aerial photography, also show
significant structure. This type of pwer spetra data, digitized from aerial
photography, also shows significant structure since the power spectral density
has an inverse power-law dependence on wave number over one or more portions
276
-------
of the spectrum with breaks in the spectra at characteristic wave numbers. A
number of examples of typical discharge and meander spectra are shown.
Loucks, Orie L. 1970. Evolution of Diversity, Efficiency, and Community
Stability. Am. Zoologist 10:17-25.
The response in species diversity associated with successional change in
vegetation, or in a more general sense, species diversity as a function of
time in any system of primary producers, has been the subject of much
speculation but little direct study. All evidence available shows that
pioneer communities are low in diversity, that in mesic environments the peak
in diversity in forest communities can be expected 100-200 years after the
initiation of a secondary successional sequence (when elements of both the
pioneer and the stable communities are present), and that a downturn in both
diversity and primary production takes place when the entire community is made
up of the shade-tolerant climax species.
The natural tendency in forest systems toward periodic perturbation (at
intervals of 50-200 years) recycles the system and maintains a periodic wave
of peak diversity. This wave is associated with a corresponding wave in peak
primary production. Specialization for the habitats in the early, middle, and
later phases of the cycle has figured prominently in species-isolating
mechanisms, giving rise to the diversity in each stage of the forest
succession. It is concluded that any modifications of the system that
preclude periodic random perturbation and recycling would be detrimental to
the system in the long run.
MacArthur, Robert H., and Pianka, Eric R. 1966. On Optimal Use of a Patchy
Environment. The Am. Naturalist 100(916):603-609.
A graphical method is discussed which allows a specification of the the
optimal diet of a predator in terms of the net amount of energy gained from a
capture of prey as compared to the energy expended in searching for the prey.
The method allows several predictions about changes in the degree of
specialization of the diet as the numbers of different prey organisms
change. For example, a more productive environment should lead to a more
restricted diet in numbers of different species eaten. In a patchy
environment, however, this will not apply to predators that spend most of
their time searching. Moreover, larger patches are used in a more specialized
way than smaller patches.
Maizell, Robert E. 1960. Information Gathering Patterns and Creativity. Am.
Documentation IX:9-17.
A comparison of creative and "noncreative" research chemists with respect to
the ways in which they use their professional and technical literature. The
creative chemists differ from the "noncreative" in that the former read more
technical literature on the job, are less relectant to use literature of
greater reading difficulty, are less influenced in their independence of
thought, read more extensively and consult more frequently the older material,
are more inquisitive and have broader cultural interests. The findings of the
study are believed to be helpful in planning library and information services,
in refining future inquiries into the ways in which scientists use recorded
277
-------
information, and Improving tests for the identification of creative ability
among chemists.
Marsden, J. R., D. E. Pingry, and H. B. Whinston. 1976. Environmental Data
Management: the Identification of Outliers. J. Environ. Econ. and Mngt.
3:154-163.
Increased emphasis on pollution control and abatement has often necessitated
development of large scale data bases. While sophisticated techniques have
been developed and employed for data storage and manipulation, parallel
developments in analyzing the accuracy and reliability of the data have been
absent (see, for instance, the broad spectrum of requirements outlined in the
Federal Water Pollution Control Act Amendments of 1972).
This paper centers on the latter and sets out a procedure for data
editing and "outlier" identification oased on an application of discriminant
analysis. A hypothetical example is included along with some suggested
applications.
Pianka, E. R. 1966. Convexity, Desert Lizards and Spatial Heterogeneity.
Ecology 47(6):1055-1059.
The number of lizard species in the flatland desert habitat is correlated with
several different structural attributes of the vegetation. It is shown that
both the horizontal and vertical components of spatial heterogeneity are
correlated with the number of lizard species. The habits of the twelve
component species are considered briefly as they relate to the partitioning of
the biotope space. Three species are food specialists, eight display various
substrate specificities, and only one species appears to be truly "convex."
Two tests of the present interpretation of these results are proposed, and
some speculations concerning Austrialian flatland desert lizards are made.
Pakulak, A. J., W. Sawka, and R. K. Schmidt. 1974. Analysis of Nesting
Habitat of Canada Geese Using Remote Sensing Imagery. In: Proceedings
of the 2nd Canadian Symposium on Remote Sensing. Guelph, Ontario.
p. 365-371.
The purpose of this study was to evaluate nesting habitat of Canada geese
(Branta canadensis interior) in the Little Seal River area, Manitoba using
recent remote sensing Imagery. Five different sets of Imagery, all taken in
August 1972, were carefully examined to determine which films best represented
vegetation and landform features. Two types of color infrared photography
proved to be most suitable and were used in delineating vegetation-landform
(habitat) units on a study area map. These units were then compared with
goose nesting data collected in spring, 1970. Canada geese nested in 6 of the
8 designated units but showed a marked preference for birch-willow and gravel
ridge habitats. In general, results suggested that remote sensing imagery
could be used to describe habitats in other, largely inaccessible goose
breeding areas. Such an approach, if applied on a large enough scale, would
provide new and relatively, inexpensive ways of estimating annual production
of Canada geese in the Eastern Prairie Population.
278
-------
Simkerloff, D. S., and L. G. Abele. 1976. Island Biogeography Theory and
Conservation Practice. Science 191:285-286.
The application of island biogeography theory to conservation practice is
premature. Theoretically and empirically, a major conclusion of such
applications that refuges should always consist of the largest possible single
area can be incorrect under a variety of biologically feasible conditions.
The cost and irreversibility of large-scale conservation programs demand a
prudent approach to the application of an insufficiently validated theory.
Tans, William. 1974. Priority Ranking of Biotic Natural Areas. The Michigan
Botanist 13:31:39.
A detailed description in outline of the priority ranking system of bio tic
natural areas used by the Scientific Areas Preservation. Council of the DNR,
Wisconsin is provided. Such a system must be devised so that results among
individual researchers will be somewhat similar. The more factors involved in
evaluating an area, the more averaging there will be and a simple additive
meaningful numerical score for each area is hard to devise since it is a
composite score and therefore masks the criminal criteria relating to ultimate
preservation by acquisition, availability and threat. The final scoring
system decided on retains separate scores for biological and physical
characteristics, availability and threat for quick and accurate comparison of
areas.
Terbough, J. 1973. Preservation of Material Diversity: the Problem of
Extinction Prone Species. Contribution to the American Society of
Zoologists Symposium: Toward a System of Ecological Reserves. Houston,
Texas. 37 p.
Preserving diversity in a world of rapidly shrinking land resources will
require a prompt and universal response based on an appropriate application of
ecological knowledge. Every nation should possess an inventory of its
biological endowment. Agencies in charge of parks and wildlife should
consciously adopt policies that are designed to minimize the pace of
extinctions. The common practice of declaring parks in remote or unused
portions of the landscape, or around scenic attractions, may fail to serve
this purpose. Large reserves are needed to preserve natural vegetation
formations, animals at the top of the trophic pyramid, and widespread species
with sedentary habits and poor colonizing ability.
Endemics or rare habitat types can frequently be protected with
realtively small investment in land, provided appropriate tracts can be
identified and sequestered in time. The nesting grounds of colonial species
can be spared with even less land withheld from production as they are usually
located on offshore islets that are unsuitable for agriculture. Migratory
species present more difficult problems since appropriate action often
requires international cooperation.
279
------- |