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

-------